/diy/ - Do It Yourself

have some ohm

Hey /ohm/ not entirely sure if this is the right place, if not please redirect me.

So I am taking apart my gfs hair straightener to try and fix it but I cannot figure out how to take the joint apart. You can see in the first pic I took a small screw out and it looks like there is a pin in there to hold the sides together, on the other side (next pic) the same pin seems to have a hex head. I can't seem to turn it or push it out from the screw side. I've just hit it with some lube and I'll give it a crack in a bit but I'm pretty stuck. Any ideas?

Next pic

>>1218431
>pic
wow

>>1218467
t-thanks?

>>1218452
both of those that I disassembled had a pin there, one had two hex washers but I don't know why, the was just a metallic pin hold in place by the two side caps.
You might break it so be prepared to replace it, the only thing you can do to easen it up a little is compress that part a little to relieve the tension from the spring, but if you compress too much you're just gonna make it harder to pull out.

>>1218430
There are ready-to-use usb-hid libraries for thoose stm32 boards http://www.stm32duino.com/viewtopic.php?f=13&t=576

>>1218431
>bump limit reached on old thread
Actually it hasn't.
It's only at 301 at the moment.

>>1218431
That is an awful image.

>>1218563
'delicious fumes' shoop was better

I didn't actually need to be able to desolder those LEDs to see what was powering them, putting a resistor in parallel with them showed me that since the LEDs dimmed, they're on a constant-current supply. Which is what I'll do when on the road too, maybe after I replace the blinding blue LEDs with some red LEDs.

So if you have a constant current supply of 40mA and put two of the same model LEDs in parallel and connect them to this, is it likely that one will get significantly more current than the other? Since I don't want to be stepping the voltage up and complicating the circuit needlessly, I'm stuck with 5V to power 9 LEDs, and want to know if I need a supply for each of the 9 LEDs, or just for each different sort (3).

>>1218620
Remind me, you run it all from a power brick and charge it from the dynamo, UPS-style?

>>1218620
Ok, looks like pic related is actually a pretty decent constant current source, provided it isn't dependant on the beta of the NPN or whatever the equivalent is for the FET. But I'm still a little on the fence about what's ideal. I don't really want to buy 9 MOSFETs (because they cost more than the bargain bin BC547s).

>>1218635
No dynamo, though that is an interesting idea. Just a normal 5*18650 battery bank, and I'll be pulling current from the USB-out because it's well regulated and constant.

>>1218638
Ok, thanks. 18Ω means 33mA. You need no mosfet for 33mA, that's overdone. What's the forward voltage of the LEDs (just to check dissipation)?

>>1218578
>tfw no dab-core solder

>>1218620
>So if you have a constant current supply of 40mA and put two of the same model LEDs in parallel and connect them to this, is it likely that one will get significantly more current than the other
Yes, and what's more, an LED's Vf falls with rising temperature. One could eventually hog all the current and go pop.

>>1218636
>I don't really want to buy 9 MOSFETs (because they cost more than the bargain bin BC547s).
Then don't. :^) (R1-R3 just for current measurement, R5/R6 adjusts *approximate* multiplication ratio)
In any case, with the regulated supply, you might be fine with plain old resistors as current regulators.

>>1218643
Leds are:
6* 2.0V, 20mA
2* 3.2V, 30mA
1* 3.6V, 35mA
I'll be putting them in parallel, so that's one lot of 120mA, one lot of 60mA, and one lot of 35mA.

I saw a circuit that used another BJT instead of the FET, does it not rely on the beta?

>>1218645
>One could eventually hog all the current and go pop
Then it's a good thing I bought that 100-pack of BC5XX transistors!

>*approximate* multiplication ratio
Will a trimpot be enough to keep these constant, or will the temperature drift be significant? As far as I see it, if I'm relying on the forward voltage of something staying constant enough to keep my LEDs at the same current, then I might as well be using dropper resistors. That said, I also don't want to buy a bunch of voltage regulators, they're also somewhat expensive at the local shop.

LTspice for macOS is a fucking joke at the moment, it tells you that you haven't updated in 299 days when you've redownloaded in the last 50.When you do try to update it takes far too long giving you no indication that the program has done anything but freeze, gives you a command-prompt window showing that some files were copied, and goes back to showing a frozen window. Then when you quit after a while of inactivity and load it again you find it telling you that you haven't updated in 299 days. I can't even use this version at the moment because voltage sources stopped working.

>>1218650
>don't want to buy a bunch of voltage regulators
Voltage regulators have the benefit of supplying a lot of current, and usually have a configuration with some more currents to permit more. The cheap-as-dirt LM317 is 1.5A with ~1V minimum dropout. You can even gang them, so you could use a full-size one for the initial regulation, and then a bunch of LM317Ts for each strand if you wanted to. a small value resistor is obviously recommended if you're going from a voltage source of course.

Additionally, the voltage regulators usually can be put into a current-source mode with an output current determined by the Vref divided by config resistor. For the LM317 Vref is 1.25V, so you could get 40mA from a 30-33 ohm resistor, or 57mA from a 22 ohm resistor. All constant current sources that aren't switch-mode generate a lot of heat, but that 22ohm config only would generate 71mW, easily enough for a quarter watt resistor. It'd be the regulator you'd have to worry about depending on your Vin.

>>1218648
Are those recommended currents or absolute maxima?

>>1218650
I neglected to clarify it's actually it's R5/each leg's emitter resistor.
The tempco of output current appears to be negative and a few uA/degC. You'll want to try to keep all the transistors at the same temperature, as Q2-Q4 temperature coefficients of current output are about -1%/degC! You might consider a transistor array in DIP package for this.

>>1218657
Linear regulators are pleb-tier.
May as well get a boost converter, connect all LEDs of a current in series, and set up for current control.

>>1218657
>The cheap-as-dirt LM317
Costs me 2.74 USD at my local store, where online it's 8.3c. That's a 3300% markup, neat. Sucks that overseas shipping takes a month. If I only need one or three of them then I'll bite the bullet, but if I need all nine then I'm buying 30 of them online.

Total power dissipated by a perfect linear regulator would be between 463mW and 571mW.

How about a LM334 adjustable current source? It looks really good as long as the 10mA in the maximum ratings is a joke. If not, what would you even use one for?

>>1218670
Recommended currents, though for the 30mA and 35mA they're both recommended and maximum, which doesn't make too much sense to me. Those ones are 5mm Cree LEDs, if that helps. So if all the transistors stay at the same temperature, will there be not temperature drift in current?

I'm already running a flyback transformer, I guess I could run another secondary off it, but I assume your suggestion requires a boost converter to keep the output voltage arbitrary and fully adjustable. How accurately could one set the current output from one of those? I guess I'd need three of them too, which is fine as long as I can use the three 1mH axial inductors lying about, or something smaller.

All this is going inside a sardine can too, so I hope these components aren't susceptible to fish smells.

>>1218648
Found it, 3.8..3.9V @ 35mA, 25°C and -2mV/K. Worst case at -20°C in winter: 3.9+45*2mV=4V, harmless.
Erm.. I would use resistors.
The six 20mA as three groups of two in series, the two 30mA in parallel or single and the 35mA alone. Since you have constant 5V there's nothing to gain, neither from regulation nor energy-wise.

>does it not rely on the beta?
It does, but very little. You don't need to drive it down to B=10 or 20 like you would for Uce = 50mV (I use the BC337-40 in such cases, pic). What's really annoying is the 0.6V loss for the current sense resistor. There must be a solution and there is a solution. I need a coffee..

>>1218679
>-20°C
I'd be lucky to get below +5°C or above 30°C, I guess I'll be using dumb current limiting resistors after all. I'll see if I can find datasheets for all my LEDs, from which point I'll find the minimum safe resistance values and adjust with individual trimpots.

>>1218674
>10mA in the maximum ratings is a joke. If not, what would you even use one for?
Precision reference current for data conversion or timer, perhaps?
>30mA and 35mA they're both recommended and maximum
You might get away running both of those at 30mA with minimal loss of brightness. If not, just add one more. LEDs are cheap.
If all the transistors are thermally bonded, the current shouldn't drift more than a few percent, and only toward zero. You could add one transistor per LED if you like.
1 millihenry is probably too big for most switchers, but at 3000Hz that might just work.
You might also look into ICs that are designed just for the purpose of boosting a voltage to a constant current for a series string of LEDs, such as the LM3410, which your local shop may or may not have heard of. Or, adapt a traditional boost converter similar to Pic related, where the desired output voltage is "programmed" as Vfb + Vleds @ desired current. We are getting a bit away from the discretes only rule, I'm sorry.

>>1218683
>LEDs are cheap.
Not these ones, $8 each for some Cree super-flux LEDs with 90° viewing angle at 3000mcd gives approximately 5.5 lumens of radiant flux. At 69c/$ they're the best white LED my local store sells, assuming my approximation about the 50% brightness point being the firm boundary of a cone is good enough. The shop is a little behind the times because they're not selling any raw COBs, but at least that made it easy enough to document every LED they sell on a spreadsheet to sort statistics by.

I'm sure I could get better performance from a Chinese COB bought online, but I set myself a goal to make this from locally sourced parts, for whatever reason. My rule isn't actually discretes only, but locals only, though they pretty much are the same thing.

The Cree datasheets don't list a temperature coefficient, so I'll have to assume based on their chemistry(?). I'll also have to do a voltage regulation analysis of my power bank at different levels of charge and different drains, which should be fun.

Or is pic related better? Both NZD, by the way.

>>1218682
Jaycar is shit.

Hi guys,
newb here
Q: I'm working at a small office in a multi-story residential building, and we have the issue that if someone rings our unit's doorbell (at the buildings main entrance) we've to run to a 2-way-intercom-door-buzzer box to buzz them in.

my idea was it to have it automated, as the door bell and buzzer and 2-way intercom are all in one box, that - and here's where my newbness shines - "some" IC, that has the ability, that if a current is on a line (the door bell ringing) that the IC waits like 5 seconds before it closes the buzzer-in circuit for the buildings main entrance door.
preferably the IC has a line for a manual on/off switch too.

I'm sure something like that exist, but I don't know what search terms to use to find the correct thing.

thanks-a-thon in advance

>>1218689
>locally grown
Ah, that clarifies matters. If you want to try the current-controlled boost conversion route, probably better to work backwards from whatever the local shop stocks. Even the hoary old LM2577 would do if it weren't for its ridiculously high quiescent current. The general idea, which was suggested by LM317-anon earlier, is pretty similar across implementations, but the details, suitability for the lowish currents in question, and other component values will vary by the individual type.
On a related note, the LM723 is old enough that they might stock it for use as your EL power supply controller.
Regarding soldering irons, since neither reads out the actual temperature I say go for wattage. The 10W will likely disserve you as soon as you need to work on fuse clips or large ground planes or the like.

>>1218697
You could do it easily enough with two 555s or a single 556, along with a handful of extra components. The manual on/off would be accomplished simply by interrupting the doorbell signal with a SPST switch.

>>1218697
get a scrap piece of 2x4, put it between the door and the frame.

>>1218700
>You could do it easily enough with two 555s or a single 556, along with a handful of extra components. The manual on/off would be accomplished simply by interrupting the doorbell signal with a SPST switch.

thanks for the info, can you elaborate more on that, please?

>>1218705
>get a scrap piece of 2x4, put it between the door and the frame.
not a good idea, because surveillance cams due to neighborhood. landlord wants entrance doors closed at all times

>>1218682
>dumb current limiting resistors
If you wire the six LEDs as 3p 2s (4V), 30mA and 35mA separate you end up at 5V*155mA=0.775W. That's hard to beat with any 'active' solution. Enough dynamo power left to keep the brick charged.

>>1218696
Yep. I've got Surplustronics locally too at least.

>>1218700
>if it weren't for its ridiculously high quiescent current
And being $21. They stock all of those mentioned components, but I'll model the lot in a python code to figure out which arrangement would work best in terms of temperature variation, cost, complexity, and additional power usage. I don't think LTspice has the accuracy (or lack of fucking bugs) to do it for me, and I should have more control with a code. Thank god for scipy.linalg.solve().

Maybe I'll find some way to completely offset the temperature coefficients by combining positive and negative coefficients, but that would take some intense preparation.

>>1218711
Oh shit, yeah I forgot that those ones were below 2.5V, I'll definitely series those. I wonder what the easiest way of hooking a dynamo up to a bike is? Magnets on the rim with a big field-winding strapped to my front fork? Or a little motor shaft on my derailleur? But getting enough power to feed the EL inverter might prove difficult. I'll do some tests on the chintzy solar panel the power bank came with, I'll probably be able to mount it somewhere when I'm done.

>>1218706
All-in-one intercom systems are kinda weird, and I'm not an expert on them. Maybe another anon knows better about them.
If you need elaboration on how to use a 555, let's just say there are whole books on that. I suggest buying a time-delay relay module instead. What you need is called a "triggered delayed interval timer" in industrial circles. Here's a 12V version:
http://timers.shop/Multi-Functional-5V-20V-Time-Delay-Relay-Timer-5-amp_p_12.html
Here's the industrial mains-powered version:
https://www.alliedelec.com/macromatic-thr-3856u/70274479/
You will probably also need a second relay, and possibly other electronics depending on the intercom/doorbell system, to receive the signal from your doorbell and convert it to the dry contact trigger for the time delay module.

>>1218713
>$21 for an LM2577
Good god anon, they really don't allow any fun down there do they.

I wonder if it would be any more efficient to step up the power to this thing? It has a built-in dimmer, a feature which I might have attempted to install, but at just one channel I don't think it's a good choice. No datasheet either, so we'll never know.

>>1218717
That looks better suited for the 1W and above LEDs.

>>1218717
https://surplustronics.co.nz/products/6952-led-driver-lithium-battery-charger-dc-dc-constant-voltage-constant-current-adjustable

>>1218719
How the fuck can you get to their website? It only gives me "Sorry, something went wrong".

>>1218721
their front page is currently busted, everything else seems fine

>>1218717
But it says full datasheet available. Which website?

>>1218723
https://www.jaycar.co.nz/dimmable-constant-current-led-driver-module/p/AA0595
It's not there, it's usually in a "downloads" tab next to "description" and "specifications".

>>1218719
Anyway, that thing has its own buck/boost converter, right? I'd need only three of them to power each set of LEDs in series, but with some rework I bet I could get some of the components to be shared among the boards. It is a pretty space-heavy fix, though.

I initially planned on using 6* 30° red LEDs for the rear lights at 30° angles from one another to spread the light out so I can be visible from all angles to the rear, 2* high-lumen white 90° LED on the front for a similar purpose, and a single high-mcd white 15° LED on the front for a more intense bright beam, but I'm starting to have second thoughts. Mainly whether I need so many LEDs when my bike will look like something out of TRON even without them. Pic related was what my two light holders were to look like, though I've already changed the rear design to incorporate a tuna can. The rear light mounts under the seat on the seat's own mounting bars.

Active brake lights are also an idea I'm flirting with.

>>1218727
The surplustronics one is boost converter only, so would need series string of some kind

>>1218729
It would work for the 6 rear lights and two white lights, but not the single spotlight, unless I run it together with the two front lights.

Also I crash way too often (why I'm making lights in the first place I'm afraid) to put a protruding bike light on my handlebars, so I'll probably mount it on the unsprung arch over my front wheel, since the control switch won't be mounted on the front light regardless.

A few threads back some helpful anon talked about his "folder of useful schematics/circuits". Does anyone have a package like that to share?

>>1218729
>boost converter
Input Voltage: 5-35V
Output Voltage: Continuously adjustable from 1.25 V to 30V
That's step-down. Didn't the guy in the video also say so?

>>1218715
Thanks a lot for the info, will see if it helps me.
Those all in one intercoms are rather common here (Germany)
You just pick it up like a phone to kick in the 2-way intercom, and push the green button for buzzing them in
>pic related, my apartment building's intercom-buzzer-door-bell

>>1218741
So the key button unlocks the door and what does 1..6?

>>1218747
>So the key button unlocks the door and what does 1..6?
they're just placeholders for the casing (no buttons in them). In the event the system is suppose to be capable of more

>>1218751
Ok. Can you buzz/open by pushing the key button without first picking up the phone?

>>1218780
>Ok. Can you buzz/open by pushing the key button without first picking up the phone?
yep
doorbell is inside it too

>>1218780
>>1218783
As you can see by the pic, it's rather primitive. Not even an IC (neither on the back) on the PCB.

>>1218431
Why are you jewing Winfield Hill out of his authorship credit for the Art of Electronics? I actually know the man.. he works at Edwin Land's private research Rowland institute

>>1218784
That looks friendly. I know that these systems have been optimized up to a sort of bus structure/digital addressing to reduce the number of wires to a bare minimum. This looks much simpler. There seems to be a diagram at the bottom where the white phone cord comes in, can't decipher that but it looks important. Did you already try something like google.de -> siedle model x internal wiring? Maybe you can poke around with a multimeter and find some DC/AC that's accessible. Be sure to keep the receiver on hook while doing that.

Functionally you need to detect the bell (honk, beep, chime whatever), use that signal to trigger a delay (I would start the time delay after the bell button has been released) which then triggers a relay or switch that pushes/shorts the key button for a second or three and then all resets to wait state. These are all very common functions, the challenge is to integrate it into the siedle without changing anything, merely adding an overlay that does not interfere with the normal function and can easily be removed if need be.

>>1218731
I guess at UTC+12 it's not the right time for a message but I found a buck-boost thing at Ali: www.aliexpress.com/item/Boost-Buck-DC-DC-Adjustable-Step-Up-Down-Converter-XL6009-Power-Supply-Module-20W-5-32V/32627799347.html They always seem to have 2 coils of about the same size (33µH). This one has another one (2µ2) as a output filter.

When I want to power an arduino with an external power supply, do I but the wire with 5v running through it into the VIN pin or the 5v pin.
Intuitively Id think the VIN but Ive been looking at certain schematics and they run power into the 5v pin.

>>1218869
If you have regulated 5V you don't want to put it into the linear voltage regulator because it drops the voltage and needs about 6V to work. So use the raw pin (unless it's a 3.3V Arduino).

>>1218878
I still haven't taken the plunge and started drafting boards for SMD parts. Does it save time overall? I know I've had to do a lot of troubleshooting of soldered protoboards because I fucked up some interconnects, something that would never happen on a PCB. I've done some basic PCB design at work but never did anything well enough to fab it.

>>1218880
You can etch you own boards if you're not sure if it's worth to fab them. However this is pretty time consuming and 2 layer boards from China cost pretty much nothing if you can wait 2-3 weeks to get them.

>>1218883
How long does it take to draft them? Are there libraries available that have all the common SMD component footprints?

>>1218884
they have all new fangled software to help you with that now, kicad is open source but i don't know if you can just design a pcb with it, you might have to lay out the circuit then assign footprints to parts then layout the pcb and it checks your netlist to show you what to connect where. it has all common parts, footprints etc but you can make your own if you like.


OF COURSE there are other softwares out there

>>1218884
If it's one of the many standardised SMD footprints it's available of course. If it's not a very rare and special IC you can usually find it easily.

>>1218884
Might take a few hours to get used to it, but eventually creating each part will take you a few minutes. It's really not that hard to get used to.

After one or two boards I found routing smds far easier than through hole, and after soldering one or two boards I now prefer smds. far easier to rework too, even with a regular iron. If you're starting out, stick to smds with pins that extend from the package and perhaps 0805 and larger passives. These are quite easy to solder.

Libraries exist for just about every layout software there is, though I would really recommend that you figure out how to make your own parts. It's rare but sometimes parts in libraries have mistakes and that can be frustrating to use. If you make a mistake on your own part you've no one else to blame but yourself.

You might already be doing this if you're used to through hole, just in case you are not, I'd suggest printing out 1:1 scale your pcb artwork and placing components on these before sending to a pcb house.

Best of luck.

>>1218808
I think I'd need some sort of specialised constant-current buck-boost module, not a voltage one, otherwise I'll still run into the same uncertainties. In the scope of things, I think a linear regulator of some sort will use less power with the small LEDs I'll be using, not to mention being able to be made without offshore parts. Hopefully spice works today.

How big of a capacitor would you need for it to ignite fuel when shorted?

>>1218908
Shorting it would be less likely to start ignition than putting it across a tiny spark gap or fuseable link, but I've definitely seen sparks come from a 3" diameter 4" tall capacitor charged to 240*SQRT(2) volts when shorted with a screwdriver.

>>1218880
>>1218884
If you are doing simple stuff then usually all the stuff is already in the libraries. But if you are doing anything slightly sophisticated you will need to make a new part/footprint, it's just a fact of life in ECAD.

SMD is a lot easier if you have good magnification, depending on how good your eyes are.

>>1218907
Good morning. I think you're right. The 5V - small LEDs affair has been settled and can be further optimized as soon as you get hold of the datasheets for all your LEDs, or even better, measure them yourself, not difficult. The ELwire thing is something else, there may be room for improvement.

>>1218939
Should i go for some transformer based spark generator instead? Power comes from 12v battery

>>1218786
I just pasta it forward. Thanks for the update, may the next OP be sure to include it.

>>1218880
>Does it save time overall
It can, if you use good tweezers, solder paste, a good magnifier, and a reflow oven. The $30 USB webcam microscopes are bretty gud for this. It definitely does open up a broader repertoire of parts you can use, which can save quite a bit of time on the design end. I'd still try to avoid BGAs and really fine pitch stuff at first.
Personally, these days I try to avoid through-hole stuff. With SMT I can etch and fab my own boards without a drill press. It's kind of lame that they still bother with through-hole in le making world.

>>1218946
I was still going to work out whether a simple dropper resistor would have significantly worse temperature drift than a BJT based system, and compare their power drain. I'll buy the LEDs today, take voltage measurements at different currents, and get their characteristic exponential curves. I might try to do that with some BJTs too, but since each transistor's beta is different, I'm not getting my hopes up.

Now while I'm pretty sure a flyback transformer is the only way to efficiently step up voltage by a factor of 20, I still need to figure out why my shitty astable multivibrator won't work. I don't want to be using a reverse-bias BJT relaxation oscillator again.

>>1218982
>I was still going to work out whether a simple dropper resistor would have significantly worse temperature drift than a BJT based system, and compare their power drain
The negative tempco of the LED will work against the positive tempco of many resistors and in favor of not frying the LEDs.
Power dissipation will probably be about the same vs. resistors since they are both serving exactly that function of converting electricity into heat.
>each transistor's beta is different
If you happen to get BJTs that are still on the tape, it's likely they won't deviate in Vbe(on) and beta too much from one to the next.
>shitty astable multivibrator won't work
Steam locomotive wheels were made slightly off-balanced so that they would start more reliably. Perhaps your astable multivib design needs something like that?
Alternatively, 555s are common and cheap, and would probably be easier to make work over frequency and even to PWM-regulate.
Maybe worrying too too much about efficiency is over-engineering. Might switch back to designing the EL supply for a minute and see just how much power it needs, but 600mW for the LEDs alone allows over 30 hours of riding from a 20Wh supply.

>>1218982
When you have the LED data you can easily gauge the overall efficiency. I already posted my calculation, temperature is not a problem. The beta spread plays no role because it is high enough. As long as you have 5V it doesn't matter which way you control the current, active-linear or passive-resistive, efficiency will be the same.

For the ELwire, have you considered a simple resonant push-pull converter? Two transistors, two resistors, center-tapped primary winding, core either open or air gap. Frequency will be set by secondary inductance and C of the ELwire. Have a link for the ELwire you use, C per meter?

>>1219001
>both serving exactly that function
Technically the BJT designs do drain a little extra current, but that shouldn't be very significant unless the design is really fucking bad. Since the LEDs themselves are drawing fairly little current, I thought I should at least take it into consideration. But if the LEDs and resistors are heating up from their own power consumption as opposed to strictly ambient conditions and dissipate their heat at different rates, there could be an increase of current somewhere, so I'll prepare for ±10°C between parts just in case.

>still on the tape
They were in the same bargain bag, I don't think so.

>30 hours of riding from a 20Wh supply
One of the cells may or may not have made a bubbling noise when I was soldering on to it, but it didn't burn my room down or get very warm at all when I left it to charge for half a day so I think we're good on the "not blowing up" part. Still, flames shooting out the back would complete the aesthetic of EL wire + obnoxious eurobeats.

>>1219002
>When you have the LED data
I fucking forgot to buy them, same with the extra fuse. But I have a soldering iron again, plus a bag of varied ceramic caps, some actual resistance wire that doesn't need to be 30m long, and a couple of 1k trimpots. But until I do buy them, I'll see if I can model a dark/lit street and see how far the light from different combinations of LEDs will be visible, if I want to change my arrangement.

I haven't bought the EL wire either yet, but it's the high brightness stuff that Adafruit sells, here's the datasheet:
https://cdn-shop.adafruit.com/wanyuPL3.pdf
I'll be using the "blue-green" EL wire because it's the brightest. It isn't UV resistant, so I might look around at other brands, but it's the only EL wire a local import store stocks (nicegear for any other Kiwis) and so doesn't have Adafruit's godawful 50%+ shipping markup on it.

>>1219021
The new soldering iron is rippin', though it does show its price when the iron is almost at temperature and the "heating up" light flickers while making a noise. Maybe that oscillation can power my transformer. I miss the extra 10W and natural rubber cable of my last iron, but not burning all my flux is pretty neat. I'd swap the cables if the temperature controlled iron didn't have tri-lobe screws.

>>1219021
>±10°C between parts just in case
Good enough. They "should" maintain their stated tolerance across temperature and load. Vbe(on) is fortunately easy to measure and sort with any diode check function on a multimeter.
>lithium ion cell
>soldering on to it
Aw hell no. Is it at all bulged? If so, get it out of there. Otherwise, I kindly counsel you to keep that power pack away from your junk from now on and never leave it unattended while charging. There's a reason Li-ion packs are spot welded and never soldered.
I do think you should consider 1W-class LEDs for your headlamps if you can get them. It might be worth trying a sign shop for something brighter if you come to need them.

>>1218697
>if a current is on a line (the door bell ringing) that the IC waits like 5 seconds before it closes the buzzer-in circuit for the buildings main entrance door.
>>1218707
>landlord wants entrance doors closed at all times
But it's OK for the door to unlock for anyone who rings the bell?

>>1218878
So Im guessing thats why the VIN pin is sometimes labeled 9v, because you need to input more voltage.
Also what happens if you have a 3.3v supply?

>>1218878
SAVED

>>1219161
You use a board built or modified for 3.3v. Ladyada's got a guide for conversion.

Question on LED durability.
I've got a bunch of 3.0-3.2 V 20 mAh white leds.

Will it be a problem to driver them from either
A) a 3.3 v regulator, over driving by .1 volt
B) a 5V regulator, with a 22Ohm resistor, drawing 21 mAh instead of 20 mAh.

The leds won't be on long, just bursts of a few seconds. I just don't want to wire it up, and then have it die on me a few months later because I was driving the leds slightly too hard.

>>1219168
A) will likely be a problem, LED current vs voltage goes up very steeply, at 3.3V you'll probably be quite a bit beyond rating.

B) approach works fine, but your math's wrong, you should have 100 ohm or thereabouts.

>>1219171
Oh, I should mention 4 leds in parallel.
This calculator, http://ledcalc.com/#calc, says I'd need a 22.5 ohm resistor, but I've only got a 22 ohm, and then straight to a 100 ohm.

Sounds like a slightly lower resistance should be fine, thanks.

>>1219176
You can use four individual 100 ohms for each LED, or 22 for the set's close enough too.

>>1219178
Thanks!
Space is tight, so I'd rather use the 22.

>>1219176
I'd go with the 100 ohms on each. With a single current regulator for the set, one LED might have a "slightly" lower forward voltage and therefore hog more than equal current, getting hotter and thermally running away only to go pop, subjecting the rest to more stress and imminent failure.

>>1219182
Huh, ok.
I didn't realize that could be an issue.

So, regulator, then each led with it's own resistor?

>>1219182
Just reread to see that you're lighting them only for a few seconds at a time, in which case the single 22 is more than likely okay. If you were leaving them on for much longer, the individual 100s would be a much better idea.

>>1219193
Got it.
I'll think I'll try to fit 4 100 ohm resistors, just to be safe.

>>1219085
>spot welded and never soldered.
It had a metal strip spot-welded across all the cells to put them in series, but the wires connecting them to the PCB were originally soldered. I just replaced the wires with not broken ones.

Now I was planning on going for a simple, "just enough for them to see me" light array, hence the wide viewing angles, not anything to illuminate a dark road. But I'm certainly considering going back on that, along with adding active brake lights and indicators. I'll compound some data on LED COBs from Mouser datasheets and compare their luminous efficacy. Then I'll look for some real local COBs, hopefully between 0.6 and 3 watts. If I'm running those beefy lights then I'll also want to dim them, which I probably won't be able to do with the same frequency as the flyback transformer because I plan on changing its frequency, because doing so changes the EL wire's colour. If I put a flashing-light oscillator on it too, then that's three shitty oscillators. I've got my work cut out for me. Meanwhile, I'm cycling home in the dark without lights, I think I'll just tape some cheap LEDs to my handlebars and seaypost for now.

>>1219211
>to put them in series
Meant parallel. Fuck the "seaypost" too.

>>1219211
Ain't it great how projects grow to fill the available ambition?
There isn't much reason to dim a bike headlight. I'd expect you'd want to see as far as you can always.
So how much do 555s cost in your neck of the woods? And how do you feel about microcontrollers?

>>1219216
>Ain't it great how projects grow to fill the available ambition?
Not for me, I never get anything done. But this time I'm limiting myself both in only using local ingredients, and in making everything out of fish cans.

I thought I'd want to dim it for the purposes of both battery life and not blinding pedestrians/drivers, the former being important if it's pulling 5W or so.

555 costs 2.70 NZD, 556 costs 2.40 NZD, so I guess their costs are insignificant when compared to their profit margins. Or they're just fucking with me. Either way I am somewhat trying to avoid large DIPs, and µCs are completely out of the question for me, it just takes the fun of physical engineering and troubleshooting away and replaces it with code-monkeying. I know it's stupid, but making my own primitive 555-style timer from BJTs would be something I am considering, though it's probably far too bulky.

A discrete comparator-based relaxation oscillator could work, which it looks like I might be able to do with only 2 or 3 transistors per oscillator. Since LTspice is working again, I'll give that a shot. Needless to say, spice does not like the reverse-bias BJT oscillator.

>>1219244
555s are $1.20 at surplustronics

>>1219264
Correct you are, here I only searched there for LM555s, didn't realise they would stock them as NE555s instead. Also, I found pic related, but I would seriously have to revamp my power supply if I intended on using it. Pretty overkill.

>>1219211
pleas don't go crazy with this, lighting enthusiastic cyclists are just a nuisance.
>"just enough for them to see me" light array, hence the wide viewing angles, not anything to illuminate a dark road.
thing that gets me about cyclists is they adorn themselves in hundreds of tiny flashing lights and expect themselves to be visible, which they are not. look at bikes, cars. headlights are visible not because of blinding brightness, flashing or number they are visible because they are large in size. a laser is visible from a distance and bright to be seen but it doesn't tell you anything as a road user.
please at least have a gander at lighting regulations for vehicles to get the angles and brightnesses actual vehicles use when designing this thing.

>>1219211
>active brake lights
on a bicycle? absolutely pointless.
>indicators
how are you going to ensure the separation between them on a bicycle to make them useful from any kind of distance? why are you trying to subvert other road users expectations?


i fucking dread to think what you are doing with EL wire.

get yourself a hi vis vest.

>>1219244
Oh, right.
>buy 2, save 30 kiwicents
Interdasting. It's almost as if they rolled dice to figure their pricing.
>making my own primitive 555-style timer from BJTs would be something I am considering
Far too bulky and all sorts of """fun""" matching up diff pairs out of stock, but the motivation as an off-bike exercise isn't entirely nonsensical. That "cont" lead on the 555 is your key to regulation and feedback, btw.
Also take warning from >>1219278 however bitter they may be: you'll want to make those lights appear larger than they are, through some sort of diffuser-reflector setup, especially at the rear..

>>1219278
I'm sure the EL wire outlining my frame will make me identifiable as a cyclist and help with with the "pinpoint brightness" problem, while the LEDs will make me visible head-on and tail-on, and they are required by the law. I could try an EL panel on the front and rear each to get more of a "brightness over area" thing, but I already have a relatively reflective orange cycling raincoat which I wear, which should be fine.

>subvert other road users expectations
The last person I crashed into asked me if I had a bicycle licence, I don't think their expectations are existent if the first place. The main motivator towards indicators is that I'm constantly changing gears and tapping the brakes, and often find myself unable to indicate because I'm in the middle of a gear change as I go about a corner. I lie, the main motivator for brake lights and indicators is simply to have fun designing and building the things. I might as well install an obnoxious speaker if I go that far, but I probably won't.

I find it best to strobe the rear lights for sheer eyecatchingness and find most cyclists around do the same, but I'm not a big fan of strobing front lights.

>>1219282
Perhaps some sanded sheets of acetate could make a nice diffuser? Especially with a 3W, 240lm, 120° viewing angle LED I found for less than 1/2 the price of the shitty Jaycar 5mm LEDs. No datasheet, but for that price I'm buying one anyway. At that power output I bet it's worth investing in a down-converting constant-current supply, because who wants to dissipate 2W through resistors? I find most rear lights have a bunch of shitty red LEDs to make them fairly diffuse in the first place, and my plan of 6 LEDs is probably fairly good on that front, but adding a diffusor couldn't hurt. Diffused lighting is also more aesthetic.

I was also looking at building a discrete comparator instead of a 555 to get an RC oscillator made, since the sinusoidal output would (could?) be much better for the transformer.

>>1219282
>however bitter they may be
i'm not bitter, i don't like to be blinded by obnoxiously bright lights while on foot or distracted by disco-on-wheels while driving

>>1219290
>I crashed into
you crashed into them lol? maybe you should be diy-ing some glasses instead?
>often find myself unable to indicate because I'm in the middle of a gear change
you need to anticipate better, unless you are the first person in the history of the world to realise that hand signals on a bike are impossible to do?
from any distance at all unless you have a 1 1/2 ft boom on the back of your bike indicators won't be any practical use.


>I find it best to strobe the rear lights
here we get to the crux of the matter, a flashing rear light will make your brake light pointless, nobody will be able to tell if its a brake light or just a flashing/pulsing/strobing rear light.
this also affects the indicators to a degree.

>sheer eyecatchingness
there is a fine balance between being there to be seen and being distracting and annoying.

>but I'm not a big fan of strobing front lights.
seriously, thank you.

>I'm sure the EL wire outlining my frame will make me identifiable as a cyclist
EL wire has absolutely nothing in terms of visibility as a decent set of retroreflectors on the bike and on you.
please check local regulations as to what colours you are allowed to display. as much as you might think i want you to get arrested i'm really just trying to help you out.

>simply to have fun designing and building the things.
great, good for you anon godspeed.

>>1219297
>a flashing rear light will make your brake light pointless
I'm cutting those extra features from the design anyway, they weren't terribly serious suggestions.

Reflectors are good and all, but having something that actually emits light actively would stop the reflectors from being as blinding if I were to cover my whole bike in them. The point of the EL wire is to clearly outline the frame independently from the added visibility of the lights (and reflectors, those are now on my to-install list thanks).

I'll see if I can change the degree to which the rear lights strobe, preferably something I can change if need-be. I was wondering how good an idea a diffusor on a front light really is, and about how much it would decrease the light output. Jaycar do sell a parabolic LED reflector and not the LED for it to go with, so I'll see if it fits this cool 3W thing. Now I might have to consider using a heat sink.

>>1219211
You should consider a microcontroller for all that extra functionality. A PIC series would get the job done.

>>1218887
The simplest way for me to sum up the KiCad workflow would be:

1.) Create schematic. Libraries for most common parts are available.

2.) Associate components with footprints. Footprint libraries are pulled from a Github repo, so unless you're working with uncommon components, odds are you won't have to make any footprints.

3.) Lay out your PCB.

Creating symbol and footprint libraries is easy, but avoid doing it when you don't have to. KiCad really tries to help you not have to reinvent the wheel, so accept that help instead of remaking a library for the umpteenth time.

>>1219303
I find designing and constructing hardware more enjoyable than software. It's not like I'll have 50 different settings or anything, just a PWM dimmer or two, a variable frequency oscillator or two, and a constant current power supply or two. All good fun for a bunch of BJTs.

I'm trying to make an audio spectrum analyzer from a microcontroller and an individually addressable LED strip. I'm doing some tests and simulations in matlab to figure out the best-looking algorithm. One thing I noticed is that the Fourier transform of music doesn't look at all like I'd expect from the spectrum analyzer in my music player.

This graph is the FFT of 480 samples of The XX's Intro at 5 seconds.
https://youtu.be/H4UtBYUMVJk?t=3
At that time, the song is one note played repeatedly and a bass note playing constantly. Yet the most prominent frequency is 17360Hz, which is barely even audible. Can anyone explain what's going on?

Code below.

data2 = []
for n=1:length(data)
data2(n) = .5*(data(n,1) + data(n,2));
% data2(n) = (data(n,1));
end

start_point = 5;
data2 = data2((fs*start_point):(fs*(start_point)+480));
player_fft = fft(data2);
P2 = abs(data2/length(data2));
P1 = P2(1:length(data2)/2+1);
P1(2:end-1) = 2*P1(2:end-1);
data_fft = fs*(0:(length(data2)/2))/length(data2);
plot(data_fft,P1)

>>1219432

my guess is your data has been corrupted by MP3 encoding. try it with an old-school CD instead.

>>1219456
The file sounds fine, though, whether I play it in a music player or in Matlab.

>>1219432
>high frequency content
That's what plucking is.
>The file sounds fine
That's what psycho-acoustics is for.
You should listen to the content through a band pass filter so you hear what your spectrum analyzer is hearing.

>>1219570
Hmm, alright. At that point in the song, my music player's spectrum analyzer, which I'm trying to model mine after, looks like this. Any idea how the binning works on this? The frequency range is still the audio band, yet it seems to be way heavier on the lower frequencies. I'm guessing it's some sort of logarithmic scale - maybe inverse log so that the bins are are wider at the lower frequencies and the high frequencies' representations are attenuated by narrow bins.

I might try making some MP3 files of various frequency sine waves to see how the player displays them.

>>1219573
Tested with some linear frequency sweeps. Definitely big bins at low frequencies. I'll try to work out a formula to make some for myself with a consistent exponential change in bin size when I'm not doing homework.

I've also gotta figure out if I'll add in a bit of the frequency content of neighboring bins, since the frequency resolution is so low in my display (~30) bins. Might make it prettier.

>>1219432
480 samples is a bit short, try 4800.

Audio spectrums are usually shown as log frequency and log amplitude, not linear.

>>1218884
Sure, KiCad also has some assistants to create common package types for common and uncommon pin counts algorithmically. For example, to create a DDR3 BGA footprint you just create the regular BGA and depopulate manually.

>>1219573
>some sort of logarithmic scale
Ah, scaling by the square root of frequency might look better as a first approximation. You should definitely add in the content of nearby bins to simulate filter Q (look it up).

>>1219600
>4800
Why? I'm dumping it all into 30 bins anyway. I want to have a lot of wiggle room on processing time when I put this on a microcontroller.

>>1219602
More samples helps get decent resolution at the low end.

>>1219603
Good point, thanks.

>>1219604
May also want to try a constant Q transform, which changes from the constant absolute frequency resolution of an FFT to a constant relative resolution like a traditional bank of bandpass filters. May not be the most practical way to do it though.

>>1218431
is it normal for me to be afraid of using leaded solder? I use lead free and as you can imagine it's a chore to work with

>>1219614
Looks interesting. I'm probably going to use an fft and approximate a filter rolloff on either side of my bins since I'm more familiar with that method. But I'll read up on the implementation of the CQT and see if it looks easy.

>>1219620
Leaded solder is pretty safe as long as you ventilate the area, wash exposed skin after handling, and stop eating it.

>>1219620
Kind of. When I work on existing electronics I find myself sucking up all the old solder on joints I'm working on to replace the low with lead solder, which isn't terribly productive, so you've got me beat there. But any fumes you'd be inhaling are from burning flux, not lead, so there isn't really much risk unless you're getting particulate lead all over your hands and not washing them.

>>1219626
Honestly, though, it tastes really good. I don't chew and swallow because that's just stupid but if you give it a lick every so often, no harm done.

I'm trying to interpret serial data from a very old keyboard.
Each key sends a 10-bit code over one wire. The first bit is always 1.
Is timing the right way to do this? Detect the first bit, then poll the data line again after a bit's worth of time (~0.8ms) for each subsequent bit? Or is there a better way to do that?
Pic related, it's 1001001001

>>1219629
Could be 12500bps async serial. Can you set up a UART for that nonstandard speed and see what happens?

>>1219620
Buy better lead free solder.

I have used decent lead free and it was just as easy to use as lead.

>>1219634
I can see on the scope that it's not ascii or ebcdic. I'm pretty sure it's something completely unique.
This is a keyboard from a Perkin-Elmer scientific computer, it probably dates from the late 70s.
It has a DIN connector but it's definitely not an IBM PC type interface. It uses 4 pins, two for power and ground, one for data, and one for the bell.

>>1219628
A true Roman graces this general with his presence!

>>1219643
Many old workstations used async serial for their peripherals and had UARTs on the computer end to communicate with them. What you have on the scope looks just like an 8-bit async serial byte on the wire, with parity tbd.

>>1219646
Romans didn't get sick from their lead pipes because the water in them was always flowing and therefore the amount of lead in the water was of low significance. Get your facts right.

>>1219650
You first. http://io9.gizmodo.com/5877587/the-first-artificial-sweetener-poisoned-lots-of-romans

>>1219629
How do you know the first bit isn't 0 if you're triggering the capture on the first high?

>>1219639
>shit nobody said ever

>>1219651
You were referring to lead salts? Then I misspoke. Sugar of lead was certainly a killer. The link sends me to the AU site which doesn't have that article, sorry.

>>1219646
>What you have on the scope looks just like an 8-bit async serial byte
You're right, it was. I was able to receive ascii text at 1200 baud, 8 data bits, no parity. I didn't recognize it because it was little-endian and inverted.

>>1219693
>12500 baud... must be painter's brain
Cheers m8, enjoy your bad-ASCII keyboard

>>1219693
>little-endian
Those were the days. Not that I'm speaking from experience or anything.

I've got a few random devices throughout my place or for buddies that I need to fix loose ports on. Everything works fine, they're just those defective-by-design type where the actual electrical connection is the only mechanical connection to the board. What's the best type of epoxy or various goos to use to secure it in place, and build a bit of a structure up around it to help secure it? Is there a good non-conductive quick cure epoxy you guys can recommend?

>>1219812
Silastic is what the pros use, but silicones, hot glues, and epoxies also work. You want something that isn't too rigid and adds some strain relief, otherwise where the wire exits the compound will be almost as weak and prone to over-bending as when it was just soldered to the board. Epoxies can be quite rigid, so watch out for them. Dielectric strength and heat resistance are also factors, but if you can search about for some actual PCB-grade silastic then you're good as gold. If you're worried that the wire will simply pull out of the compound, tie a knot in the base that the compound will physically hold, instead of relying on the sticktivity of the glue. You'll probably only need to worry about that if you have something like teflon wire insulation however.

>>1219827
Ouch, Silastic is expensive. I'll see if I can get a hold of some in the future at a reasonable price, but I'll get BSI's epoxy in the meanwhile, rigid one to mound connectors and flexible to mount wires. One of the easy ones I have is just the USB port on a keyboard, it's come loose once already, just going to mechanically fixate it for hopefully a permanent fix this time.

>>1219812

instead of going around gluing connectors (like USB, DC power jacks, headphones jacks, etc) why not do something more direct to solve your problem. get a marker and write on your forehead, "I AM INSANE, PLS TAKE ME TO NEAREST MENTAL INSTITUTION"

>>1219830
If there's a better way I'd love to do that instead, just gonna need to know what that is. Most of what I've got are things like this exact USB port on the right, where the only things keeping the rear from lifting up are the solder joints, which are starting to fail and disconnect. My thought was to build some epoxy on the top and sides of the connector to give it some mechanical strength, and then fix the solder joints. Some of the fixes for friends include laptops that have loose power connectors after being physically damaged by drops, where the same type of issue is present.

How would I go about attaching a smartphone battery to a PCB that would require it to be attached to regular positive-negative wires? I'm trying to replace a battery like this with a Galaxy Note battery, but I don't know how to go about dealing with the lack of wires on the Note battery. The closest I've been able to gather is that it probably involves a JST connector, but I don't know how to go about applying it.

Sorry if this is a stupid question, I know very little beyond how to solder.

>>1219840
The smartphone battery will have some sort of plug/socket thing on top, right? Unless you want to fabricate something out of little pieces of copper sheet, I'd advise looking around for a plug it will fit into. Though if you don't mind sticking some wires/plates in there and fixing them in place permanently, that always is an option.

>>1219848
Once I've got a battery connector, what's step 2? Solder the wires to the connector or something?

Alright, that took way too long. It's a good thing I enjoy tedious algebra. Bike anon here. So I've been looking at driven LC oscillators for the transformer because there's nothing quite like a sinusoid, but figuring out how to change the frequency (doing so changes the EL wire colour by a bit) was racking my brain, because to use a normal little trimmer cap would require a massive inductor. So I decided to replace the capacitor with two different capacitors with a potentiometer in between, so changing the potentiometer would change the total reactance. A couple of hours later and I got pic related out of it, where frequency in hertz is on the y axis, and resistance out of a 20k pot in ohms is on the x axis. While it's not very linear, I think I should be fine with it. Now the problem is figuring out a transistor-driven oscillator circuit that works with it.

>>1219854
Yes, but you'll need to know the pinout of the battery/connector. Some batteries say on them, but if this is to replace a 2-wire battery and not one with an extra wire for temperature sense or cell balancing or something, you'll probably just need to know the voltage outputs. Your $10 multimeter should do fine with that, provided it's set to measure voltage and not current. Google also works. Be careful not to heat up that plastic too much when you solder, I very much doubt that it's made of a decent thermoset, and will probably be prone to melting, thus letting the pins sag out of alignment.

I'd watch out for the two cells having different levels of protection circuitry (the thin board visible beneath the Kapton tape on your picture), so google into that as well.

>>1219860
Thanks, that's very helpful.

I'll be using this connector, and I have to ask: there are three pins, but only two wires. Does it matter at all if the middle pin is ignored during the soldering process? Will that affect anything? This is of course assuming that the voltage is correct and won't cause any issue, which I've checked and confirmed.

And perhaps a silly question, but what's the best method for making sure that the battery's contacts remain firmly in contact with the connector? Tape, or some sort of glue?

>>1219878
It almost certainly won't hurt the battery if the middle pin is left disconnected, but the battery's protection circuitry might simply let no current pass or mess with you in that respect, in which case you might have to tie the pin to a rail or set it to a voltage with a voltage divider. You should be fine though.

So it's one of those pressure contacts as opposed to one with slots? Yeah, that would be a problem. If you've got the patience I'd mock up a box out of wood or metal to hold the two together, but otherwise use tape if it's temporary, glue if it isn't. For the glue, something like hot glue or silicone are probably ideal, because they'd be able to have a significant volume to grip the connector. Epoxy is also an idea.

Where do I get a 1:50 transformer?
I want to build a charge pump with it, but that's the only part I can't find.

>>1219880
Not easy. Either miraculously find one, rewind one coil of an existing transformer, wind a complete transformer from scratch, or see if you can order one. What kind of charge pump uses a transformer anyways?

>>1219883
I'm not sure if charge pump was the correct term, but a charge pump works by inducing a voltage on the coil by turning it on and off. By using a transformer I could induce a higher voltage on a second coil and use that one to charge up a capacitor instead of only a single coil.

But you're making a good point here, I could just remove the transformer and go with a classic charge pump design. Probably much easier than getting my hands on a fitting transformer.

>>1219880
Unwind a few coils off the secondary of a 220Vrms-to-5Vpeak transformer. Better hope it's not potted.

>>1219886
I think you think a charge pump is a boost converter.

>>1219886
Are you using this like a flyback transformer (non sinusoidal)? In which case it might actually be arbitrary what size transformer you need, I'm not too sure if the inductive spike you get by suddenly open-circuiting the transformer could cause a voltage spike larger than 50*Vp in the secondary. You could try using a single inductor and an appropriately sized zener diode (or several) if you just want to charge up a capacitor as a shitty boost converter, though it will probably make more waste heat.

>>1219889
I think he's referring to a flyback converter.

>>1219858
You can't tune the resonance that way, it's not how the physics works. Suggest you simulate it. Intuitively I'd say that the frequency will barely change until the series R comes very close to zero, it acts more like a switch. Your graph looks like x^2+y^2=r^2 but resonance listens mainly to w^2·L·C=1.

The colour change is not noticeable without a close-by ref colour, not worth the effort. What does change drastically is the brightness because current depends on frequency (elwire is capacitive).

A critical look at the datasheet (thanks for the link) will tell you that this type of elwire is not bike-compatible. Better wait for RGB ledwire.

>>1219896
>not bike-compatible
How so? I know it's not UV rated, but I figured I'd give it a layer of epoxy or polyurethane or something to amend that. You're right about the brightness change, which doesn't bother me too much because I'm relying on the LEDs for illumination. Frequency alteration is for the most part a pointless gimmick, but if I can easily implement it inside the sine-wave generator that I'll build anyway I don't see why not. I'll probably go with the white EL wire if I do find a circuit to make the alterable frequency work.

Now when I say I did the maths, I mean I did the maths. From z = X_L+(X_C1+R1)(X_C2+R2)/((X_C1+R1)+(X_C2+R2)) all the way to booty town. The end result gave me exactly the expected frequency (w=1/√(LC)) when the potentiometer was turned all the way to either side, (1mH inductor, 160µF cap gave ~400Hz, 2.8µF cap gave ~3kHz), so I'm assuming that the maths are right. Check it yourself if you feel like a noose isn't quite your speed.

Now this applies for the circuit in pic related, so whether I can get a transistor-driven circuit to use it (each arrangement actually has two different resonance frequencies, the other being in the fractional hertz) is another question entirely. The circuit I just tried wouldn't let me put any kind of resistor in series with the single cap, so I should find a design that does.
(the png was too big, so have a gif instead)

>>1219300
>the rear lights strobe
The ideal frequency is exactly 0hz because its much harder for someone to track and follow a flashing light that is moving while the light is off. It makes it much more difficult to estimate your speed and anticipate your position.

>>1219943
Not a slow flashing. Either fast (~10Hz) or not at all. Ideally I'll make only some of the LEDs flash, or even better make them just dim periodically instead of turning off periodically. If I can change the degree of dimming and the frequency of oscillation then I should be able to get the correct balance between "eye-catching", "distinctly a bicycle and not a motorist or pedestrian", and "seizure inducing". My motto being, if you don't know what you should permanently set it to, it needs more potentiometers.

I also need an emergency off button for the blind attack:
https://youtu.be/m8JunMGb0x4

>>1219923
>(w=1/√(LC)) when the potentiometer was turned all the way to either side
No doubt. What is the Q of the circuit when it is not? Will any type of oscillator run stable when the pot is in between?
>another question entirely
Maybe, but the only diy-related question in this context.
>wouldn't let me put any kind of resistor in series with the single cap
Does it tell you why that is?

>>1219880
>Where do I get a 1:50 transformer?

you will find that CCFL Inverter transformers (the lil ones that generate 800V for LCD backlights) typically have 50:1 or 75:1 ratios. they are easy enough to pull from dead monitors and sidewalk TVs.

>>1219959
>correct balance between "eye-catching", "distinctly a bicycle and not a motorist or pedestrian", and "seizure inducing".
i already told you the correct balance is 0Hz.

eye catching is not a good thing, drivers should be constantly looking at everything around them in order to identify anticipate and react. having a flashing/pulsating/fading light in the corner of their eye when they are trying to look out for other hazards is extremely shitty.

just because cunt drivers say they didn't see you because you weren't lit up like a Christmas tree when they knock you down doesn't mean they weren't busy texting or something else equally stupid.

flashing it has to be brighter for the same perceived intensity which absolutely ruins your night vision.

i'm not going to keep arguing with you, i'm just telling you its a really shitty thing to do.

>>1220018

this guy is 100% wrong. a tiny red light could be mistaken for anything, even a dashboard reflection, whereas a flashing red light screams at you, ''CYCLIST HERE, PLS DONT KEEL''

>>1220028
I really hope you don't drive if you are that retarded.
Even if you were stupid enough to mistake a red light on the road outside of your windscreen for a light on your dashboard how would it flashing make it any better?
Identification of cyclists is a non issue. To anyone. Nobody has ever seen a light on a bike and hit them because they didn't realise it was a bike. Either the bike wasn't ' there too be seen' or the driver wasn't driving with due care.
If you have a license to drive with the best will in the world I suggest you surrender it voluntarily.

hey ohm,
I want to use the box in pic related to house some electronics.
for in and outputs of wires of the box and a fan on the side I will need to (probably) drill some holes on the side.
The box is reinforced with fiberglass.
Is it still safe to drill? Do I need to wear a mask? And do I need to clean the box?

I know this isn't exactly /ohm/, but you guys probably worked with housing electronics. So maybe worth a shot

Rate my design please.

I'm planning to build a modular coil gun. At the end I want to be able to chain as many coil modules as I want to increase its power.
I'm also using the JP headers in place for the MOSFETs that I haven't decided on yet.

Main module:
While off, the gun discharges the capacitors through an 1M resistor. For that, I added the PNP transistor that is controlled by the 5V power supply. Once that one is off, it discharges the caps, otherwise not.
While on, the gun uses a simple charge pump to charge up the capacitors. To ensure safe charging, the ATTINY85 reads the voltage and stops charging once ~95% capacity is reached.
Pin 6 produces a PWM signal for the charge pump in place of an oscillator.
Once charged, the READY signal is on, indicated by one or more LEDs. When the trigger is pressed, it sends the FIRE signal for a second.

Question: Is it fine if I just build my own coil for the charge pump? If so, how many coils should I use here?

>>1220053
Coil module:
While FIRE, an IR LED lights an IR receiver. Is a bullet between the receiver and light and FIRE, the coil is enabled until either !FIRE or the bullet moves out of the way.
READY is used for an LED indicating READY state.
All the signals are passed through to the next coil module. This way I can chain as many as I want, while they're all working in parallel and sharing their capacitors.

>>1220052
You probably should wear a mask of course but if it was me and my mask was further than arms length away I probably wouldn't bother.
Try to avoid breathing it in and just vacuum it up when your done.

There are punches you get for metal enclosures that cut a real neat hole for conduit entry that you tighten with a hex key but you have to drill a small pilot hole first anyway so that's not much help I guess sorry.

>>1220054
Put it in the box "to house some electronics".

>>1220052
>The box is reinforced with fiberglass.
use your shittiest drills if you don't know how to sharpen bits.

yeah it's safe, do it outside and as >>1220056

says, vacuum the inside & out

>>1220057
I don't understand.

>>1220056
>>1220071
thx guys, the box itself isn't steel btw, it's plastic reinforced with fiberglass. I'll also tape the plastic before I drill.

>>1220057
give me a break, English isn't my first language and I'm tired

>>1219923
>gif
>not animated

>>1219959
>>1220018
Commercial bike flashers in flashing mode run at about 4-6Hz fwtw

>>1220099
>it's plastic
yeah we got that, fibreglass is rough on drill bits. take it nice and slow, i know people say let the drill do the work quite a lot but really, let the drill do the work.
>tape the plastic
fuck why didn't anyone else think of that. genius!

>>1220106
uk the legal range is 1-4Hz.
i haven't seen a single bicycle with a flashing light come anywhere close to that.
more often than not there is more than one flashing light rendering enforcement a complete waste of time.
cyclists truly are a law unto themselves.

>>1220053
Sorry mon frere but your schematic is basically illegible:
>drawn right to left
>using connectors in place of the correct schematic symbols
>using EAGLE instead of KiCad
Even a draft on paper would be preferable.

>>1220106
>>not animated
It's quite common for circuit diagrams to be in .gif format, and it was a format that let me keep the edgy background transparency when the .png was too big.

>>1220110
>the legal range is 1-4Hz
In what country? I'd better look up my own legislation too.

>>1220147
>It's quite common for circuit diagrams to be in .gif format, and it was a format that let me keep the edgy background transparency when the .png was too big.
No, seriously, m8, this is super-prime b8, usually found only in the backrooms into which not even a shot of black and a secret password is enough to get you admitted.

>>1220147
>In what country?
If you had quoted the whole line you would have your answer!

Didn't you see the frequency when you were checking the regulations for brightness, angle etc? Or you didn't bother with that?

Of course the optimum frequency is 0Hz

>>1220153
>If you had quoted the whole line you would have your answer!
I thought he said "uh" not "uk", my bad. The last time I checked my country's laws, it just says the lights and reflectors need to be visible from 200m:

Compulsory equipment:
>red or yellow rear reflector that is visible from a distance of 200 metres when light shines on it.
>Good brakes on the front and back wheels.

When cycling at night or when visibility is poor, cycles must have the following:
>One or more steady or flashing rear-facing red lights that can be seen at night from a distance of 200 metres
>One or two white or yellow headlights that can be seen at night from a distance of 200 metres. Only one of these headlights may flash.
>Pedal retroreflectors on the forward and rear-ward facing surfaces of each pedal. If the cycle does not have these the cyclist must be wearing reflective material.

200m says nothing about how many lumens or candela they should be, so I'll have to wing it and go big. Though strapping a light to a street-light and walking 200m away to see if it is visible at least constitutes as a qualitative analysis. I think it's the candela value that matters from a distance, not the lumens, so I'll get a bunch of LEDs of different mcd values to test. It also says nothing about frequency, other than:

When considering lights it is important to be mindful that:
>Headlights should be attached to handlebars and pointing down.
>Your lights can be a hazard if used incorrectly. You must not use cycle lighting equipment in such a way that it dazzles, confuses, or distracts so as to endanger the safety of other road users. Angling your front lights down toward the road helps prevent this.

Being able to dim the EL wire should help with this, as all I really need is a visible-enough outline, not a neon lightshow.

I'll also add a buzzer to act as a horn.

>>1220161
>so I'll have to wing it and go big.
yeah fuck everyone else eh

>the candela value that matters from a distance, not the lumens
you want a wash really so lumens should be your concern

>When considering lights it is important to be mindful that:
if you are putting it on the handlebars mount it on the offside, away from the pavement/sidewalk so you don't look so small to other road users.

the physical size of illuminated area (reflector) on your cycle is far far far more important than the brightness.

given the same amount of energy a 50-100mm diameter disc is much more visible and much less likely to blind others than a 3-5mm single led point source.

look at e.g. laser based headlights on new(ish) audi/bmw, they use a high power laser but bounce it off a large area reflector rather than shooting it out directly. plus some led ring light around it too. large lit area - much easy to see.

Rate my craftmanship /diy/
Also try to find out what is it

>>1220169

>>1220175
?

>>1220169
When it comes to straight distance that I can be seen from, cd matters more. I'll also want to get a decent lumen value anyways, and that 3W 120 degree "240lm" LED should be great for that. By "go big" I meant that you can always dim a bright light but you can't make a smaller light brighter. I'll be installing proper reflectors also, but I'll put my lights behind decently sized diffusers, making the illuminated area closer to 3*3cm or larger. I agree that point sources are awful and blinding, which is why I picked EL wire over LED strips. Where to mount the light has still got me wondering because a light on top of my handlebars will take the brunt of any impact when I crash, so it will probably be under the handlebars or on the stem, hopefully that's allowed.

>>1220173
What seem to be speakers connected to what seems to be a battery, a board that has a cartridge-like pin thing, and another board that doesn't have the pins that we can't see.
Could be anything, an audio amplifier perhaps?
One meant to go int a specific system?

>>1220185
> What seem to be speakers
I lol'd
> what seems to be a battery
I lol'd harder
> cartridge-like pin thing
> board that doesn't have the pins that we can't see
Lol'd pretty hard
> audio amplifier
Well that's the part of the device. It is, um, a prototype for wifi speaker. Custom design with insulative tape and bare wires. Gotta replace speakers to something better.

>>1220173
Is it a jpeg?

>>1220200
I wonder how did you figure that out

>>1220199
You really can't expect people to be able to guess anything from that blurry thing.
But yeah, replace the wires, especially since you can easily short out the battery as it is.
Also consider a battery protection circuit if you don't already have one.

>>1220204
The next prototype will have the discharge protection, it wasn't included. I have already disassembled the current one so I won't die of battery explosion today. Thanks for the advices.

>>1220183
>when I crash
>>1219290
>The last person I crashed into
>>1218731
>Also I crash way too often

ffs sell your bike already you are clearly too disabled to use it

Going to replace the glue logic in my cp/m machine with a CPLD. I've done FPGA stuff on boards, but I'm not sure how to program CPLD's as standalone devices. Any tips?

>>1220236
In terms of hardware there's not much to worry about, get your JTAG right, set up power properly and it should just work. A lot easier than an FPGA

For the digital design the only major difference is there's a lot less capacity for arithmetic and memory, but for just replacing a bunch of 74 series glue, you're all good.

>>1220267
Thanks.
Going to be good replacing 3 chips and 30 traces with just 1 now. Shame you cant get a 5v fpga without messing with logic level converters.

Not sure if this is the right place to ask but I have what looks like a cable port in my bedroom and our modem is connected to one on the other side of my house. Can I just get a cable and a modem and plug them into the port in my room to make a new network?

Not sure if this is the right thread, but it is electronics related.
Trying to fix pic related. It's a police 8000w xq-t8626 flashlight.
The thing on the left is the battery holder, charging circuit, and switch, the thing on the right is the led part.
The positive is connected through the middle hole thing, but does someone have a clue how they're getting the other polarity?
I'm on vacation, so I don't have anything that would help with with determining it.

What are the steps in taking the next step from prototype to manufacturing? Specifically with small electronics

>>1220311

you cant put 2 modems, either cable or DSL, in parallel. only 1 can be on at a time. also, it's likely your ISP will only let you use the modem they sold you, and whose settings are hard-coded inside. ISPs have become more and more restrictive about what they will support.

>>1220320

negative is run trough the case, almost certainly.

>>1220322

finding a small chinese factory to make some samples, then the real deal. nobody assembles stuff locally anymore.

>>1220327
>negative is run trough the case, almost certainly.
I thought so too, but that just seemed like a really weird thing to do, so I wanted to check.

>>1220330
It's not weird, all cars do it.

Is there any way to 'smooth' a joule thief type circuits output? I assumed a ceramic capacitor across the output would give a nice smooth output but it didnt work. My ultimate goal is to power a 3 colour, slow flashing LED (which wont work since the led is constantly reset by the 'flickering' output)

>>1220305
>Shame you cant get a 5v fpga without messing with logic level converters.
did you even google m8
http://www.microchip.com/wwwproducts/en/AT40K40

>>1220330
>really weird thing to do
Nah, most cars have been doing that for decades.

>>1220380
>but it didnt work
Did it not smooth the output, or did it stop the circuit from working entirely? It's essentially an AC signal, so maybe adding a (Schottky?) diode to stop the charged voltage from leaking back out again and a decently sized (10µF?) electrolytic cap would work. Ceramic caps are pretty small, so unless the thing's oscillating really quickly, a ceramic cap probably won't do the trick, but the problem is more likely to be the lack of a "rectification" diode. Changing the inductance of the coil(s) will probably change the frequency, which you could try as well.

>>1220380

a cap wont work coz it charges when the output is high, but discharges as soon as it goes low. to prevent the discharge, you put a diode in series.
(you may need a schottky diode if the frequency is high. you can pull schottkys out of any old switching power supply.)

So this is what I'm pretty sure the constant current/voltage LED/battery charger is. Buying one will be cheaper than building one even if I used offshore parts, so this is probably what I'll use to power my 3W LED. I'll test the LEDs in the weekend, but until then I'm still messing about with frequency generators.

>>1220380
You need a lowpass filter. A C by itself is not good enough.

Is there a way I could find a modern equivalent to replace the 2N5953 JFET?

>>1220390
>>1220391
>>1220411

thanks all! got it working - the cap was infact discharging on the 'off' cycle.

thanks for all the help

>>1220420
Nice phaser m8. I haven't run an analysis or anything like that, but in this application a 2N3819 would probably replace (note, different pinout) with only adjustments to the bias pot. Changes to the bias circuitry and sweep generator are unlikely to be necessary.

>>1220420
>>1220428
What's a phaser? Why does it input stereo and output mono?

>>1220438
>What's a phaser? Why does it input stereo and output mono?

that's what phasor means; it's an electrical engineering term for a complex signal and this is a cirtuit that realizes the concept visually. pretty cool when it works. I built a convolution filter with a similar layout for my dog.

>>1220438
Oh, so just a vibrato or maybe an auto-wah, not a phaser?
>>1220438
>Why does it input stereo and output mono?
It doesn't. It just uses the ring of your input plug to switch the low side of the power into the circuit.

>>1220420
2N5953 IDSS 2.5..5 mA DSG
BF245A IDSS 2..6.5 mA DSG
2N3819 IDSS 2..20 mA DGS -- grouping?
BF245A comes closest and has the same pinout.

Anyone good with arduinos.
I plugged 8 switches into pins 0-7 on an uno.
I pulled up each pin but pin 1 (TX) is always low on my serial monitor.
Is this related to the fact that its plugged into a computer. As in would this still be an issue if I used an external power supply other than usb.

>>1220690
Pins are shared with peripherals.
If pin 1 is TX as in transmit that is converted to usb to allow your serial monitor to work then yeah its getting used by the serial monitor, the chip will set it up as an output to use serial output.
I guess I don't know never used arduino

For a 9V to 450V charge pump, how big has the coil to be / how many coils are needed / what diameter or length? I'd assume tiny 5 coils is just not enough.
Also, do I need to use insulated wire or not?

>>1220698
If you make a coil from un insulated wire its not a coil it's just a dead short.
Usually enamel coated wire is used because the insulation is exceptionally thin so you can get more windings in.
Retards call it ' magnet wire' because of its other important property: Being no more or less magnetic than any other wire

>>1220720
It's called magnet wire because you use it to make electromagnets.

>>1220698
ctrl-F charge pump -- 10 matches
Are you recycling your old posts?
>>1220720
It is called magnet wire (not magnetic wire) because it's used to wind electromagnets and magnetage runs on windage and amperage :)

>>1220698

A charge pump doesn't use coils. As for step up, depends on the power you need.

For low power you could consider modding a CCFL inverter :

http://www.hardhack.org.au/hv_reg_power

>>1220737
They should call it electromagnet wire then
>>1220738
Fffffgfffffffgggfffffgffgfuuuuuuuuuuuuuuuuuuuuuu

>>1220738
I'm just asking different questions that are related to the whole thing, if that's what you mean.
I really want to get this coil gun thing going, but as it turns out, there are a lot of details I need to figure out.

>>1220742
Boost converter, my mistake.
I don't want to modify anything existing, I want to put everything on a single custom PCB. It's much easier to design the case if I can influence the dimensions of the PCB. Plus, I build one from scratch which is kind of cool itself.

>>1220698
Photoflash units regularly do this from 6v or less. You can find plenty of circuits around the web.

>>1220747
You have much to learn, anon. You need to search the web more, read more, and experiment more, just like we did, rather than try to optimize circuits you don't fully understand.

>>1220747
Look at existing boost converters from AliExpress and the like, they do everything efficiently and down to a cost. A buck converter I'm going to be using has an inductor of 33µH in it, if that gives you a sense of scale, and its easily small enough to just buy instead of winding.

I heard people here mentioning that boost converters get less efficient when stepping up to higher voltages, but I'm not too sure.

>>1220756
>You have much to learn, anon. You need to search the web more, read more, and experiment more
Well yes, hence the questions. I'm pretty sure I got the boost converter theory down, it's just that I saw a bunch of DIY coils online that looked like they didn't have any insulation at all.

>>1220762
That helps a lot, I really thought I'd need something bigger and need to do it myself. I found some inductors for cheap on ebay, I'll just try and slap everything together to see if it's working well enough.

>>1220762
>I heard people here mentioning that boost converters get less efficient when stepping up to higher voltages, but I'm not too sure.
True. However, there is also the flyback topology which can be made less inefficient for a given step-up ratio and can reach higher step-up ratios compared to a boost converter.

I ran my tool battery to cut off in my radio then forgot about it for 2 days
Radio drained it too far and now my battery chargers will not accept it even after opening it up and forcing it to charge from my bench supply to 19v (full charge is 21v)
I cant even get the thing to allow a tool to run from it

Any ideas on how to get this thing rolling again?

>>1220828
Did you try looking up the protection chips on the board to see if they might have gone into a lockout mode and need to be reset?

>>1220828
Does it hold the bench charge? Maybe its actually fucked.

Someone help a simpleton understand this circuit diagram. It's a 32x12 LED grid. I understand the columns anodes are wired together and the rows cathodes are wired together. What I am unclear on is the 50 pin breakout (actually just a 2x25 ribbon cable). Does the 1st pin get soldered to column 0s anode? And then the 2nd pin to column 1s anode and etc? And then once I get to pin 33 is it soldered to row 0s cathode? And if I have that right is it soldered into the preexisting solder (where the wire was soldered to it) or is it soldered beside it but still on the anode/cathode? Sorry if retarded.

>>1220856
Any two nets with the same name are connected. They just didn't draw the wires because it would be a mess.

>>1220311
You can use a MoCA adapter at each end to run ethernet traffic through. Kinda expensive though, and often slower than cheaper powerline adapters instead. WiFi may also be a reasonable option. Best and cheapest way if you can is just run an ethernet cable.

You can't just add a second modem, you only get one connection to the ISP.

>>1220859
Oh I feel stupid. So pin 1 would be soldered right on the anode where the wire connecting the columns was already soldered. Thanks. Last question, I assume it doesn't matter which row i solder pin 1 to as long as its column 0, right? Putting the ribbon cable right down the middle would probably be the most efficient way to do it, then just have the last 12 pins stretch across the grid and solder it to the rows

>>1220862
Doesn't matter as long as it's on the same net.

>>1220851
It will, battery is only half a year old
>>1220830
they covered all the chips in some bullshit so you cant read the numbers

>>1220879
I have a shitty solder job done because my solder is apparently bad. I tried using a different solder to patch it up but its still a patch. Can I use a heat gun to fix it? Never used the hot air gun before, just bought in case I ever tried smd.

>>1220879
Boo hiss.

>>1220964
You could try it. Pull it away as soon as the solder goes liquidus, of course.
Better might be solder sucker/desoldering braid/try again.

Does anyone know of a decent TRS plug with an integrated strain relief? I can't seem to find ones that are good. Plenty of chinese junk, but they all suck.

>>1221008
You're better of buying a good cable and using the end of it as your socket. But it's a right pain to get them panel mounted.

At least that's what I do with TRRS sockets. You can probably buy a panel mount audiophile socket from an audio store, though they typically stick to 1/4" sockets instead. You could convert all your devices to 1/4" sockets and jacks, but you can't get 1/4" TRRS if you have a smartphone and TRRS earphones to go with it.

>>1221008

do you have $4.50 to spend on each one?

but, really, i've never come across a replacement plug that DIDNT have strain relief, even the cheapest radio-shack type ones, so your story is suspicious.

>>1221065
Maybe he lives in Aus/NZ? Not sure about the ones with the spring, but the others definitely don't have them. These aren't classified as "replacement" jacks however.

>>1221071
Add your own gland and layers of shrink tubing, I suppose.

>>1221071
The ones in the pic are the cheapest dog shit jacks that come out of China. The plugs are usually undersized, the threads are shit, and the plastic is garbage.

>>1221090
I can confirm, though the slimline ones do have an anodised aluminium body. The plastic between the connections still melts too easily though.

What's generally worse, using a comparator as an amplifier (not for audio, just to get a triangle wave's amplitude up), or using an op-amp as a comparator? Am I correct in assuming the comparator's lack of stability will make it less ideal, or does that only apply to signals with ~zero amplitude for a significant amount of time?

>>1218431
topkek

>>1221138
A quick spice sim tells me that a non-inverting amplifier made of a comparator is not a wise idea and doesn't work for shit, though my voltage rails may have been too close together.

Now how would I go about amplifying a triangle wave to as close to the rails as possible and while delivering almost an ampere of current? Will a simple common emitter amp do that for me, or do I need to find a good enough op-amp? Could I run a Schottky diode and capacitor to keep track of the waveform's amplitude for feedback purposes?

I want to make my own USB cables, cause aesthetics and shit. (I Love nylon braid and braided cables)
What's the "scientific" name of the cable usually employed to make USB cables? (Shielded with four cables inside)
Is there single conductor cables with shielding? What its the scientific name of that cable?
Can use unshielded cables to build USB connectors?
I'm no planning on making USB 3.0/USB 3.1 shit for now.

>>1221170
>single conductor cables with shielding
Well technically coax, but the shielding is used as the ground conductor. What I think you're looking for is "4 core screened cable", pic related is the first thing that comes up on images.

When you do come to doing USB 3.1, Cat6a cable might do the trick, as long as you're able to use its shielding as its drain wire itself.

Building a cantenna to tap into neighbours wifi until I get my own at the new place I'll be staying at.

Should I get a wifi dongle with an antenna, or isn't it needed in such a setup? I'll be mounting the dongle in the lid.

>>1221178

>>1221181
I guess my question is actually whether an antenna on the dongle would improve the signal or not.

>>1221184
look at the picture >>1221181
in the picture >>1221181
end of dongle is placed at focal point of reflector

>>1221170
Both USB and Ethernet cables have a twisted pair structure and shielding is optional. For USB only one pair needs to be twisted. You can follow your special needs and braid as you must as long you don't fup the Z too much. USB also works well with 100Ω LAN cable. AWG26 patch cable has 4 pairs of which you need two, maybe your favourite colour is among them..

My dog ate part of a remote gate opener, this is what i was able to rescue.
I guess the best choice now is to send it to repair or to buy a new one but sadly that is quite expensive in the place where i live.

So here are my questions:
>Are the orange circled things mini resistors?
>Are the blue circled things fuses?
>What is the red circled thing?
>Do you think i could repair it using 'normal' sized resistors and/or other components if i have the original design of the circuit?

Cont.

Seems to be she chewed the SAW resonator a little bit aswell.
According to how it looks like, do you think i need to replace it aswell?
Hope you can help me with this, it would be awesome.

Ask me if you need more pictures and thank you very much.

>>1221300
Oranges are resistors, but the 0Ω resistors are just there to bridge over other traces. These "SMD" resistors display their value in a particular way, so if you want to replace one look that up. The grey squares are not resistors but capacitors, value unknown but typically fairly small.

The little glass packages in blue are probably diodes, but there's no real way of telling what sort of diode they are.

The three-legged component is almost certainly a transistor of some kind, but again there's no way of knowing what sort unless you're lucky enough to have one where the name is printed on top. It is probably a power MOSFET of some kind.

>>1221305
That can looks slightly dented, but it's probably ok.

Can you find a picture of a completed board online? You're probably only missing a few components, which could be fixed by sanding down the conformal coating on the traces near the edge and soldering components and "bodge wires" onto them. If the image shows a board with a mystery component or two, you'll probably have to trace the board out so as to be able to infer the values of the components by the circuit's function. For example, two of the traces going to the transistor are thicker than the other, so it's a good bet that the thinner trace is the gate, that the pin next to it is the drain, and the pin opposite both is probably the source of an N-channel FET. Knowing what's written on that little 8-pin package is also crucial for this sort of circuit backwards engineering.

Are you sure you can't just wait until it comes out the other end of the dog?

>>1221156
Power MOSFETs, probably.

>>1221320
>a power MOSFET
in a remote? no, Q1 is the oscillator/transmitter but part of the 'inductenna' (http://i.4cdn.org/diy/1501176894741.png)) has been chewed off. the 8 pin chip is the data encoder.

>>1221320
Do you know how could i remove the green thing from the yellow labeled copper tracks if the dog defecated that spare piece and i had the chance to solder it again?

The damaged can's name is 'HD R433M', looked for it at the internet and it said that the component was a 'Surface Acoustic Wave' resonator.

>>1221355
The 8-legged integrated circuit's name seems to be 'HT6P20B'.
Hope i don't to reprogram it...

>>1221300
>>Are the orange circled things mini resistors?
>>Are the blue circled things fuses?
>>What is the red circled thing?
(component id)
D1 - Diode
C1 - Capacitor
R1 - Resistor
F1 - Fuse
T1 - Transformer
L1 - L= Inductor
JP1 - Jumper (wire)
Q1 - Q=Transistor
U1 - U=Integrated Circuit

more complete list here: http://en.wikipedia.org/wiki/Reference_designator

>>1221355
Fine sandpaper lightly applied removes the solder mask, and the copper too if you overdo it.
I'd be a bit concerned about the doggo having a sharp piece of laminate moving through its digestive tract. Would the services of a veterinarian be appropriate?

>>1221361
It's already programmed and wouldn't lose that programming just from a doggo attack. If you can't restore the original RF circuit, you could probably attach one of those chink SAW transmitters to its output to restore the function.

>>1221363
I've been finding small parts of that last piece around the 'disaster zone'. Seems to be she just chewed it without eating anything. A good relief..
The only thing i can't find yet is the head of the infrarred diode.

>>1221373
You appear to have the whole board there, then. IR diodes are easy to source and replace. Looks like you lucked out, mate.

>>1221373
If it has a 433MHz SAW it has no IR diode. That's just a normal indicator LED that lights up when you push a button, see pic. Looks like you get that thing restored. The green solder mask can be scraped off with a razor blade or a cutter and a piece of wire will be needed to bridge the gap when soldering. Good luck.

>>1220405
So I bought a couple, and it appears they don't work properly at 5V (even though the website told me it should). I can adjust the voltage from ~4.3V to ~1.2V, like I'd expect from a buck converter, but the current control doesn't work much at all when I short-circuit it through my ammeter or through 10-20Ω of resistive wire.

My guess is the ~5V isn't enough for the 5V LM7805 voltage regulator to power the op-amps properly, which makes perfect sense. Do I need a boost converter now? Because having 3 switched-mode power supplies in a row feels fucking stupid.

Can I mod this to work off 3V op-amps with a 3V regulator, or will that just make it awful?

>>1221479
If you'll be running it from 5V regulated, just replace the 7805 with a short from in to out.

I'm trying to use a delay function on a PIC microcontroller. The delay function is included in one of the heads I'm using. I think it needs some specific datatype, because if I give it an int or a float it gets the delay totally wrong. I'm not totally sure of the pattern but it seems like it uses some arbitrary value for the delay no what number you give it. However, if I type a number in as an argument for the function call without casting it as a variable and giving it that, it works (albeit off by a factor of 2 or 4). Can anyone explain what's going on, or how I can determine the proper datatype if I can't find the header file that documents it?

>>1221508
>The delay function is included in one of the heads I'm using.
>I can't find the header file
if the compiler can find it so can you! come on anon.
the new mplabx xc compilers have __delay or something macros but you have to define your clock frequency so it knows what to do.
i guess you aren't using that then?
you say you are casting so i hope you are at least using c and not asm but the way you say casting as a variable we might be talking at cross purposes...
is it off by a factor of 2 or 4 or neither or both? varies? can you figure out why and when it changes?

i hope you aren't just in asm and doing something silly like
>movlw 'variable'
>call delay

need input!! show us what you have.

>>1221501
Holy shit, you're a genius, why didn't I think of that. Can I put a bodge wire across the regulator and not fuck it up, because it's SMD and surrounded in other shit and I'm not at all confident that my conical tip can get enough contact to desolder it, and I don't have any copper desoldering braid.

>>1221513
It should be safe to do that but I'd probably cut the ground lead first just to keep it from consuming current or causing trouble.

>>1221510
Found the header file. Yeah, I'm working in C.

Delay code is at https://pastebin.com/vEYGsZeJ
I need to find a way to get that to accept floats. I'd like to mimic __delay_us and simply change the "unsigned long" to a float, but I'm not sure that'll work since I don't know what _delay is. There aren't any more mentions of _delay in this file, and I'm not sure where to look next to find the _delay function that __delay_us calls.

I defined my clock frequency already.

I think __delay_us gives the proper delay, maybe half as much, if a non-decimal number is provided. I tried giving decimal numbers less than 1 and it doesn't work at all.

>>1221531
>doesn't work at all

I mean, it delays, but I think it's the same problem as giving float variables to the function, where the delay is fixed at some arbitrary number no matter what I feed it.

>>1221534
>floats
>on a pic
You could just multiply your fractional seconds by a million and use __delay_us, couldn't you?

i'm very new to electronics and i'd like to do a small project, i want to mount LED lights on a hardhat and have a battery pack and a dimmer switch on the hard hat, but i'd need something VERY bright and perhaps a housing with a reflective insides, is there a special kind of LED i should search for? or a good place to purchase housings or casings?

>>1221540
It's fractional microseconds. I'm trying to control a an individually addressable LED string that's set by timing the high and low bits differently. Shortest time period I need is .4us. It's cake for the mcu since I'm running it at 20MHz, but the library doesn't seem to innately support it.

>>1221531
ok so what happens is some compile time magic, all of these delay functions don't really exist, when you compile the code the computer generates a delay loop in assembly and inserts it instead.
so if you think about it you can't use variables because the delay is generated at compile time, the value of the variable isn't known until runtime.

you can see that __delay_us, __delay_ms are just macros that calculate the number of cycles passed to _delay. _delay is intrinsic, there is no source code available, its compiler magic, do not question it.

the calculations shows divide by 4000.0, the .0 should make the calculation float compatible? so you should be able to use floats BUT the final outcome number passed to _delay will be rounded to a whole number because _delay delays for cycles, these are atomic, you can't delay half a cycle thats madness.

have a read of the compiler manual, search for _delay etc. http://ww1.microchip.com/downloads/en/DeviceDoc/50002053E.pdf

so variables won't work, floats kind of should.

what you ahve to do for variable delays is get the old pencil and paper out and write some inline asm to do a busy loop and work out how long it will take based on how long each instruction takes by reading the datasheet.

e.g.

delay: movwf counter
start: nop 'nop does nothing, burns one cycle
nop
nop
decfsz counter
goto start
return

you have to do the maths though because for example nop uses one cycle but decfsz is a branch and uses either 2 or 3 cycles depending on the outcome of the test. gosh this takes me back.

of course if you wanted to be boring you could just loop as many times as you needed through the __delay_xs(xxx) with xxx as a constant yeah?

>>1221546
Oh, those WS2812 bastards. This is the part where you would do better to write some asm to do your own bit timing, or see if someone has already done so.

>>1221548
>loop as many times as you needed through the __delay_xs(xxx) with xxx as a constant yeah?
Only if you don't mind to add the time it takes for you to loop outside the __delay loop.

>>1221548
Oh, that's interesting. That explains why changing the variable declarations didn't change the delay timing. I tried passing some decimal numbers in the argument and it didn't seem like I was getting the right delay out, but it was after I was already pretty frustrated so maybe I messed something else up.

Would decrementing a number in a C loop work equally well as assembly? I have a tolerance of 150ns. If I did my math right, one instruction cycle takes 200ns so that's pretty scary.

>>1221550
I liked it for its simplicity until I had to actually interface with it.

>>1221550
>to add the time
who do you think you are my dad or something gawd. of course you are correct

>you would do better to write some asm
yes of course this is (always) the best solution. and the most fun!

C1 is supposed to be an electrolytic. Does my polarity look right?

>>1221550
>>1221557
Oh, and I just realized that since each instruction cycle has to be counted for these things to be timed right, setting the outputs is going to mess up the timing between transitions. Do I need to do that in assembly, too?

I'm strongly considering switching to a rectangular RGB display if I can find one that's cheap and reasonably big. I was just going to make a rectangle out of the strip anyway.

>>1221556
>Would decrementing a number in a C loop work equally well as assembly?
not really, you don't have any control over what the compiler hands you at the end of the day, it could even change between compiles (it won't (shouldn't?)) and you wouldn't have a leg to stand on because its allowed to do as it wishes, it can throw in nops just for fun (actually for reasons) and you have no way to stop it.

you write it in assembly its what goes onto the chip, asm is just mnemonics for the instructions themselves.

>200ns
for 20MHz yeah that is spot on, 4 clocks per cycle. or whatever the nomenclature is.

>>1221560
>to be timed right
all instructions take time, including setting ports and pins.
if i were you i would be writing the bit banger entirely in asm. get what data you want lined up and ready in c then push it out to the pin with asm.
are you uncomfortable with assembly? its fairly trivial in small doses, this would be a great project to start with!

just be aware if you want any kind of decent rate your pic will be burning practically all of its time updating your display and not have much time for anything else, you might run into problems working out what to pump out to the display in the first place! if its set and forget thats another story.

>>1221559
what is the offset of V1, half of the time your cap is going to be getting its arsehole reamed.
are R2&3 supposed to be connected to anything?

>>1221560
Unless the LEDs will tolerate holding a high signal between bits, you probably should set up a framebuffer in RAM and spool it all out at once. Hard, but possible!
Another option is to abuse the SPI or USART interface by sending bit patterns that approximate the bit high/low times required.

>>1221559
Why not use a ceramic?
I presume there's a junction missing at R1/R2/R3, in which case you have drawn it backwards.

>>1221564
I'd be happy with 10Hz on the LEDs. I'm planning on divvying the strip up into 8 segments of 30 LEDs so 1.25*24*30+50us=950us. So that should be fine.

I haven't done anything with assembly in the past. I guess this is a good way to get my feet wet, but I was hoping I could quickly move on to the rest of the problem. I'm making an audio spectrum analyzer and the math behind the frequency analysis and binning is much more interesting to me than messing around with the display.

http://stevephillips.me/blog/using-ws2812b-pic-microcontroller

This guy says it's actually a lot simpler than the datasheet would have you believe. I only need to care about the time that it's high, which is quite a relief. I'll give his method a shot and if that doesn't work, I'll try my hand at assembly.

>>1221522
>cut the ground lead
I think I'll desolder the pots first.

>>1221559
Don't opamp/comparator inputs act like massive resistors to ground? In this case I'm not sure what ground means, when you've got R4 and 5 to define a virtual ground at the positive input V+. But assuming V+ = V-, the right side of the cap will be biased at 2.5V, while the sine wave oscillates between 5V and 1V. This means the capacitor might end up getting biased both ways so you should go for a bipolar cap. But in any case, the average voltage at the left side of C1 is 3V, so it probably is in the correct orientation.

>>1221568
>Why not use a ceramic?
10mfd ceramic?
how large are they?

>>1221574
I can flow the solder on the single front lead/pin, but can't seem to bend it back with either a thin screwdriver or a pair of pliers. Any tips that don't involve running two irons at once or heating up one set of pins to 400° while I melt the others?

>>1221566
>are R2&3 supposed to be connected to anything?
>>1221568
>I presume there's a junction missing at R1/R2/R3
Fuck me, I need to get these pcbs fabbed again now. haha, shit!
>>1221574
>This means the capacitor might end up getting biased both ways so you should go for a bipolar cap
Will do, didn't have any ceramics around so I thought I might force an electrolytic. I won't.

>>1221603
Well you can use a bipolar electrolytic capacitor.

>>1221574
Oh dear oh dear, I didn't expect a SOT-223 for some dumb reason. With no desoldering braid I'd get the whole thing warm and then twist it with some sturdy, disposable tweezers, or look for traces that could be cut with a #11 blade.

>>1221597
5mm encapsulated packages are available, pic related. The electrolytic will be fine as long as you put + toward the higher end of the circuit, which will be the virtual ground at R1/R2/R3 rather than the ground through the external equipment.
>congratulations! adding missing traces is a rite of passage

>>1221065
Nah, I was looking at Digikey and all the ones they had in stock didn't have strain relief.
>>1221090
True story, and those are what I'm trying to get away from.

>>1221607
>SOT-223
Don't worry, nothing a kitchen knife and hot iron couldn't solve. I feel like a druggie. Though I did split the SOT-223 into 3 pieces, one of which may or may not have been the silicon die. But she's clean (enough) now, and I only have one burn/blister to show for it.

I'm going to make the bodge wire with some 1mm solid core wire from a 5-core grounding cable, but to clean up the area I wove my own solder wick, and it worked incredibly well. I should make a rope-twisting jig for this!

Unfortunately I delaminated a trace when reefing on the iron and knife, but at least it was the V+ trace not the op-amp+ trace, so that just means the bodge wire will be a little longer.

>>1221627
Oh shit, I delaminated the other one too.

>>1221630
Nothing like solid copper wire for circuit aesthetics. The solid bodge wire should keep that trace in place, but when I get around to putting epoxy or silicone on the pots to stop them from moving with vibration, I'll want to glue the traces down too. Now to put the pots back in place.

>>1218431
What are those little things for? i accidentally unwelded a little one and is a pain in the ass to weild it again

Probably a stupid question but I want to use this Arduino bluetooth shield with a 3.3v Arduino (the M0 Pro). The shield has a switch marked VDD, with the two options being 3.3v and 5.0v.

Is that what this switch would be for, for specifying the input voltage?

I looked through all of the documentation on Linksprite and can't find a concrete answer.

>>1221634
HAWT
May I recommend 30awg wire-wrapping wire like the pros use? It fits in vias better, for one thing. Also, something less aggressive will work fine for the pots in case you should have to adjust them for drift over time. Nail polish is good and cheap if you don't want to shell out for Loctite.

>>1221636
It's a capacitor, m8. It resists changes in voltage. Probably pretty important. Use a through-hole component lead to pin it down while you solder both ends.

>>1221638
Looks like it. The switch sets the power supply voltage of the shield, which limits the I/O voltage accordingly.

>>1221636
Can't see what the via is connected to but it's probably for decoupling the V+ for the IC on the other side.

If you can't get solder it back on, get a cone tip for your iron and git gud.

>>1221640
>Looks like it. The switch sets the power supply voltage of the shield, which limits the I/O voltage accordingly.

Thanks man

>>1221640
>Nail polish
Perfect idea, I was just thinking that even silicone might be a little difficult to remove if I want to adjust the pots.

Is there a liquid that will dissolve solder but not copper? I was just thinking that if you could drop a bit of this on a board you could clean up all the excess. My two thoughts are mercury, or some sort of acid. Maybe gallium would work, or some liquid gallium alloy like Galinstan.

>>1219303
What do you use to program PICs? The PICkit 3 seems to be the method of choice.

>>1221644
Doesn't sound like anything I'd want to handle. Lots of flux and hot air or a touch with a wedge tip will help it look more or less like it's supposed to be there. Solder wick will get it off the board.

>>1221647
Last time I needed to, I built a parallel-port programmer out of stock on hand that was right terrible but did the job. My SOIC-8 "socket" was a home-etched board with a chip clip (like for crispy snacks bags) holding the PIC in place.

>>1221647
You could make your own hardware, which is extra easy if you pick a PIC with built-in USB interface.

>need inductor to finish theremin
>search digikey for 5mh inductor
nuthnin.pgn
>In a fit of despiration search mouser
out of stock

Where do you guys get your inductors?

>>1221636
>unwelded
>weild

this thread....

>>1221662
Millihenry? Are you sure? 4.7mH is the nearest "standard" value. Perhaps that would work just as well. I see Mouser has a hundred types of that value in stock.

>>1221654
>a home-etched board with a chip clip holding the PIC in place.
That's all kinds of kludge.

The LED driver board works again and is showing the correct behaviour on large loads, but on small loads it stays on constant current mode regardless of the voltage setting, which doesn't bother me either way as I'll be using it to power one 3W LED. I actually bought another because I thought they looked so damn useful that I'd just go ahead and make a bench top-power supply with one, with constant current, constant voltage, and a Li-ion battery charging mode that I haven't figured out yet. All it needs is a dial each for voltage and current, jumper wires to panel-mount LEDs, and 3 potentiometers. And a binding post or two.

So on that note, are there any good looking panel meters out there in the 0-30V range and 0-3A range? A look at the usual suspects just gives me plasticky dials or matte black seven segment displays, and I'm looking for something a little more artistic.

>>1221671
>good looking panel meters
Or maybe I've found a reason to buy a few of these...

>>1221667
Thanks dude!

Didn't even think 4.7 would be that common.

Why people use a nonwhole number as standard I don't know

>>1221676
E-series preferred numbers matey

>>1221676
Each common number is the "12th root of 10" times greater than the previous one, and there is also a series of twice as many less common ones which are the "24th root of 10" times away from one another. All rounded to 2 sig figs, of course. It's theoretically a better way of dividing up the values because it stops you from getting the discontinuity of going up by one ohm at a time util you hit 10, when you start going up by 10 ohms at a time.

>>1221671
>all kinds of kludge
You've a flair for understatement, anon. The driver circuitry was a corner of one of my breadboards, made with a 74HCT245 and a bunch of transistor-based level shifters.
At a guess, there might be a minimum current enforced by the wiper resistance of the trimmer.

>>1221672
>alien style
Kek! Cyrillic alphanumeric VFD displays. I'm tempted, just to mess with the burger political footballers trying to gin up a new cold war.
In seriousness, I'd probably go with serial LCD panels because graphs.

Bump limit approacheth... any opinions on whether I do another lolpic or put up a nice comfy workbench photo I grabbed off the net?

>>1218883
EasyEDA. i 3 sets of 5 boards delirvered to my house in 3 days. it was like $27.

>>1221691
>easyeda
>Big Price Drop(75%OFF)! $2 for 10pcs, 2-layer, PCB sizes <= 10cmx10cm. (Each order only includes one chance of $2 for 10pcs PCBs)
Now I gotta figure out something I need ten of.

>>1221682
>VFD displays
Actually they're electroluminescent, and so they need 200VAC. Here's a not-English website full of them:
http://www.155la3.ru/electroluminescent.htm

>serial LCD panels because graphs
You mean dot-matrix, right? I'm not really sure what I would want to display other than the numbers, decimal points, and "V, mV, A, mA"; anything involving dynamic memory for a graph over time would be way out of my scope. With that in mind, maybe a 14-16 segment display would suit me. What kind of aesthetics can you get out of a dot matrix anyways? With only gimmicky tickers/scrollers and RGB, I'm not yet sold. But the EL display is a little bit much for a shitty power supply, even more-so when it comes to nixies.

>>1221686
Use pic related.

NEW THREAD

>>1221713
>>1221713
>>1221713

NEW THREAD

>>1221716
MOLDY BREAD, DON'T GO HERE

ACTUAL NEW BREAD

>>1221721
>>1221721
>>1221721

ACTUAL NEW BREAD

>>1221723
Meat, no gluten.

>>1221559
why the heck do you have positive feedback going to the op amp

Anyone know a sight to help me identify old componets whose datasheets I can't find on google? Like the one in the pic

>>1223139

yeah, it's called ''googles''

https://commons.wikimedia.org/wiki/File:KL_AMI_S6800_Black_Background.jpg

https://en.wikipedia.org/wiki/Motorola_6800