Paul and Sabrina's Cheap 3 Phase Inverter (AC Controller) with Field Oriented Control

[MPaulHolmes]

The wires are:
8 wires for resolver. It goes right to a plug on the motor.
+12v and gnd
CAN

[thingstodo]

Originally Posted by MPaulHolmes

The wires are:
8 wires for resolver. It goes right to a plug on the motor.
+12v and gnd
CAN

8 wires for resolver
2 for CAN
2 for +12V and Gnd
3 spare wires

If they all go to the motor .. the motor controller and the motor both need liquid cooling so they are likely to be close together? If a 3 foot connector from the controller to the motor would keep all the signals together, clean and dry ... maybe go for the connector.

[JANI-DIP]

Hi, is anyone have P&S resolver to encoder board pcb, schematic and BOM (final version) for leaf motor?

[laapmetot]

I'm not sure its the latest version but I bought this board from Paul Holmes and it worked together with the leaf motor and his controller in 2018, and I got information from him. Too big to share here so drop me a pm if you want

[laapmetot]

I have the 200kW ac controller as designed by paul holmes which I bought a couple of years ago. With a lot of help from Paul I got it to work on a leaf motor, but unfortunately never got it to work with another motor so I shelved it since the leaf motor was too large to fit the go-kart I was building.

Now I have a new project and thought to try it again. I learned some more, and hopefully enough to get it sorted out this time.

Unfortunately I couldnt reach Paul anymore and was trying to find information online and found this forum with a long thread of this exact controller, so I will post here and hopefully this will help me, and others in the process as well.

[laapmetot]

What I try to do is to get the motenergy ME1616 PMAC or PMSM permanent magnet synchronous motor to work with the P&S controller.

The controller itself is functional: I have a leaf motor which runs fine, both with the standard resolver with P&S interface board to convert resolver to encoder, and with custom code from paul to run a sincos sensor.

Current sensors have correct values, current settings and throttle settings are ok. So I know for sure the electronics are ok and basic setup too.

Since both the ME1616 and the leaf motor are PM motors, I didnt expect much trouble. famous last words :-)

First thing i did:
try motor type 4 which is generic pm motor.
I have a US digital encoder mounted on the ME1616: 2000 cpr, index pulse and open ended per Paul's instructions. I set the encoder-ticks to 2000 and set the pole pairs to 5 per motor datasheet.
Unfortunately the motor doesnt turn, its just sits there vibrating a bit and making "inverter noise". what I tried:
- stream-poscnt: counts up to 8000 and drops to zero and raises up again etc. So that seems good (2000cpr x 4 for quadrature sounds logical) When turning the motor the other side it runs down to zero, then jumps up to 65000 or something and goes down. Weird, not what I expected: I would expect the index pulse to result in reset to 8000.
- Tried 500cpr just to see if that helps: but no.
- swap a-b to check if encoder runs opposite direction then motor. No difference.
- tried step by step the whole angle-offset range, also for swapped ab. There is a range in which the motor " wiggles" a bit more but no clear correct value
- when switching on the controller without encoder connected, it turns slowly and rough: must be the startup sequence which runs until first index pulse.

So, no good result.

Then I tried the sincos version with custom software I got. I can stream sine and cosine values which make sense: go up and down in a range roughly between 400-800. Center value in the software is 612 so no need to correct that. Basically the same result as before: motor wiggles but that's it. I tried the same as above to no avail.

I tried checking the code: I can change small stuff and recompile, but digging in the heart of the functionality is beyond my understanding of C and motor theory. I tried reading up on both topics but you cant become a software and power electronics professional in a couple of weeks unfortunately.

Maybe somebody can help with this? There are a couple of these controllers out there as far as I know.

https://www.youtube.com/watch?v=w-5cUruwHgI

https://www.youtube.com/watch?v=hJSRw9ITxFk

[thingstodo]

Originally Posted by laapmetot

First thing i did:
try motor type 4 which is generic pm motor.
I have a US digital encoder mounted on the ME1616: 2000 cpr, index pulse and open ended per Paul's instructions. I set the encoder-ticks to 2000 and set the pole pairs to 5 per motor datasheet.
Unfortunately the motor doesnt turn, its just sits there vibrating a bit and making "inverter noise". what I tried:
- stream-poscnt: counts up to 8000 and drops to zero and raises up again etc. So that seems good (2000cpr x 4 for quadrature sounds logical) When turning the motor the other side it runs down to zero, then jumps up to 65000 or something and goes down. Weird, not what I expected: I would expect the index pulse to result in reset to 8000.
- Tried 500cpr just to see if that helps: but no.
- swap a-b to check if encoder runs opposite direction then motor. No difference.
- tried step by step the whole angle-offset range, also for swapped ab. There is a range in which the motor " wiggles" a bit more but no clear correct value
- when switching on the controller without encoder connected, it turns slowly and rough: must be the startup sequence which runs until first index pulse.

Agreed that for 2000 cpr, and a 2 channel encoder, you get 2000 A channel positive changes, 2000 A channel negative changes, 2000 B channel positive changes, 2000 B channel negative changes .. per rotation. So a count to 8000 makes sense, which then resets to 0 and starts counting up again.

Going the other way goes to a negative number. Since Paul prints them out as unsigned integers (if I remember) they will show 65535 instead of -1, 65534 instead of -2, etc. It's just how the bits in 2's complement binary notation show up. This is normal.

I don't know about the angle offset. That's not used for AC motors (which is what I have experience with).

I will dig up the values that I used for my son's Siemens AC motor. That was a couple of laptops ago, so it will take a bit of looking. Although all of that was was likely posted on this site somewhere ... it's likely faster for me to search my archives than to search through my postings on the various threads

The constants used by Paul's equations to determine how fast to rotate the magnetic field, and how much current is required to move the motor smoothly don't make a lot of sense to me. I don't pretend to understand them.
But I have some numbers that worked for a larger motor (no permanent magnets, though). Maybe you can then guess at some numbers that you can try and that may make your motor operate more smoothly.

Paul had some testing code to try a bunch of different constants and pick the 'best' one. That did not work so well. I think it got removed somewhere along the line.

[laapmetot]

thanks for the feedback! Actually I'm beginning to think Paul didnt finish the PM-encoder part of the controller. The AC part works fine, the leaf-resolver part too but the PM section just refuses to work and I'm not able to fix it unfortunately.

[mpgmike]

Originally Posted by CollinK

So... I'd like input on this:

I'm thinking of switching to dsPIC33 chips before I finalize a PCB design. This chip is newer than the dsPIC30 chips and programming wise uses the same instructions and everything. dsPIC30 chips are not recommended for new designs anymore. However, there are some points to consider:

1. Every dsPIC33 chip is 3.3v. There are no exceptions. So this would require a 3.3v regulator on top of the 5v regulator already in use. And I can't just get rid of the 5v regulator because other things need it. This is easy, though, I can just use a 3.3v regulator off of the 5V source.
2. dsPIC33 chips can only sink or source very low currents (4ma) as opposed to the 25ma possible with dsPIC30 chips. This means one would need a bunch of transistors. Granted, I probably should plan on that anyway.
3. dsPIC33 chips do not come in DIP40 packages (which is what the current layouts all use). You must do 28 pin if you hope for it to be through hole. Otherwise it jumps up to 44 pin QFN.
4. dsPIC33 chips have no EEPROM
5. They do, however, add other nice stuff like ECAN, programmable PLL, more interrupts, more timers, 12 bit ADC, DMA.

I'm really strongly looking at using 44 pin QFN for the dsPIC chip. It could be the only surface mount chip required. I don't think it will be too terrible to solder and boards could be sold with the dsPIC already soldered on for those who don't feel adventurous. I could use dip28 chips but I would almost certainly end up needing two of them.

I'm on page 63 of this 3xx page thread, but had to interject before I lost my place. The dsPIC33EV series is 5 volt! The dsPIC33EVxxGMxxx chips are motor controller versions. I concede that they don't have EEPROM, but they do offer Flash programming (which can be manipulated to act like EEPROM). Now, back to my page 63.

[mpgmike]

OK, read through the DC and AC Controller threads. Quite an education. Thanks to the blood & sweat from the folks that worked through the bugs to realize working systems.