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

[thingstodo]

Update for Sep 11

The latest hex file I received from Paul is SensorlessFOC.hex (but he's still working on the sensorless part) dated Sep 11 at 6:22 pm

Loaded into the controller - OK

config shows 0 for raw-throttle-signal.

Start capture to foc test 1. run-pi-test - no sounds and eventually prints every number from 62 up to 20000. Wiggle around the connection on the throttle. It seems the cheap potentiometer is not up to the task! When it begins reading, center it.

run-pi-test again. Again, all number printed out.

cycle power, config shows 522 as a raw throttle now. Try run-pi-test again. It gives me no numbers and a message that states
'No values passed the test'. Hmm

run-pi-test again, just for the fun of it.
This time I get 3 numbers. 3534, 4030 and 5456. 5456 sounds familiar, but it different from my last successful run. But it's close.

Set p 5456 and i 88
run-rotor-test. It hums for a while but does not turn. It seems to abort the test early. All 0's printed out.

Cycle power to the controller
set p 5456, i 88, encoder-ticks 4000
Try running the motor with the throttle. It appears that the encoder signal is not getting through. I need to review my notes and see if I put 24V into the encoder or 12V into the encoder. I have a resistor network on the output to drop the signals to 5V. I remember that the signals never got above 10V ... but I don't remember the supply voltage I used.

I'll update this post when I find the notes .. or find my posting about the encoder ... if all else fails I still have a hex file that just reads the encoder signal and prints it to hyperterminal

It appears that I need a new set of commands from Paul so I can change the encoder-ticks, the number of pole-pairs, max-battery-amps, max-motor-amps, etc etc

I think that the max amps (motor and battery) may be limiting the test. I *KNOW* that the alignment of the coupler between AC and DC motors is not good (binding) and that the coupler between the DC motor and the encoder ... the encoder wiggles when it turns Maybe it's good that the motor did not spin up to 6000 rpm? I'm sure that the motors would have vibrated apart!

Video 1 https://youtu.be/beccIT8Gy2c

Video 2 https://youtu.be/QRyaNKy5Sss

Video 3 https://youtu.be/0dB2qA5oYBk

[MPaulHolmes]

I think anything in 3500 to 5500 is a good P value with your 120v pack. Some days they will pass the test, and some days they won't. It's just a question of if there was some weird sporadic noise on a particular test to determine if something passed. The test conditions are set very harsh, so I bet if I eased up just a bit, a whole bunch in that range would pass.

It should be set to:
encoder-ticks 1000

since that's just how many ticks are on the encoder. The software secretly makes the resolution 4 times that, but it's a secret.

for pole-pairs, you type the following:

pole-pairs 1

For max-battery-amps, max-motor-amps, the defaults should be fine (they are only really relevant for when using the throttle). default max-battery-amps is set to 400 I think. same for motor amps (line to line current).

I don't think run-rotor-test is going to have much meaning without the encoder attached. that's a guess though.

[MPaulHolmes]

I found a major bug that was probably causing all the problems. I'll send a new hex file in a few minutes.

[thingstodo]

Not sure what the defaults are, but these are the settings that come up when the controller powers up

AC controller firmware, ver. 1.0
config
p=14000 i=00225
current-sensor-amps-per-volt=0027
max-regen-position=0025
min-regen-position=0400
min-throttle-position=0624
max-throttle-position=0972
fault-throttle-position=0005
max-battery-amps=0050 amps
max-battery-amps-regen=0010 amps
max-motor-amps=015 amps
max-motor-amps-regen=015 amps
precharge-time=0005 tenths of a sec
rotor-time-constant-index=030
pole-pairs=002
max-rpm=06000 rev/min
throttle-type=0
encoder-ticks=0512 ticks/rev
stator-resistance=01024 mOhm
stator-inductance=00307 mHenry
pack-voltage=124 volts
pi-ratio=062
raw-throttle=0000

[MPaulHolmes]

It should work a lot better for you now. Also no more outputting every possible P and I all the way to 20,000. I've also attached what a typical rotor test might look like with an encoder. It runs through all possible rotor time constants from 0.005sec up to 0.055 seconds. Mine worked best today with 0.049 sec. Some days it was good with 0.031 sec. They say the time constant varies quite a bit since the rotor inductance changes based on temperature. I guess they weren't joking.

EDIT: NEVER MIND. my best was about 0.028 or 0.029, just like normal. haha. that means I need to fix the "automatic save of best rotor time constant" feature since it updated the rotor time constant with the last value tested.

Also, now if there's any fault, it will automaically be immediately displayed on the screen through the serial port. And it will say which fault it is.

[thingstodo]

Sep 13 Update

The latest hex file I received from Paul is SensorlessFOC.hex dated Sep 12 at 11:14 am

Loaded into the controller - OK

foc test 2 is the log file

set for pole-pairs 4

set rotor-time-constant=index 40 (17)

run the motor with the throttle signal

Motor is NOT 4 pole pairs


foc test 3 is the log file

set for pole-pairs 2

run-rotor-test

video shows a big bang - test aborted at 870, index 50



foc test 4 is the log file

set for pole-pairs 1

the last index is NOT the best, 83, but that's still low.

index 25

run motor with throttle .. lots of buzzing. Motor does NOT have 1 pole pair.



foc test 5 is the log file

set for pole-pairs 3

the last index is the best, 27, but that's still low.

index 50

run motor with throttle .. lots of buzzing. Motor does NOT have 3 pole pairs



foc test 6 is the log file

set for pole-pairs 2

Throttle fault on power-up

Check for a couple of items in config, wiggle potentiometer to get a reading. But it appears to need a power cycle to reset the fault.

pole pairs 2. run-rotor-test

Best value is last - 890.

Set rotor-time-constant-index to 49 (is it 0 based or 1 based?) Oh well, if I get the second last value it will be fine.

Crank the throttle - it appears I was too aggressive. Hardware fault when I try to slow down. Darn!


log file foc test 7

throttle is out of range on power up

cycle power

throttle is 257. Change to 482. Run-rotor-test

Powered down during test. It was cycling pretty badly. I should not have killed power - the test had completed (but I didn't realize that!) Oh well.




log file foc test 8

run-rotor-test finishes, but there is a hardware fault at the end of the test. So I could not select the highest value (1231) and test with the throttle.

It appears that there is a problem with my cheap potentiometer, used for the throttle. There are dead spots. Those spots give me 0 as a throttle value.

Sigh!

Log files attached. Working on the videos.

Video 1 https://youtu.be/isGhQsoZepA

Video 2 https://youtu.be/5bUCE9oz1FM

Video 3 https://youtu.be/llbOpFHC6cU

Video 4 https://youtu.be/jR54cTNkKu8

Video 5 https://youtu.be/bO53AHmxcrA

Video 6 https://youtu.be/xdrg1i52hs8

[MPaulHolmes]

Some of the dead spots could be commanding a valid huge throttle before the fault is triggered, since it averages 128 throttle reads before using it. The bang may have been commanding 400amp per phase, at zero rpm. A hall effect that looks just like a POT is around $25 from mouser i think. I got the 180 degree turn option. Its good to 10,000,000 cycles i think.

The rpm values you are seeing are actually rev per 16 sec. So it was actually 60/16 times those values in rpm. Your highest value was about 5000rpm. Dang.

I made the index start at 1 for counting rather than zero for the rotor test. So i think you would pick 41. Your motor is more sensitive to the right rotor time constant. I think I'll increase the resolution a bit so you have a wider range of values to choose from. It won't matter much, but this is a very valuable piece of info.

By the way, I have a new unloaded motor test you can try. It will tell you the stator inductance. That's one of the variables needed for the sensorless code. I'll send a hex file later for that. I'll disable throttle completely, and then you just run the code. It will automatically spin up the motor to like 5000RPM, and then will automatically compute the inductance and print it on the screen.

[thingstodo]

Originally Posted by MPaulHolmes

Some of the dead spots could be commanding a valid huge throttle before the fault is triggered, since it averages 128 throttle reads before using it. The bang may have been commanding 400amp per phase, at zero rpm.

I have video edited and uploaded. I've been adding youtube links to the status for the day. But I'll let you know about this one - a mounting bolt fell out of the ring cap and shorted across the DC bus. It was a *BIG BANG*.

A hall effect that looks just like a POT is around $25 from mouser i think. I got the 180 degree turn option. Its good to 10,000,000 cycles i think.

I'll look it up. It'll be worth it. The parts for the IGBT drivers are on back-order until .. mid november!

The rpm values you are seeing are actually rev per 16 sec. So it was actually 60/16 times those values in rpm. Your highest value was about 5000rpm. Dang.

Not sure my setup would take 9000 rpm. I need to do some work on the mechanical alignment.

I made the index start at 1 for counting rather than zero for the rotor test. So i think you would pick 41.

Good to know. My memory is not what it used to be!

Your motor is more sensitive to the right rotor time constant. I think I'll increase the resolution a bit so you have a wider range of values to choose from. It won't matter much, but this is a very valuable piece of info.

Glad that some of this testing is helping!

By the way, I have a new unloaded motor test you can try. It will tell you the stator inductance. That's one of the variables needed for the sensorless code. I'll send a hex file later for that. I'll disable throttle completely, and then you just run the code. It will automatically spin up the motor to like 5000RPM, and then will automatically compute the inductance and print it on the screen.

OK - I'll be working on the alignment late this afternoon. Maybe doing more testing this evening.

[thingstodo]

Posting video takes a lot of time.

The video is likely of SOME use to Paul, but it takes not further effort to make it available to all so I did that. My wife plans to view some of the videos when she has insomnia!

These posts have video links added afterward:

Sep 8
http://ecomodder.com/forum/showthrea...tml#post492687

Sep 11
http://ecomodder.com/forum/showthrea...tml#post493114

Sep 13
http://ecomodder.com/forum/showthrea...tml#post493274

The video is what I did and saw. The commentary on the bottom has the benefit of watching the whole video and thinking about other things that may have happened.

There are also corrections to the dialog in the subtitles - I did not go to the trouble of correcting the audio.

[MPaulHolmes]

This is awesome info. I just made a change where it will show every fault that happens rather than just one of them. Also, I can make the rotor test probably 100 times faster now too. Although we know what your rotor time constant is, so you probably never need to run it again. haha. Once you know it, it's fixed for the life of the motor.

In fact, anyone in the country that wants to know what their EVTV rotor time constant is, we can now tell them. It's a critical variable in the FOC code, and is NEVER on the name plate. And in fact even when I've called the motor manufacturers, they never even knew what a rotor time constant was, let alone had the value available.

So, just for posterity, that motor's is about 0.046 seconds.

How is your 24v supply hooked up?