Arlo1 Inverter build for Emrax 268

Wächter

1 mW
Joined
Dec 11, 2012
Messages
19
Location
Stemwede
This topic is about an inverter based on the "Hi power inverter for Nissan leaf motor. Dyno's 302.3hp p15" from Arlo1. (https://endless-sphere.com/forums/viewtopic.php?f=30&t=63982)

It will be used for an Emrax 268 (medium voltage) motor.
I'm finally in the stage were everything is complete and i can make the first tests.
Here are some pictures of the build process:

62Mn0k7.jpg



Dq9WFuK.jpg


mMCjrFj.jpg


yV7tNtD.jpg


The bus bars are water cut and the caps have 75uf each.
Right now i'm trying to figure out the right settings for the lebowski ic (latest 2.A1 firmware). The desat protection is tripping really fast. I started with the settings Arlo1 used in one of his videos.
I made a video where you can see the motor wiggling around for a few seconds and then desat is triggered:

[youtube]nWoIjBfbVrg[/youtube]

I'm running a 380V system. I can also force the motor to spin up when i'm closing the throttle really fast but only for around 2 seconds.
I will now test various settings, but if anybody has a good idea which settings will possibly improve the performance i would be grateful to hear it :D . I also uploaded the settings hex file. This is where i'm starting off from.

And here is a picture of the inverter inside the car (just for testing :D):

X3YX9Xr.jpg
 

Attachments

  • Setting.txt
    3.5 KB · Views: 320
Looking at the settings, I would go through the setup menus starting from main menu option c. Looks like some settings come from Arlo1's motor (like the motor impedance) and his controller (like the current sensor offsets). Also, at the moment it is in sensorless so it cannot deal with a heavy load on startup (wheels off the ground should be OK, but to start with wheels on the ground you'll need hall sensored mode).

Just go through all menu's from c, use autocomplete where available.
 
Yes, the wheels are off the ground at the moment and i have the hall sensors ready to be connected to the brain board.
I stay in sensorless for now, because it is not possible to calibrate the hall sensors with the fast desat triggering.
The current sensor offset measurement is also triggering desat. Same for the motor impedance measurement. I had to reduce the current to around 10A to successfully complete the motor impedance measurement. I also tried putting the values from the motor manual in there (126/118 Ld/Lq uH and 11,5mOhm phase resistance) but that didn't changed much.
I will go though the menu's again and play with the settings.
 
Yes, the polarity of the current sensors is correct.
I also tested the inverter with a 12V battery at first. With 12V i was able to perform the current sensor calibration. I also could spin the motor without desat triggering. Now with 380V you can hear the calibrating noise for one moment and then it shuts off.
 
sounds like something is triggering the desat. Could be desat too sensitive or something wrong with the gate drive... Did you try the PWM test function (without motor connected) for 5, 50 and 95% dutycycle and check the waveforms on a scope for all 3 outputs ?
 
I only checked the pwm with 50% dutycycle and it looked ok. I will also check with 5% and 90% and i'm taking a look at which phase is causing the desat. maybe it's always the same.
 
Ok, all the phases are behaving the same way. So it's not like one igbt is causing desat all the time.
I checked the gate signals again. With 5%, 50% and 90% dutycycle.

50% dutycycle:
jRkPyIQ.jpg


5% dutycycle:
IAR8ap3.jpg


90% dutycycle:
fqsS7aP.jpg


I think they look okay. I also checked the fault signal.
As soon as i give a little bit of throttle you can see the desat triggering (No arduino attached, you can see the 25us the igbt driver is shutting the igbt down):
ncrBIaV.jpg


Maybe there is something wrong with the desat circuit itself (false diode?)
 
Yes, i did that. The polarity is correct and the channels of the pwm output and the current sensor match up. There is one ribbon cable for each phase. Not really much that could go wrong on that part.
What is the relation of the current sensor with my desat problem? My understanding was that the desat circuit and the current sensor are not affecting each other. Please correct me if i'm wrong, maybe i miss out on something. And thank you for helping me on this problem :)

The motor spins when i disconnect the arduino (which is shutting down the brain in case of a desat fault).

[youtube]piRafWdmps4[/youtube]
 
Well the current sensors have to match up with the motor phases. The controller IC gets very confused when it outputs a forward rotating voltage but sees a reverse rotating current (because 2 channels were swapped). In a situation like that it will have no control over the current and it will trip the desat (or blow the fuse).

What might be tripping the desat are the PWM frequency related currents. A low PWM frequency, combined with a high voltage and a low motor inductance will give very high PWM currents. The controller IC has no real control over this as the currents are dictated by the aforementioned parameters. Maybe you have much higher PWM currents than Arlo1, either due to more voltage or lower motor inductance.
What helps here is to increase the PWM frequency, but you must be sure of course that the output stage can take this (increased switching losses etc).

When you change the PWM frequency you need to re-autocomplete the PWM menu which changes the loop sample frequency. This parameter is used everywhere, meaning that you will need to re-do most of the menu's.... (redo the FOC measurement, re-set the erpm limits etc etc etc)
 
Good news! I tried several changes in settings and what really helped me was to reduce the deadtime. I reduced it from 2000ns to 500ns and now it's running really good (wheels off the ground). I also changed the pwm frequency to 10 kHz but that didn't changed much. I think i will switch it back to 6kHz and see how it behaves.
Thanks for your help Lebowski. I will post more updates in this topic. I have to make some other changes on the car and then i will try to put it on the street (~next week).

Here is a video with the new settings and with hall sensor mode:
[youtube]CevWR_y6eKc[/youtube]
 
Hi

EMRAX is outrunner motor, so it has a lot of poles, 20 poles to be exact. This needs a high frequency to run at good RPM. Roman told me to use at least 800Hz to reach 5000rpm. At that rpm you would need to increase pwm freq to have better resolution. He suggested 16kHz or 18kHz and drivers have to cope. I think at low freq you will notice this as cogging or noise in motor if you dont use higher freq.

A
 
arber333 said:
Hi

EMRAX is outrunner motor, so it has a lot of poles, 20 poles to be exact. This needs a high frequency to run at good RPM. Roman told me to use at least 800Hz to reach 5000rpm. At that rpm you would need to increase pwm freq to have better resolution. He suggested 16kHz or 18kHz and drivers have to cope. I think at low freq you will notice this as cogging or noise in motor if you dont use higher freq.

A

I took a video of the first test drive i made last week and it matches with your description. There is much more noise coming from the motor compared to the H300 controller i'm normally using.
Here is a part of the test drive video:

[youtube]jYnFJeBHMaY[/youtube]

I will try different settings with higher pwm, but first i have to make some changes to my case and add a water cooled plate.
 
Nice drive! It is good to see Emrax run on DIY inverter.
Do you run it sensorless or you use UVW sensors?

Probably noise at low RPM is because of stepped PWM instead of sinus. Or PWM is not fast enough. Try at least 16kHz.
I tried to run motor with 48V battery and one chineese controller in sensorless. IT WORKED VERY GOOD! Then i tried to run it also with Johannes controler as well as Pauls. Roman gave me some pointers. I wasnt able to run it very good under torque. But that was 2 yrs back.
It was all a bit theoretical since my controller is in my car and i need it every day.

A
 
Something that will help reduce the noise in drive_3 is to reduce this coefficient:
Code:
n) immediate motor phase step: 48
in the control loop coefficients menu. You can set this coefficient all the way down to 0, but it may (or not) cause instability. I am not 100% sure this coefficient is absolutely necessary, but theory says it is so I put it in.... It is quite a noisy coefficient though.

Can you post the hex ?

p.s. I'm talking about the hissing noise, any sort of tone is from the low PWM freq...
 
Congratulations on running a difficult motor on DIY controller!

You could add a leaf transaxle now if you wanted more power. :)
 
Awesome to see I helped inspire some more cool projects.

I used a 200pf blanking cap as the 100pf was to fast for the slow switching IGBTS

Also I have run it to 1000hz.

15000 rpm with a leaf 4 poll pair motor.

You should be ok to at least 5000 rpm but it will depend on the current rise time in the motor you have.

I had to spend a lot of time "tuning" the inductance/resistance settings in the controller.

When I get back to Canada I will try to find my settings.

But if it measures 150uH at low current like 200 amps peak I found setting it to ~65-80uH worked a lot better for full power with field weakening.

PS nice bus bars.

I have some custom bus bars water jet with a DXF file to share when I get home as well.

-Arlin
 
neptronix said:
Nice. Any updates on this?

Right now there are no big updates. I made a custom cooling plate for the controller, but it is a bit leaky :roll: . No new tests with the emrax motor at this point. I also made circuit board that will fit inside the controller case.
I got two Nissan Leaf Motors and i will build 2-3 more controllers, but that will take some time.
I will than build a test bench for the motors so i can adjust the controllers correctly.

MJafAyc.jpg
 
If you order Fuji IGBTs we should get a group buy going.

They have a 24 week lead time right now. I was looking at possibly getting 100 of them.

Another point is the PWM frequency will effect the noise but you want to keep it low for lower switching losses.
But you need the PWM to be high enough to control the current smoothly and have enough PWM cycles at max RPM.
 
Hi

I managed to run a "black project" with Volt inverter and Lebowski brain. On 6Khz it was quite noisy, while noise subsided a lot on 12Khz. I even tried on 15kHz but it seemed motor bucked more there. I managed 4300rpm only with voltage and 5500rpm with 75A weakening.
Because of 20 pole count control is very precise. Leaf motor bucks a lot more in sensorless. Or maybe i have to ammend wiggle parameter?
Now i have to figure out how to add 3x hall sensors to Emrax :).
https://www.youtube.com/watch?v=dRgfqRJDuTw
 
Back
Top