Porsche Boxster with Siemens and DMOC

[Isaac96]

Hi all,
I thought I'd start a project thread as my car is nearing completion.
It's a 2000 Boxster S in a very nice silver color. The usual IMS bearing issue destroyed the motor so my dad and I were able to get it for $3500.
First version used a Siemens 1PV5135 of the same sort that Damien has, along with the Azure Dynamics DMOC645. The battery pack was a complete Leaf system, using the Leaf BMS and an Arduino for voltage monitoring. I also had CHAdeMO working sometimes, but didn't fully test it beyond a simple communications test.
Drivetrain is clutchless through the 6-speed manual.

Unfortunately the DMOC is a terrible inverter; it had low power limits, strange overtemperature issues, and was generally problematic. And the Leaf batteries turned out to give about half of their rated capacity, resulting in 35-mile range.
So both a new battery pack and new inverter were in order. This was before the Prius inverters were well documented, so I decided to install a Huebner inverter kit into the DMOC.

Version II is getting close to drivability. Yesterday I finished installing the batteries, 12 of the LG Chem Pacifica modules for 32Kwh and 360v. This should give a range of about 100 miles, and only 50 or so is needed for my daily commute to school (which is cancelled these days due to COVID).
Pictures are coming soon.
I have a website for the car: electricboxster.com
It has not been updated since the car was rebuilt, but it has some nice pictures and drivetrain info.

I am currently struggling with the inverter and motor.

Voltage and current sensors are working correctly; pedals are calibrated and potnom follows the pedal nicely.
The encoder also seems to work right.
However, the motor will not budge until potnom reaches 65. (I believe this is exact).
And when potnom does reach 65, the motor shakes for a second or two and then shuts off due to overcurrent.

I've been logging different variables and the 'amp' variable does not change until potnom reaches 65. 'ampmin' is set to 10, which is default.
There does not seem to be any derating happening.
I'm stumped. Any ideas?

Thanks!
-Isaac
[Parameter file is attached; firmware is 4.76]

[arber333]

Hi

I would advise you to try increasing ampmin and boost parameters. I use 7000boost and 21ampmin in my inverter. I also set deadtime to a high value of 3us. But that is just my drivers.

Try to read the wiki in tuning guide under parameters section.

Also I think Damien posted 1pv motor parameters somewhere on dielectriccar...

[Isaac96]

Hi Arber - thanks for the advice!
I've looked at Damien's parameters (which are on the wiki).
Looks like his boost is at 3600 so I'll try that first and then jack it up more if that doesn't help.

I will follow the tuning guide once I can get the motor spinning reliably, and also some cooling to motor and inverter.

How does ampmin affect the motor's operation? (Never mind, just checked the wiki. Seems helpful)
I'm not seeing amp changing as I apply the throttle until potnom reaches 65, will keep investigating and come back with more questions.
Thanks!
-Isaac

[Isaac96]

Just did a good bit more testing - still no luck.

I tried boost values of 500, 1500, 3000, and 5000.
ampmin of 5, 10, 20, 30.

It doesn't seem like the parameters are the issue. The motor simply doesn't move no matter the setting unless potnom reaches 60. Then there's an instantaneous overcurrent. The inverter feeds no power until potnom hits 60.
This is weird!
Should I try downgrading firmware? What's a good stable version for an AB encoder and async motor?
Thanks!
-Isaac

[arber333]

Ah you have AB encoder! Did you spin it with wheels in the air?
Try switching two phase wires on the inverter.
Or two signal wires of AB encoder.
If you set this incorrectly motor will not spin.

If that is good, what does your firmware say? You need sine version with AB encoder selected. My car works very good with 4.76R sine version...

[Isaac96]

Yes the wheels are in the air.
I just reversed the encoder wires and behavior was exactly the same.
AB encoder is selected.

Turns out I was on the 4.73 sine release. Updated to 4.76 and there was no change.

I changed the slipstart setting to 10. As a result the motor turned on and hit overcurrent at 30% potnom.

I am also logging amp to see what the inverter is outputting. It stays zero until the 60% potnom, then jumps to 4000 or so when the motor does its 'turn on and die' act.

So it seems that the inverter is being prevented from outputting any current until potnom hits 60% -- and then the current gets dumped all at once! Could this be linked to derating? I have no errors besides the overcurrent. And the motor and heatsink temperatures are within the normal range.

Still stuck!

-Isaac

[arber333]

Ok, can you try to rotate wheels in manual?
So put inverter in manual mode and set some sane boost.
Then add manual % current and some hz in small start like 1, 3, 5, 8, 15...etc.
When you get the wheels spinning you increase frequency
To 30 or 50hz and observe behavior.
I think you would have to have more than 50% ampnom set for this.

[arber333]

What is your AB encoder pulse count? Usually those encoders use 64 pulse count for rotation per channel. Otherwise you may try 128 or 256.

Try that first.

Also if you use 4500 boost it is used for 500v systems. I use 7000 on 380v because motor is wired at 180Vrms. Yours is probably too. And if you use 200V main bus this is not enough. Try 9000 and raise it if it is not enough.
Rotor must overcome its own resistance and for that you need to feed it volts as well as amps.

A

[Isaac96]

Good morning - stayed up late trying to get it working, still no luck.

My encoder is 64ppr.

Main bus voltage is 360v; Damien used boost of 3600 with a higher voltage. Should I just set it higher? Would that cause any problem besides maybe some inefficiency?
I will try manual mode soon when it's warmer out in the garage.

Thanks!
-Isaac
P.S. Regen seems to work correctly, if I spin the wheels when the car is powered on there is noticeable resistance and the brake lights turn on.

[arber333]

Hi

Did you calibrate current sensors? Which ones do you use?
https://openinverter.org/wiki/Schematic ... sor_boards

Hm... which board version are you using? If you have R2 board with sensor board that gathers current sensor, voltage and temperature data in one board you might have entered POSITIVE OClimit value, whereas you should be using NEGATIVE, so eg. -800A! Verify this first. I am not sure how new firmware handles this now.

WARNING! You are using seriously short deadtime. I recommend you set it to 160! I dont know what kind of drivers you use, but it is allways good to have more blank spoace between signals for testing.

Also make this parameters 200 for testing.
"iacmax": 5000,
"idcmax": 5000,
"idcmin": -5000,

[Isaac96]

I just recalibrated current sensors - ended up with the same gain settings.
Will try reversing the sign on ocurlim. From previous testing I think the sign is correct already, as negative ocurlim changes the PWM output from the STM32 which drives the overcurrent comparators.

I'm using the drivers from the Rev2 kit; previously when I scoped the IGBT outputs the deadtime seemed sufficient (but I'm not much of a scope user).
Will increase deadtime and see how it does.

[arber333]

I'm using the drivers from the Rev2 kit; previously when I scoped the IGBT outputs the deadtime seemed sufficient (but I'm not much of a scope user).
Will increase deadtime and see how it does.

Damn! Those drivers are not what i would put in a car and trust with 350V and IGBTs... Not at high power.
Did you observe if you get 5V signal to drivers?

Also you can do a nice test if your IGBTs work.
Disconnect HV and motor cables. Put 12V on DC link. Put three 12V lights on phase outputs and run manual mode. Then in manual mode enter fslipspnt 1Hz and 100% ampnom. This should cause each light to blink 1/3 second. That way you confirm your drivers and IGBT still work.

You can actually do the same with 3x incandesent light bulbs and high voltage. Just in case choose 60% power then .

EDIT:
Why do you use "potmax": 2200? Do you not have enough sensor voltage? This is a fraction of what pot should show up to 4095.

A

[Isaac96]

I just tried changing iacmax, idcmax and idcmin to 200 and -200 as you suggested. Same results.

The deadtime of 160 made the overcurrent happen slower, but still happens every time.

Something strange is happening however. On startup, every single time the inverter is turned on, a "DERATE - CURRENTLIMIT" message appears. What could cause this?

I will try to get the motor wires disconnected and see if the IGBTs are working. They all functioned correctly before installing in the car, but I'll see.

What drivers would you recommend instead of the ones I have? I am open to finding a more permanent solution once the car is running.
IGBTs are Powerex/Mitsubishi CM600DY-12NF.

Thanks!
-Isaac

EDIT The pedal only shows up to 2200. It's the original Porsche drive-by-wire with dual pots. I don't think it's a problem, but I can change for testing if it could help.

[johu]

Damn! Those drivers are not what i would put in a car and trust with 350V and IGBTs... Not at high power.

Whenever you say that I have to mention that they work flawlessly for me and others since many years. And that is even the old ones, 2.5A and -15/+15V. Of course they are not as safe as Toyota drivers...

Something strange is happening however. On startup, every single time the inverter is turned on, a "DERATE - CURRENTLIMIT" message appears. What could cause this?

What is iacmax set to?

[Isaac96]

Something strange is happening however. On startup, every single time the inverter is turned on, a "DERATE - CURRENTLIMIT" message appears. What could cause this?

What is iacmax set to?

iacmax is at 5000

[Isaac96]

Okay I chased the CURRENTLIMIT error for a bit. Whenever the main contactor turns on (and precharge contactor turns off) there is a current spike of some sort. When I set ocurlim to 10 then both the derate message and a STOP - OVERCURRENT error appeared on the contactor event.
What on earth is going on here? I am going to start disconnecting contactors and things. Maybe there's an inductive spike feeding back? I have diodes across all the coils though.

[arber333]

Whenever you say that I have to mention that they work flawlessly for me and others since many years. And that is even the old ones, 2.5A and -15/+15V. Of course they are not as safe as Toyota drivers...

Ok, ok, any day now i will continue with my Concept driver universal board design. For those uberentusiastic people who still want to build their own inverter. I have it prioritized somewhere in the middle between Prius charger board and Lebowski dual Ampera board .

[Isaac96]

I just spent a few hours in continued testing of motor and inverter.

All this testing was done in manual mode. Sometimes the car was in gear and sometimes it was not. This had no effect on the results as I tried most situations both ways.

First: Encoder disconnected.
-Boost values of 5000 and 7000.
Procedure:
1. Turn key to START
2. start 2
3. set fslipspnt (value)
4. set ampnom (value)
With fslipspnt values of 1, 3, 5, 7, 12, 15.
Resulting in the motor turning smoothly at very specific 'ampnom' settings, usually around 40.
When ampnom was increased by 3 or 4 points the motor began to shake violently and eventually trip overcurrent.

Next round of testing: Encoder connected. This is the integrated one in the Siemens motor, and I have verified that it tracks direction and angle nicely.

Same procedure, similar values. And very similar results.
At the 'good' ampnom setting the motor would move for a bit and then begin to shake violently and trip overcurrent.

So then I thought: maybe the encoder is reversed? Maybe that will help!
So I reversed the A and B encoder wires. This resulted in no movement at all from the motor and instead instantaneous overcurrent trip.

So as it stands I'm no closer. I have verified that my power stage is fully functional. Current sensors seem to work. Encoder also seems to work.

What else can I test or change?

Thanks!
-Isaac

[arber333]

Hi

But your motor did turn smoothly at some setting?
What happens if at this amp setting you increase commanded frequency to 30hz or 50hz? In steps of 5 from 15hz please so you don't trip oc. You should have wheels speeding at this speed. Maybe your motor winding does not agree with too much current.
Don't forget in manual you do not set any slip. It is just the V/Hz you are searching for. So I like to test a new Igbt power section with small AC motor like 250W of power.

Try to start in slip mode and set slipmin at 1hz and max at 3hz. Ampmin at 10 and boost at 5000. Don't use regen pot for now yet. Set both of its parameters at 4096.
Calibrate throttle before applying start pin.
What happens when you apply throttle?

[Isaac96]

Yes, the motor did turn smoothly. However when I increased ampnom by 3 or 4 points it began to vibrate.

I was able to get the motor up to 20hz before it shut off from overcurrent. I really think my sensors are being noisy. How can I reduce current sensor noise?

I set up the pedal as you advised and the same overcurrent happened immediately.

The current sensors do seem to be problematic...

[johu]

A sudden reaction to ampnom is normal at some point, you will be saturating the stator which makes it almost a dead short.
Can you plot measured currents and a picture of your setup? Can you measure actual phase current with a clamp meter?

[Isaac96]

Can you plot measured currents and a picture of your setup? Can you measure actual phase current with a clamp meter?

Yup, I've got a clamp meter. I'll try to plot reported current vs actual phase current - the phase wires are a little tough to get to.
I'll have to do all this tomorrow though.

This is the inverter:

It is now installed in the trunk of the car.
I can get more pics tomorrow when there is more light if that is necessary.
Thanks!

[johu]

Maybe you could rotate the sensors on the cable so they point outwards, away from the power modules. Ideally while observing their value.

[Jack Bauer]

That looks so familiar:) I ran a dmoc / siemens combo in the Panzer for years with a standard Johannes kit (actually the first one I bought from him) using both automatic and manual gearboxes. Pushed 130kw max and no blown transistors.

[Isaac96]

Maybe you could rotate the sensors on the cable so they point outwards, away from the power modules. Ideally while observing their value.

I rotated the sensors and recalibrated. Gain was slightly lower.

Still the same overcurrent tripping though...
I put a clamp meter on the phase wires and measured ac current. At ampnom = 25 the ac current was 3.6A. And at ampnom = 26 the overcurrent tripped.

Should I try disconnecting the sensors altogether? Or setting gain very high?
I also have some LEM HAS sensors that I sorta planned to install once I had the inverter going. Would those be more immune to noise in the signal lines?

That looks so familiar:) I ran a dmoc / siemens combo in the Panzer for years with a standard Johannes kit (actually the first one I bought from him) using both automatic and manual gearboxes. Pushed 130kw max and no blown transistors.

Your videos on the inverter build brought me to Johannes' system, so I can blamethank you for this.

-Isaac

EDIT Here is a plot of il1 and il2, as well as a screenshot of values at the point where overcurrent hit.