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

[MPaulHolmes]

My guess is that there was some sort of overcurrent event, and it hasn't been cleared.

Also, I would start at 1 with Ki, and gradually increase. In practice, I've noticed that it is typically 1/1000 or 1/100 of Kp.

If you tried Ki = 128 each time you unplugged the control board's 24v power, it could have tripped the hardware overcurrent. Then trying the other values wouldn't do anything, since I never clear the fault. It requires a cycle of power.

So, try this: turn off power to the control board, start with Kp = 941, Ki = 0, 1, 2, ... go from there. If Ki = 1 is the best, then that means I probably do need to increase the resolution of Kp and Ki after all. There is some room to do it though.

The default Kp and Ki when cycling the controller now is Kp = 1, Ki = 0, so it should show just 512 512 500 512 ... or something like that if you just cycle power to the control board, and then do "run-pi-test". "run-pi-test" outputs the error, and 512 corresponds to 75amp. So, if it keeps saying 512, the difference between 75amp (the desired feedback) and the actual feedback is 512, which means the actual feedback is 0amps. kp = 1 and ki = 0 produces a feedback of 0amp, since kp is just too small.

[thingstodo]

Originally Posted by MPaulHolmes

So, try this: turn off power to the control board, start with Kp = 941, Ki = 0, 1, 2, ... go from there. If Ki = 1 is the best, then that means I probably do need to increase the resolution of Kp and Ki after all. There is some room to do it though.

Take a look at 183 and 184. It looks to me like any non-zero ki-id value and your hardware overcurrent trips. So far I've tried 128 and 32 in my previous trials, 1 and 2 in this set of trials. All of them appear to turn off the test output.

183 has a successful test at kp-id 941 and ki-id 0 followed by changing ki-id to 1 and the output turning off

184 has the same as 183, but I changed ki-id to 2 instead. When the test failed to output, I changed ki-id back to 0 and run-pi-test did not output to the motor.

The 12V battery is charged back up to 13.04V. I added a switch to turn it on, beefed up some of the power cables so that they can take some decent current, and made a platform over the motors (ac and dc) so that I can mount the controllers there. I also need to support the encoder on the end of the DC motor .. when we get that far ..

While I'm waiting ... I'm going to try connecting my P&S 500 controller to my Warp 9 and do a .. I guess it's third spin for the controller, second spin for the motor, and first spin for the coupled motor. Results will be posted in the DC controller forum.

[MPaulHolmes]

I will change the code to allow greater resolution for kp and ki tonight. And send it tonight. At walmart at the moment. Just went swimming

[thingstodo]

Originally Posted by MPaulHolmes

I will change the code to allow greater resolution for kp and ki tonight. And send it tonight. At walmart at the moment. Just went swimming

Paul sent me the new code last night but I had already gone to bed.

Paul tells me that I am a *BIT* aggressive on tuning the Kp. I'll go back through the logs and start again at a location where there was no ringing, even if the voltage does not drop below 0, then continue tuning Kp with the new code. Paul estimates it will end up around 900 * 8 = 7200.

Then he has asked that I start with Ki at 1, 2, 4, etc until I get the best damped response that I can.

Kp and Ki are now multiplied by 8 so we have almost one more digit resolution.

I should have new data posted tonight.

[MPaulHolmes]

This dude is a machine of precision.

[thingstodo]

I loaded the hex file into the AC controller without incident.

Then things got a bit more ... interesting ...

After loading the hex file, removing the programmer, and cycling power - all is well!

I have some bad news. I turned off the 12V (so the contactors dropped out), turned on the breaker on the 125VDC pack .. and there was a BANG and some of the smoke came out.

I have a theory:
- my main contactor (Gigavac contactor) welded itself shut, so when I turned on the 125VDC, it shorted across the capacitors and something weakened. The first MOSfet on the DC controller failed (black mark) and the DC motor rotated slightly, after the big BANG and the black smoke and the ozone smell

I turned off the pack, checked voltage on the DC bus (the breaker tripped, but not the fuse) and it was 0. Not close to 0, but 0.00V. Then I checked continuity across the main contactor - 0.1 ohms, same as when I connect the probes directly together. I removed the DC controller from the DC bus on the AC controller. And I rechecked. The resistance between the 2 sides of the main contactor are still 0.1 ohms.

So I took some pictures and I'm posting them ... I'll get back to tuning the AC controller tomorrow.

[MPaulHolmes]

Is there a precharge resistor for the dc controller? Also is there one for the ac controller? So the caps fill slowly before the contactor closes?

[thingstodo]

Originally Posted by MPaulHolmes

Is there a precharge resistor for the dc controller? Also is there one for the ac controller? So the caps fill slowly before the contactor closes?

No. Yes. I cheated a bit - I only have so much space above the motors to mount parts. I'm using a single precharge, with contactor, for the AC controller and a single main contactor. I connected the DC controller DC bus to the AC controller DC bus, after the precharge and contactors. The AC controller does the 5 second precharge, then closes the main contactor and opens the precharge contactor. The contactors still both click for open and closed ... but there is 0.1 ohms across the terminals for the main contactor when it is 'off'.

So the signals from the DC controller for pre-charge contactor and main contactor are not used.

The precharge is a 10 ohm, 25 watt salvaged from a 600V controller.

The contactors are Gigavacs.

More investigation tonight.

[MPaulHolmes]

So, the main contactor is basically welded closed? When did that happen? It's possible that the resistance could be almost 0Ohm. My fluke multimeter sometimes says 0.1Ohm when it's really almost zero ohm. Fran told me once that the mosfets can fail shorted if the capacitor bank fills up too fast. I don't know the mechanism for why, but evidently he's seen that. So, do you have something like this:

+125v Battery Pack --------- 10 Ohm ---------- switch ------------ AC and DC cap +

+125v battery pack -------- 0.1Ohm (contactor) --------------- AC and DC cap +

+125v pack ground --------- AC and DC controllers ground

That would mean the precharge part isn't really doing anything, since all the current would be hogged by the 0.1Ohm part of the circuit. That could explain why the failure happened after closing the DC breaker, but before closing the main contactor. I wouldn't do anymore tests with the AC unless you remove the contactor from the circuit, and maybe fill up the cap manually by some other means. Maybe something like this:
permanently wire 10Ohm resistor from +125v pack to the AC cap + (well, permanently just for testing). No breaker in the way:

+125v battery pack ------- 10Ohm resistor ----------- AC cap +

Then, for the high current path:
+125v battery pack ------- dc breaker ---------------- AC cap +

+125v ground -------------- AC Cap minus.

Then, because the caps are always full, you can open and close the breaker any time you want, and don't have to worry. Careful about running the code with the breaker open, though, since I caught a precharge resistor on fire by doing that. haha

[thingstodo]

Originally Posted by MPaulHolmes

So, the main contactor is basically welded closed? When did that happen?

Not sure, but likely right after the firmware was flashed on the AC controller. I'll get the exact sequence written up and posted when I get home and see my notes.

It's possible that the resistance could be almost 0Ohm. My fluke multimeter sometimes says 0.1Ohm when it's really almost zero ohm. Fran told me once that the mosfets can fail shorted if the capacitor bank fills up too fast. I don't know the mechanism for why, but evidently he's seen that. So, do you have something like this:

+125v Battery Pack --------- 10 Ohm ---------- switch ------------ AC and DC cap +

+125v battery pack -------- 0.1Ohm (contactor) --------------- AC and DC cap +

+125v pack ground --------- AC and DC controllers ground

Yes. The gigavacs don't have a set of 12V contacts that show when the large contacts are closed. Motor starters in the 3 phase world normally do. I've only had a dozen failures on AC motor starters across 700+ motors in our mill during the last 12 years, but the contactors do weld themselves shut on a rare occasion.

That would mean the precharge part isn't really doing anything, since all the current would be hogged by the 0.1Ohm part of the circuit. That could explain why the failure happened after closing the DC breaker, but before closing the main contactor.

That's my guess so far. Further investigation tonight.

I wouldn't do anymore tests with the AC unless you remove the contactor from the circuit, and maybe fill up the cap manually by some other means. Maybe something like this:
permanently wire 10Ohm resistor from +125v pack to the AC cap + (well, permanently just for testing). No breaker in the way:

+125v battery pack ------- 10Ohm resistor ----------- AC cap +

Then, for the high current path:
+125v battery pack ------- dc breaker ---------------- AC cap +

+125v ground -------------- AC Cap minus.

Then, because the caps are always full, you can open and close the breaker any time you want, and don't have to worry. Careful about running the code with the breaker open, though, since I caught a precharge resistor on fire by doing that. haha

That will work. I will likely replace the gigavac with the one from the precharge circuit and leave the resistor across the gigavac as you show. The 80 amp DC breaker I have on the pack (the one that tripped during the failure) is also my pack switch.