John in CR
100 TW
Congrats on getting her rolling outside and so close to buttoning it up to have to address mechanical limits. Your persistence is an inspiration to us all.
Hi power inverter for Nissan leaf motor. Dyno's 302.3hp p15
- Thread starter Arlo1
- Start date
Ohbse
10 kW
What sort of speeds were you getting up to before braking? Camera perspective makes it pretty hard to gauge but looks pretty quick once you get into the throttle!
I don't suppose you have some external measurements for the leaf drivetrain do you? I'm having evil thoughts about converting a classic mini or maybe a smart roadster, but they both have very constrained engine bays. It looks like you've got heaps of space after the casing on the right hand side of the engine bay which gives me some hope it could fit into something super tiny.
I don't suppose you have some external measurements for the leaf drivetrain do you? I'm having evil thoughts about converting a classic mini or maybe a smart roadster, but they both have very constrained engine bays. It looks like you've got heaps of space after the casing on the right hand side of the engine bay which gives me some hope it could fit into something super tiny.
Trying to shorten this video but the youtube editor is painfully slow and my gopro app crashed.
So here is something fun to watch. This is at ~1/2 power.
The phase amps were set to 450 and I could not pin it, I was rolling on the throttle until ~90% throttle.
Also the variable regen is super cool It feels like it can slow the car like nothing when the mechanical brakes don't seem like they will be enough.
All the weight is on the front wheels there is nothing in the trunk and the gas tank is removed.
Skip ahead to 3:20
[youtube]5ylj-6ORmik[/youtube]
So here is something fun to watch. This is at ~1/2 power.
The phase amps were set to 450 and I could not pin it, I was rolling on the throttle until ~90% throttle.
Also the variable regen is super cool It feels like it can slow the car like nothing when the mechanical brakes don't seem like they will be enough.
All the weight is on the front wheels there is nothing in the trunk and the gas tank is removed.
Skip ahead to 3:20
[youtube]5ylj-6ORmik[/youtube]
Ohbse
10 kW
Arlo1 said:
I realise what was going on, I mean literally hitting the brakes to slow down - what peak speeds are you hitting in your street? Just hard to gauge as noise is a big indicator with anything ICE and quick that's obviously missing here.
Congrats on the first power induced traction loss
Things are getting better.
This is still at 1/2 power or less.
I have most of it narrowed down to making it work better in hall sensor operation (startup)
Once its in sensorless its AWESOME and I can roast the tires (on the dry as well )
I added caps to the Desat circuit and that helped it a little it got rid of some of the desat trips which I think were just noise going into the board that is used to shut the brain down. I replaced the hall sensor wire with a shielded ribbon and grounded 1 end. I now think I have noise on the throttle wire or the hall wires. But its coming.
Oh and I put the camera down at the end of the second video because a cop car showed up as I was using my cell to get video with 1 hand and driving a car with no seat no seat belt no insurance no plates and uhm... well I had safety glasses on....
[youtube]HMC-pwsC0Bg[/youtube]
[youtube]TEqsplvkzvc[/youtube]
I will try to make it a bit better then I will start finishing all the other things needed to make it my daily!
This is still at 1/2 power or less.
I have most of it narrowed down to making it work better in hall sensor operation (startup)
Once its in sensorless its AWESOME and I can roast the tires (on the dry as well
)I added caps to the Desat circuit and that helped it a little it got rid of some of the desat trips which I think were just noise going into the board that is used to shut the brain down. I replaced the hall sensor wire with a shielded ribbon and grounded 1 end. I now think I have noise on the throttle wire or the hall wires. But its coming.
Oh and I put the camera down at the end of the second video because a cop car showed up as I was using my cell to get video with 1 hand and driving a car with no seat no seat belt no insurance no plates and uhm... well I had safety glasses on....
[youtube]HMC-pwsC0Bg[/youtube]
[youtube]TEqsplvkzvc[/youtube]
I will try to make it a bit better then I will start finishing all the other things needed to make it my daily!
SlowCo
1 MW
I truly enjoy following your project and the progress your making.
As I mentioned earlier all this is way over my head but therefore also awesome to see what is possible.
As I mentioned earlier all this is way over my head but therefore also awesome to see what is possible.
Damn. I spent most of today working on this.
Sensorless is awesome! Will roast the tires at 1/2 power
Sensored sucks! Will not even take full throttle and as I push further on the throttle it seems to make less power!
In sensored is has a harmonic or a oscillation that I can't seem to go away.
Its like it goes though about 6 cycles making increasing power then back to low power and the oscillation starts over.
So today I adjusted the sensor gap its a sensor that is on the end of the motor shaft with a magnet centered in the end of the motor shaft. 1 sensor spits out uvw
I also removed the 4k7 pull-up resistors from the hall sensor input.
I added an 470 uf electrolytic cap to the brain and a 47uf ceramic cap to the sensor board.
I re-routed the sensor wire with a shielded wrap to the out side of the motor casing.
I removed 2 phase wires and swapped them to get it running FWD in FWD not that that should matter but there was a couple settings that made reverse WAY more powerful then forward.
I calibrated the halls in forward and reverse and the timing was about 18 deg apart so I adjusted it to the middle by modifying the HEX file. (yeah today I learnt hex)
I also tried a little more advance or retarded timing but I didn't play to much as it did not help
I then went back to being calibrated in forwards. I think the next thing I will try is calibrating the halls at different RPMs If they are changing timing with RPM then I will try to calibrated them at low speed and see what that does. Reverse and forward have different rev limits.
I played with the control loop co-efficients and I think its a good chance that the solution is in there.
When I changed the phase control loop for drive 2 to 500 and 50 reverse all the sudden got to be almost like it should have been it was reasonably powerful at ~ 1/2 throttle. 550 and 55 seemed even better. then I tried 600 and 60 but It did not solve it.
Then I tried 480 and 24 and that did not run well at all was a lot worse then 480 and 48
I tried playing a bit with the control loops for drive 3 but its not worth working on till I sort out the sensored operation. As most of the time in sensorless with the phase amps set at 425 it will lay rubber at speed on dry pavement with no problem.
HMMM....
Sensorless is awesome! Will roast the tires at 1/2 power
Sensored sucks! Will not even take full throttle and as I push further on the throttle it seems to make less power!
In sensored is has a harmonic or a oscillation that I can't seem to go away.
Its like it goes though about 6 cycles making increasing power then back to low power and the oscillation starts over.
So today I adjusted the sensor gap its a sensor that is on the end of the motor shaft with a magnet centered in the end of the motor shaft. 1 sensor spits out uvw
I also removed the 4k7 pull-up resistors from the hall sensor input.
I added an 470 uf electrolytic cap to the brain and a 47uf ceramic cap to the sensor board.
I re-routed the sensor wire with a shielded wrap to the out side of the motor casing.
I removed 2 phase wires and swapped them to get it running FWD in FWD not that that should matter but there was a couple settings that made reverse WAY more powerful then forward.
I calibrated the halls in forward and reverse and the timing was about 18 deg apart so I adjusted it to the middle by modifying the HEX file. (yeah today I learnt hex)
I also tried a little more advance or retarded timing but I didn't play to much as it did not help
I then went back to being calibrated in forwards. I think the next thing I will try is calibrating the halls at different RPMs If they are changing timing with RPM then I will try to calibrated them at low speed and see what that does. Reverse and forward have different rev limits.
I played with the control loop co-efficients and I think its a good chance that the solution is in there.
When I changed the phase control loop for drive 2 to 500 and 50 reverse all the sudden got to be almost like it should have been it was reasonably powerful at ~ 1/2 throttle. 550 and 55 seemed even better. then I tried 600 and 60 but It did not solve it.
Then I tried 480 and 24 and that did not run well at all was a lot worse then 480 and 48
I tried playing a bit with the control loops for drive 3 but its not worth working on till I sort out the sensored operation. As most of the time in sensorless with the phase amps set at 425 it will lay rubber at speed on dry pavement with no problem.
HMMM....
Arlo, maybe I'm totally wrong but... I can't forget lebowski's remarks here: https://endless-sphere.com/forums/viewtopic.php?f=30&t=36602&hilit=lebowski%27s+motor+controller+IC&start=1200#p1057635
about the control loop coefficients for different motors.
Trying to re-engineer lebowski's thougths and including newer learnings (PWM-freq.=21kHz, f_sample=29kHz) my calculations show:
* 'Fast' motor (Colossus) L=10uH, R=10mOhm, L/R=1ms -> ratio_12 = coef_1/coef_2 = ~15
* 'Slower' motor (MAC) L=139uH, R=80mOhm, L/R=1.7ms -> ratio_12 = coef_1/coef_2 = ~25
* Hub motor (MagicPie) L=560uH, R=60mOhm, L/R=9ms -> ratio_12 = coef_1/coef_2 = ~130
* Nissan Leaf motor L=230uH, R=10mOhm, L/R=23ms -> ratio_12 = coef_1/coef_2 = ~300
So maybe your actual ratio of 10 in drive 2 could be too low? What are your coefficients in drive 3 where it's running so nicely?
about the control loop coefficients for different motors.
Trying to re-engineer lebowski's thougths and including newer learnings (PWM-freq.=21kHz, f_sample=29kHz) my calculations show:
* 'Fast' motor (Colossus) L=10uH, R=10mOhm, L/R=1ms -> ratio_12 = coef_1/coef_2 = ~15
* 'Slower' motor (MAC) L=139uH, R=80mOhm, L/R=1.7ms -> ratio_12 = coef_1/coef_2 = ~25
* Hub motor (MagicPie) L=560uH, R=60mOhm, L/R=9ms -> ratio_12 = coef_1/coef_2 = ~130
* Nissan Leaf motor L=230uH, R=10mOhm, L/R=23ms -> ratio_12 = coef_1/coef_2 = ~300
So maybe your actual ratio of 10 in drive 2 could be too low? What are your coefficients in drive 3 where it's running so nicely?
okashira
10 kW
If it works sensorless then I dont think your control coefficents are the problem
I think the problem is you are using Hall effect.
The Leaf motor comes with a resolver. Why are you not using that?
I think the problem is you are using Hall effect.
The Leaf motor comes with a resolver. Why are you not using that?
okashira
10 kW
But how many hall sensors? You want at least 3° electrical accuracy for good FOC performance. The resolver can get you that... but a hall?
Can you code in speed and acceleration based interpolation of the hall outputs?
Can you code in speed and acceleration based interpolation of the hall outputs?
okashira said:
My UVW sensor is 1 unit giving 100% perfect and actuate hall timing which is calibrated by the brain.
http://www.rls.si/am4096-12-bit-rotary-magnetic-encoder-chip
The Code with a FOC algorithm in sensored mode will work right now by adding a bunch of timing to the old code it works great in sensored mode.
Lebowski has a v2.8 code coming that will help this all a lot I am using 2.6 right now and tried 2.7 and its almost as good.
Lebowski
10 MW
Halls are only used to start the motor, so low rpm. And in 2.80 it has a PLL for hall start that does interpolation (basically a low bandwidth phase filter).okashira said:
But once the motor turns standard running is done in sensorless FOC so not depending on hall accuracy.
Ok things are getting better. I am able to give it full throttle in sensored operation and I am able to give it full throttle in sensorless at higher speeds here is a couple videos.
Tonight I found I was able to tune the sensorless control loops and get it to allow full throttle at lower RPM right now I am running peaks of 550 phase amps and 250 battery amps.
[youtube]mu8zUJZZdx8[/youtube]
[youtube]V9ow9vxk7FU[/youtube]
Tonight I found I was able to tune the sensorless control loops and get it to allow full throttle at lower RPM right now I am running peaks of 550 phase amps and 250 battery amps.
[youtube]mu8zUJZZdx8[/youtube]
[youtube]V9ow9vxk7FU[/youtube]
zombiess
10 MW
What's preventing even higher power at this point? Tuning?
It looks like it's going to move quite well, going to surprise lots of people with it. I love sleepers.
It looks like it's going to move quite well, going to surprise lots of people with it. I love sleepers.
zombiess said:What's preventing even higher power at this point? Tuning?
Traction. It lays rubber the same at 450 phase amps as it does at 550 phase amps
But full power ( actually full torque) at a launch is limited by code. Lebowski has new code to solve this. Changing how the sensored operation works. Right now Sensored operation is week and inefficient and I can't run sensorless at really low speeds and full power yet. Tuning has helped me get full power at lower speeds in sensorless but its not done so yes more tuning will help. I did about 40-50 runs in front of my shop laying rubber Wednesday night.
It looks like it's going to move quite well, going to surprise lots of people with it. I love sleepers.
I plan to leave it partially ugly for that reason
I took my brother, my wife and an electrician friend for a quick ride then let them all drive and lay rubber they all loved it and got out giggling!
Phase amp settings at 450 amps PEAK (~320rms) with battery amp settings at 275amps (just kept it down for testing and saving the battery)
Both of those numbers are 1/2 of what I am aiming for but I will not bother turning it up more till it has some better tires and some drive shaft loops as they are welded.
This has been so worth all the work. Its now ready for some rust repair and to install the chargers and drive it daily....
Thank you so much everyone for all your help and support!
[youtube]JYO59Lun2Es[/youtube]
Both of those numbers are 1/2 of what I am aiming for but I will not bother turning it up more till it has some better tires and some drive shaft loops as they are welded.
This has been so worth all the work. Its now ready for some rust repair and to install the chargers and drive it daily....
Thank you so much everyone for all your help and support!
[youtube]JYO59Lun2Es[/youtube]
SlowCo
1 MW
Awesome result!
Time to order some serious tires something like:
http://www.toyotires.ca/tire/pattern/proxes-r888 8)
Time to order some serious tires something like:
http://www.toyotires.ca/tire/pattern/proxes-r888 8)
zombiess
10 MW
I run Proxes R888 on my Z06 at 640RWHP and love them, it hooks on the street in 2nd gear and hazes them in 1st.
A big warning with them is not to drive them if its < 10C outside because the rubber compound becomes brittle, so might not be the best tire if you plan on any cold weather driving.
A big warning with them is not to drive them if its < 10C outside because the rubber compound becomes brittle, so might not be the best tire if you plan on any cold weather driving.
zombiess said:
Thanks Zombies. I have winter and all season tires but when I want to have fun in the summer I will get something sticky like this!
I lowered the PWM frequency today back to 6khz (from 10) and its working great! 6k is my goal the oem uses 5k and I think 6 is a decent number to keep current under control as well as not work the cap to hard. The cap has a thermistor so I will keep an eye on it and up the PWM a touch if needed.
I just have some rust to work on and a few tweaks and its time to get ripping on the highway.
The power during sensored operation is still low but I did make it better and I have been lowering the transition right now its at 1500 rpm it gets about 1 wheel revolution until it transitions into sensorless and roasts the tires FEALS SO GOOD TO BE HERE!
Ohbse
10 kW
Awesome vid!
The soft start might actually be a very good thing for prolonging the lifespan of your welded shafts, i've exploded my share of shafts and CV joints and it's always the shock loading that does them in. After I started pre-loading the drivetrain on the ebrake before launching on slicks it vastly reduced the breakages.
Have you thought about selling a package for enthusiasts to achieve the same results? I'd be keen and would love to contribute towards all your investment (time and dollars I'm sure!)
The soft start might actually be a very good thing for prolonging the lifespan of your welded shafts, i've exploded my share of shafts and CV joints and it's always the shock loading that does them in. After I started pre-loading the drivetrain on the ebrake before launching on slicks it vastly reduced the breakages.
Have you thought about selling a package for enthusiasts to achieve the same results? I'd be keen and would love to contribute towards all your investment (time and dollars I'm sure!)