Further improvements to electrical stability from lessons learned in my more powerful designs
1A 5V auxiliary output
The intent is to refresh the CFOC design so that it can actually be made by DIYers again in this chip shortage. I also intend to actually make these things with full heatsink/enclosure and sell them at the most affordable price that I possibly can. Thereās not a hard ETA right now.
UPDATE!
A small batch has been made and is available through the below vendors.
Only got as far as performing basic power supply functions check, programming it, and seeing if it boots up. So far so good but got a long way to go to see how it does. A lot of new things to test on this baby.
One thing Iām going to change is adding more caps. Shorter ones but slightly fatter and more of them like in the pic below.
Iām hoping DRV removal will improve reliability as well. I was mostly satisfied with V2 apart from the oddball generic drv fault that popped up when doing hard launching or trying to throttle against the hill. Though it may seem to be rooted somewhere in FOC sensorless mode as the V2ās ran a lot better once I connected the sensors. Canāt wait for this to be out to do a long needed upgrade. @shaman Are there any estimates on how much this unit will cost assembly wise?
It was $40ish per unit when doing 10 units from JLCPCB. Plus the cost of the caps, FETs, and cables. I maxed out JLCPCBās service which included having them do the connectors. One could save a little bit more by doing all the though-hole stuff yourself.
after the Raiden Zescs died out I was waiting for your DRV-less design to come to action, just hoping for those screw terminals for power and motor phase connectors, I just love it, itās so easy to fix cable lengths and switch vescs between builds when you have those setup
Can you eli5 how a drvāless design is different from the usual one, no specific really just more of a curiosity on my behalf, ie what is used instead
The DRV chips we use are MOSFET gate drivers with some extras, their job is to take digital signaling and use it to switch on and off the gates of a MOSFET quickly and efficiently. Basically fet gates need fairly small amounts of energy to switch on or off, but pumped into them quickly to minimise the time spent transitioning between off, half on, and on.
A normal logic out from a microcontroller canāt do that so you either use an integrated package like the DRVs or build a set of drivers yourself. You need 6 drivers per motor (3 phases and a circuit for high and low on each. I think this requires 6 but I might be missing a way to do it with 3) that each take up a bit of board space and add complexity to the layout. DRVs also have extra features like handling low power internal supplies like the 5v used by the processor, and they have current measurement circuits to accurately gauge motor current
(Apologies if you know this already and were looking for a more detailed explanation of how to implement a gate driver or something)
Alright so I started taking a look at the gate drive waveforms and current sense quality. The gate driver Iām using is just barely strong enough to turn off the MOSFETs in a reasonable time. Not ideal but itās something I can work with by adding some āhelpā to boost the drive strength. Iād like to use this gate driver since itās well stocked right now.
The current sense signals are a bit noisy and I think I found why. Has to do with the switching frequency of the primary buck converter. I can adjust the switching frequency of the buck converter to be higher and also add some key capacitors in places to help squash the noise.
So far thereās nothing detrimental. Stuff I can deal with one way or another.
