/diy/ - Do It Yourself

>>1688864
>Not with a "dumb" bang-bang, it won't. there's a delay required after commutating to allow for Zero-crossing to occur "cleanly", which depends on the feedback itself to work properly.
But what about delay before zero-crossing? It is obvious from that graphic.
Maybe it is better to detect fall and rise too? I have seen videos of people probing BLDC motor controllers on e-bike, and at full throttle, "ass-shaped" waveform appears, which might confuse rise/fall detection...
>Also, Back EMF coming from a BLDC is notoriously dirty, so signal conditioning and some logic is required.
This sucks, since I don't have a scope, thus I can't make a filter. This would be a problem with PWM current limiting for sure.
But on the other side, I think those motors don't get more than 200-ish Hz, so low-pass should do.
>you then update the frequency of drive to reestablish syncronous operation.
This is the idea of motor control.
You can make them spin in really dumb way, just by going though that 6 states with somewhat reasonable low frequency, but this is not useful at all, you can't ever run fan from such motor.
>Lol no they can't. duty cycle control of a brushed motor or open loop stepper control is one thing. Ad Hoc Arduino for BLDC with feedback is non-trivial.
Proper motor control without hall sensors is hard to make on Arduino. It is too slow to reliably detect zero-crossings.
For that you need to go really low-level and use AVR-C or even asm without any gimmicks. It is possible, because commercial ESCs on Atmega168 and 328 exist.
>You either want to solve a discrete logic problem,
Probably this, much faster than compiling shit, and finding lost ";".
>>1688829
Die young. Fuck living up to 80 years, getting Alzheimer, shitting pants and such. Better die at 40 from lead poisoning combined with cancer from flux fumes.
>>1688888
Checked. Boiling mercury will destroy soldering tip. Don't do this. They expensive