Looking more like encoder is indeed the issue following a few extra tests that I wanted to do to eliminate an IGBT being broken which is something I have previously experienced.
- Manual mode with encoder disconnected - smooth start and rotations of motor achieved.
- Manual mode with encoder connected - not so smooth but was able to spin wheels (not sure if SW was using signal in manual mode?)
- Auto mode and all connected - not easy to get wheels to spin at all with everything connected
Im interested to know what has changed in the software which has made a difference - removal of encflt parameter that helped deal with the noise?
I'm adapting a cheap 100ppr encoder from ebay. We'll see how that goes. Something like this :
Look forward to see how you get on Damien, good luck!
Took the Avago encoder off last night and could see it was indeed 256 pulses on the disc, all evenly spaced.
Perhaps 256 pulses and the slop in the drivetrain (im using original single reduction gearbox off the Edison van) will causes more erratic pulse signals? So i have just ordered one of these to try: https://www.ebay.co.uk/itm/AVAGO-Optisc ... 1438.l2649
Luckily postage was cheaper than advertised
Looks like I will have to adapt outer shell of old encoder to this one to mount it, but i hope 50 ppr will at least help highlight something? Enova clearly used two channels, which may have helped get it working for them?
Was interesting to see 4 unconnected wires into motor casing - they measured 320 kOhm accross them, so not sure what they are for?
I now have my optical 50 ppr encoder and i didnt get much of a change with single encoder mode, but the reactions of the set up seem less violent. I get loads of strange noises (like the old internet dial up noise i think) from the inverter when it tries to run the motor which still barely turns, unless i go into manual mode, so im now less convinced it is the encoder itself, but i can get some nice sounding ev inverter whine (like i used to get 3 years ago with v2.97? SW?) but somewhat pulsed wheels turning with AB encoder mode, so i dont know why this is so different to single mode, apart from the obvious pulsing probably because theres only one signal input?
Im going to focus on the switching of T1 next as I have only seen it switch cleanly turning the wheels by hand, so when powered I hope to see it switch ok. All good learning
Ripped out the original encoder today on the rear drive motor on the range rover and replaces with this thing from ebay. Very big improvement. Not perfect, but much better. Never thought I'd be thanking the clown who cut out the floor...
I don't understand what is going on, but I can see what seems to be a poor signal from the encoder? I did get smooth running in both manual unlimited frequency it seems, but only up to 1 or 2 Hz run mode via pedal for some reason, but then i keyed off and on and i couldnt even get manual mode to work properly, so possible something else is going on, or i changed something in my haste to try different parameter values?
I used my scope channel 1 - top line encoder signal to T1 base.
Channel 2 lower line T1 emitter signal.
Manually turning wheels in neutral:
Manual mode 0 to 20Hz:
Attempt playing with udcnom values 0 , 50V (actual bus voltage), 300V (motor plate) until it trips with 300V set:
Another attempt with different parameters set: (Channel 1 disconnected)
Im thinking about changing R3 & R4 to 500 Ohm resistors, but I doubt very much that's any cause for the problem.
Ripped out the original encoder today on the rear drive motor on the range rover and replaces with this thing from ebay. Very big improvement. Not perfect, but much better. Never thought I'd be thanking the clown who cut out the floor...
Thats great news Damien, think i will be looking for something similar.
Instead of the 5V from sensor board that i was using, I found that a separate 5V power supply has made a huge difference, and now i get the inverter to respond smoothly and the encoder signal is clear. I used a bench power supply first, and i have one of these https://www.ebay.co.uk/itm/12V-to-5V-Du ... 1976793008 and it works well too, so i will probably use it.
Now i need to get the wheels spinning faster than the 60 rpm or so I can get at the wheels with 50V test pack. Played with the parameters for a while but no great rpm improvement, so any ideas for parameters would be appreciated. Currently i have: boost 6000, fweak 121, fslipmin 1, fslipmax 2.7, ampmin 4, udcnom = 50, fmax 150, deadtime 63, idcmax & iacmax = 300. I think I have pot parameters wrong - dual channel, but both follow 0 -100 % throttle with values like 500 - 2500 min to max, so I guess pot 2 needs to be opposite max 500, min 2500? Cheers
thanks arber. Unfortunately no luck with that, but it spins so I think boost is ok? I tried 18000 and it didn't go any quicker. Ive attached the parameters that I last tried, would greatly appreciate if those with more experience can have a look and spot any major reason why I cant get much more than about 60 rpm at the wheels. The last time I spun the motor quickly was 3 years ago on something like version 3 SW. Could it be that there is no load sensed as I have the wheels in the air? Im also using an old web interface.
cheers
Thanks Tom
Ive not measured rpm, but I rekon its 60 rpm at the wheels and yes it seems to be the limit, which is going to be around 500 rpm for the motor.
No noise that I have noticed. The inverter seems to be very smooth, and I hardly get any whine now either, not sure if that's good or not?
I have tried fmin around 0.06, but that's a good shout as I have had it up just less than fslipmin for my last efforts.
Just tried fmin 0.05 and as much as I could think of changing that may influence the low rpm at full throttle.
I have no idea what is going on, and I have had parameters set the same as 3 years ago when it all ran rather well Theres only the odd parameter not showing now like encflt, and a load of new ones with the latest SW. Tempted to put old SW back on, but im reluctant given the improvements which protect the hardware better.
OK first you want to change "deadtime": "63", to something higher. I advise this to all. I didnt see which inverter you use and which drivers, but i would advise you set this at least 85 or even 180.
Is your motor really 4 pole? Is it possible it would be 6 pole?
If not that, can you use another inverter and/or different ACIM motor with this configuration?
Also i see you are using a Rev1 board. So you have single channel encoder interface. There is no AB signals in Rev1 board. Just channel A through transistor. The other pin provides 5V to encoder. Maybe you want to change thransistor type?
Can you rotate motor in manual mode? Remember this is V/Hz. So you provide 80% ampnom and 1Hz fslipspnt. When it starts to spin you add more fslipspnt. 3Hz, then 5Hz then 10Hz and 20Hz and 30Hz.... 50Hz, 70Hz...100Hz etc... Main point of this excercise is you will run out of voltage at some point ant you then read the speed from interface.
Now run it in encoder mode and read speed as the motor rotates.
You may want to use more voltage...
Have you checked the encoder signal with a scope as it approaches the problem speed?
I found the ‘on’ time was getting smaller and smaller compared to the ‘off’ time as the speed increased. The on and off pulses are not equal. Not sure if that makes sense but here’s a photo:
The width of the ‘off’ pulse is correct, the ‘on’ pulse becomes almost non existent at the speed that I get stuck at. It seems that this is what is causing my issue but I’m not sure of the root cause. Wondering if the encoder is not up to the job but if you have the same problem that would be very interesting.
As you can see I also have a lot of noise. This is on the 5V supply and the 12V whenever the forward or reverse is selected. Doesn’t seem good but I’m not sure if it’s causing the problem. The noise is present at all speeds so seems to cope with it ok. I did try adding a low pass filter on the 5V supply but that didn’t fix the problem. I didn’t have a scope at that point to check if the noise had been reduced though.
I also tried a separate 5V supply as you did but it made no difference for me. In fact it made it worse - I got clunking from the motor at lower speeds that work fine with the 5V supply from the board.
Thanks arber333 & Tom. I will try manual mode again and measure the encoder signal at the transistor again, as it does make some sense a limited signal could be causing the problem without looking into the SW code. Last time I tried manual mode I got to 20Hz at the motor easily, but this doesn't need encoder signal if I recall right. Intresting what you say about deadtime. I did have it 193 for sometime as that's what I measured with my scope. It seems to make little difference, so yes I agree stick with the longer time to avoid any shoot through.
I have the 50 ppr HEDM encoder fitted, and I read it needed a quality power supply - < 100mV ripple I think it was, also it has limited signal current output - 5mA, so im thinking I check the resistance to the base of the transistor, maybe it is too low? I do have different resistor values on my early board as I notice these changed at some point. Perhaps remove R1? Hope to get some time today to take a look, I let you know how I get on.
Just had another go and found that the encoder channel connected to the inverter was not working, it does appear to be broken. So I took channel B which I showed on my last encoder signal video (used for my speedometer) which still worked with 5V signal output and connected it to the inverter instead of channel A. As I have seen before there was little voltage on the signal even before switching the inverter on, so I disconnected R1, took a 1k Ohm resistor and connected inline to the inverter encoder input which appeared to limit current and show slightly higher voltage signal, but still lower than what Toms picture shows on his inverter.
I can see the signal is poor and I suspect switches the transistor on all the time or at least creates an odd signal when the inverter expects 50 ppr in my case. I guess the use of the encflt parameter in the past meant I could previously run the car? I don't know where the noise is from, since it is nice and clean until it is connected to the BC547 transistor base. As arber333 suggested, could be a different transistor could help, but that's something for now I wish to avoid. Just need to get the car running for now. I will try the Omron encoder I have ordered before I look at that. I will take a look at the Enova board to see how they treat the encoder signal, see what I can learn.
Manual mode runs sweet, so im sure its the encoder signal causing the low rpm issue.
Heres the encoder signal, noise much like in Toms picture.
I believe the Enova combI board has a very similar encoder circuit to the V1 board. This is the original schematic of the Enova board:
I have now removed all of this and all I have is a low pass filter (10k resistor, 1nf capacitor) between the encoder input and the Olimex. This made no difference at all for me.
It does seem very strange that we are both encountering similar problems with different encoders and different inverter boards. The only thing the same is the motor but I don’t understand how the motor could be causing this.
Another thought, do you know exactly what motor frequency/rpm you are limited to?
Just wondering on the off chance that it’s happening at the same encoder signal frequency as mine. My motor gets to roughly 35Hz so approx 1000rpm (4 pole motor) and I have a 36 ppr encoder so the encoder signal is around 600Hz.
Hm... with 3V3 you could be using 470R for R3 for sharper curve. Maybe that will make signal clearer.
I am still not convinced; you might have other issues with your control mechanics. Try running faster than 60Hz in sensorless. You might have to raise the voltage for the motor run faster.
Did you consider that using a push pull signal may invert transistor signal and what inverter sees is decreasing input which comes to an equilibrium at some PWM value?
Maybe you would rather bridge transistor contacts directly and use a voltage divider to lower signal to 3V3 level. So remove R11 and R3 and put a voltage divider into T1 pads. You may also remove R4 or use its pads to make divider.
I would also replace C1 with 1nf.
Johannes used T1 so encoder voltage wouldnt be an issue. Now i just use 470 pullups (5V) on encoder connector with open collector encoders. I use 10K line resistors directly into Olimex pins and i guess they are 5V tolerant at low enough current.
I now have the range rover driving with the new encoder but it looks like the enova motor suffers the same "defect" as the ansaldo I struggled with some years back.
Jack Bauer wrote: ↑Mon Jun 15, 2020 7:08 am
I now have the range rover driving with the new encoder but it looks like the enova motor suffers the same "defect" as the ansaldo I struggled with some years back.
Are you talking about the over current problem that you had with the ansaldo? Or was there another issue with the ansaldo that I’ve missed?
Thanks for all the replies, I will have to study these since im a beginner when it comes to electronics Tom3141 - I will try and get exact motor max rpm tomorrow with the 50 ppr encoder, perhaps you can send me your parameters and I can set exactly the same? I guess that my V1 board configuration may make some differences to the max rpm I see, which I rekon is 500 rpm of the motor, and you get 1000 rpm.
Since I have a spare motor and inverter, I took a look inside. No different to what I recall seeing inside 3 phase induction motors, but it was a long time ago and I am no motor expert. Looks like quality kit to me. Not got the rotor or phase connection end out yet tho and i wonder whether the motor is connected in star or delta? Anything else I should look at inside? - edit: looks like a Delta connection as the only terminals are the 3 that connect to the inverter directly. Makes sense to me, I guess that's the same for all 3 phase EV motors?
So far im now sure it is a 4 pole, 2 pole pair motor.
Pole pairs = 60.freq/Ns = 60.121 / 3600 or 60.305/9000. Both = 2 pole pairs.
Motor has 36 slots. The number of slots per pole & phase = 36 slots / 2 x 2 pole pairs x 3 phases = 3 (which you can see from the top)
I plan to look at encoder input onto original Enova inverter control board. I know they used both A&B channels, but how exactly i dont know.
I've attached the parameters I've been using. When I was testing at 30V I used basically the same parameters.
I've ordered an Omron encoder to try on mine as I've run out of other options to try. The arduino style optical encoder that I'm currently using seem to fix all the problems apart from the limit at 1000rpm that only appeared after I switched to this encoder. I'm thinking the arudino encoder might either be faulty or just not up to the job, it was only £2.50 on ebay!
The Omron encoder that I bought seems to have fixed the issue. It now spins up past 1000rpm and didn't have any overcurrent/encoder errors although I can't test it with the wheels on the floor yet as it isn't mounted properly. I'm going to try to get some parts made to mount it next week.
The only issue I can see with this encoder is that it's only rated up to 5000rpm. This equates to around 63mph in my car but I'm going to stick with it for now if it works ok once I test it with the wheels on the ground.
I'm powering it directly from my 12V system (via a low pass filter). I've fitted a pull-up resistor to 3.3V on the output and I'm using a 10K 1nf low pass filter on the inverter board.
