Porsche Boxster with Siemens and DMOC

[Isaac96]

I just did a ton more testing and changing of things and got rather better results. Turns out I forgot to place 1nf caps on the current sensors, which was increasing the noise. After installing caps there the noise has greatly decreased.

I'm still getting the problem with the angle variable changing. I can get the motor to run with pedal commands but it is jerky and hits overcurrent rather quickly. I downgraded to firmware 4.12 as joromy suggested and that did not improve things.

What could cause the angle variable to constantly change without signals from the encoder? I looked at the code a little and it's all over my head; the current sensors were bad enough.

Thanks!
-Isaac

[Isaac96]

I'm pretty sure the encoder is open collector, just because of its wide voltage range. I'll throw 12v at it and see what happens.

The encoder is now being powered from the ENC_IN_5V pin on the control board; R7 (which is intended to supply power for an IR diode there) has been bridged as advised in the Rev2 schematic.

Can you source power for the throttle pedal somewhere else. Maybe pedal is causing some voltage fluctuations.
I source throttle Vcc and GND at JP4 which also supplies sensor board.
I actually provide encoder power from 15V IGBT driver power supply which is super decoupled and strengthened by some tough caps.

SUCCESS!!!!
This was the clue to my problem! It turns out the gate drivers were pulling the rest of the power supply low and causing all the weirdness with overcurents. I powered the gate drivers with a separate 5v supply and the motor works perfectly!!!
(well, besides a little shaking sometimes. I'll follow the tuning guide in the wiki once I get the car down and rolling )

So I guess it's time to go write up my failures in the troubleshooting wiki.

Thank you Arber and Johannes!

-Isaac

P.S. I'll have lots of pictures and some videos soon

[Isaac96]

I'm pretty sure the encoder is open collector, just because of its wide voltage range. I'll throw 12v at it and see what happens.

The encoder is now being powered from the ENC_IN_5V pin on the control board; R7 (which is intended to supply power for an IR diode there) has been bridged as advised in the Rev2 schematic.

Can you source power for the throttle pedal somewhere else. Maybe pedal is causing some voltage fluctuations.
I source throttle Vcc and GND at JP4 which also supplies sensor board.
I actually provide encoder power from 15V IGBT driver power supply which is super decoupled and strengthened by some tough caps.

SUCCESS!!!!
This was the clue to my problem! It turns out the gate drivers were pulling the rest of the power supply low and causing all the weirdness with overcurents. I powered the gate drivers with a separate 5v supply and the motor works perfectly!!!
(well, besides a little shaking sometimes. I'll follow the tuning guide in the wiki once I get the car down and rolling )

So I guess it's time to go write up my failures in the troubleshooting wiki. Do we have a specific page for that yet? I only see a Tesla one so far.

Thank you Arber and Johannes!

-Isaac

P.S. I'll have lots of pictures and some videos soon

[johu]

Congratulations
Of course that's a bit of a design problem then... how much are the gate drivers pulling?

[Isaac96]

I actually have no idea. I'm running them off a LM2596 for now but I will try my lab PSU and see what they take. The DC/DC converters, Murata MGJ2D051509SC, claim to draw 2W each and there are 6 of those. So it could be 12W, which is 2.4A, which is probably too much considering that the LM2596 for the logic maxes at 3A.

In other news I'm getting shaking on startup still. I will try to tune carefully and document the tuning process and see if I can make it repeatable.

[arber333]

I actually have no idea. I'm running them off a LM2596 for now but I will try my lab PSU and see what they take. The DC/DC converters, Murata MGJ2D051509SC, claim to draw 2W each and there are 6 of those. So it could be 12W, which is 2.4A, which is probably too much considering that the LM2596 for the logic maxes at 3A.

In other news I'm getting shaking on startup still. I will try to tune carefully and document the tuning process and see if I can make it repeatable.

Maybe you could just dedicate one LM2596 solely to power the drivers. I think Johannes 5V power converter is a good design. I managed to use it for a long time without brownout.
What is the rating on your IGBT drive resistors? Did you use 10R or 6R8? I found out 6R8 was crisper response from IGBTs.

However i needed to use longer dead time. With my design inverter when i first tried to drive off with desat drivers they became saturated and shutoff. I spend hours trying to calibrate little RC link to desat. Later on i found out i just needed to add more dead time and increase boost to 7000 to get a good start. That is what i found for my motor. Dead time 3us is somewhere around 180 in STM32 terms...
Now with Chevy Volt inverter i use 3.1us deadtime and it works good on every PMSM or ACIM motor i tried with.

[johu]

What is the rating on your IGBT drive resistors? Did you use 10R or 6R8? I found out 6R8 was crisper response from IGBTs.

Latest kits come with 4R7 to make use of the 5A drive capability.

[Isaac96]

]

Maybe you could just dedicate one LM2596 solely to power the drivers. I think Johannes 5V power converter is a good design. I managed to use it for a long time without brownout.
What is the rating on your IGBT drive resistors? Did you use 10R or 6R8? I found out 6R8 was crisper response from IGBTs.

However i needed to use longer dead time. With my design inverter when i first tried to drive off with desat drivers they became saturated and shutoff. I spend hours trying to calibrate little RC link to desat. Later on i found out i just needed to add more dead time and increase boost to 7000 to get a good start. That is what i found for my motor. Dead time 3us is somewhere around 180 in STM32 terms...
Now with Chevy Volt inverter i use 3.1us deadtime and it works good on every PMSM or ACIM motor i tried with.

I built up a LM2596 for driver power so that's working nicely now.
The BOM which came with my kit states 10R resistors so that's probably what I have. I'm keeping deadtime at 160, since I don't exactly need the best performance (though it would be nice). I've still got the car on the lift so once I get the BMS hooked up and the cooling loop done I'll tune properly. It sounds great though! Excellent throttle response and the regen seems to be working very well.

[Isaac96]

It's been a while since my last update. But I have great news:
Today I finally managed to drive the car under OpenInverter control!
I threw out the DMOC (after blowing 2 IGBTs and 4 stm32 boards) and installed a Volt/Ampera inverter with Arber's Rev2 to Volt interface board. However, I was having issues with EMI throughout the vehicle, which trashed the encoder signals for some reason. Thus pedal control was impossible. So for the last couple of weeks I worked on shielding the phase and power wires going into the inverter, with some gorgeous EMI shielding braid and HV orange heatshrink. I also optoisolated the encoder for good measure.

As a result, I was able to drive the car around the block! (It threw some TMPHS messages because of bubbles in the inverter, but those came out eventually). It's jumpy right now and the regen is way too powerful. But that can all be fixed with some tuning. It already feels more powerful than the DMOC setup
I'll get some pictures and maybe video up soon. Super happy to finally have a properly working car!! Thanks everyone for your assistance!

-Isaac

[mdrobnak]

Glad to hear! I can't wait to see what you can get out of it.
Congrats on getting it working.

How are you interfacing with the rest of the car? I know 996 has some CAN stuff, I am under the assumption that 986 is exactly the same.

-Matt

[Isaac96]

Thanks!

My vehicle is MY2000 -- I haven't been able to find anything running off CAN. Which 996 years are you familiar with? I'd love to be able to use more CAN in the vehicle, but as far as I know the Boxster got nothing until at least 2002.

-Isaac

[tom91]

Thanks!

My vehicle is MY2000 -- I haven't been able to find anything running off CAN. Which 996 years are you familiar with? I'd love to be able to use more CAN in the vehicle, but as far as I know the Boxster got nothing until at least 2002.

-Isaac

How does you Rev counter work? Also the fuel gauge?

[Mhs]

As a result, I was able to drive the car around the block! (It threw some TMPHS messages because of bubbles in the inverter, but those came out eventually). It's jumpy right now and the regen is way too powerful. But that can all be fixed with some tuning. It already feels more powerful than the DMOC setup
I'll get some pictures and maybe video up soon. Super happy to finally have a properly working car!! Thanks everyone for your assistance!

-Isaac

Well done, would love to see the pictures and videos

[mdrobnak]

Thanks!

My vehicle is MY2000 -- I haven't been able to find anything running off CAN. Which 996 years are you familiar with? I'd love to be able to use more CAN in the vehicle, but as far as I know the Boxster got nothing until at least 2002.

-Isaac

Ah, yes. I' not sure what year vehicles I've worked with. but the 996 race car was definitely CAN driven. I think that was a 2004 though.

If it's not CAN driven, to answer Tom's question - it's probably a standard tach signal for a 6-cylinder engine. Fuel gauge is harder. That's going to be a resistor in the stock config, not sure what I'd do with that. (I'll have a similar problem on my BMW, the instrument cluster is what reads in the fuel level, same type of setup.)

-Matt

[tom91]

Thanks!

My vehicle is MY2000 -- I haven't been able to find anything running off CAN. Which 996 years are you familiar with? I'd love to be able to use more CAN in the vehicle, but as far as I know the Boxster got nothing until at least 2002.

-Isaac

Ah, yes. I' not sure what year vehicles I've worked with. but the 996 race car was definitely CAN driven. I think that was a 2004 though.

If it's not CAN driven, to answer Tom's question - it's probably a standard tach signal for a 6-cylinder engine. Fuel gauge is harder. That's going to be a resistor in the stock config, not sure what I'd do with that. (I'll have a similar problem on my BMW, the instrument cluster is what reads in the fuel level, same type of setup.)

-Matt

Matt I am asking as I have converted a 2002 late year Boxster, this is a canbus driven gauge, Speed, Rpm and temperatures are directly from canbus. The fuel gauge is the hard one as It uses a saddlebag arrangement, sender till about 1/4 then it uses consumption of the engine to figure out the rest.

[Isaac96]

How does you Rev counter work? Also the fuel gauge?

Okay this is a little embarrassing...

A year ago or so (it's been a long project) I had an Arduino running the tach with a simple square wave into pin 9 of the OBDII connector. I think I just had a simple open collector transistor with a resistor inline. I was able to make the tach display whatever I wanted.
Then of course I messed up. At the time I was using the DMOC controller with the GEVCU to control it over CAN. So of course I wanted the RPM from the GEVCU to control the tach. I edited the GEVCU code to output a similar square wave on one of its open collector inputs. Then I hooked it up -- but WITHOUT an inline resistor!

Let's just say the tach doesn't work anymore.
More recently I tried to control the stepper directly with an Arduino; that will eventually be how it's done, but for now I haven't made circuitry small enough to fit in the cluster.


I was also working on the fuel gauge for a bit, trying with different transistor and MOSFET methods to replicate the resistor. Chances are I'll do the same as the tach and just directly control the stepper (and also the low fuel light).


So let's just say neither works.
-Isaac

[mdrobnak]

Matt I am asking as I have converted a 2002 late year Boxster, this is a canbus driven gauge, Speed, Rpm and temperatures are directly from canbus. The fuel gauge is the hard one as It uses a saddlebag arrangement, sender till about 1/4 then it uses consumption of the engine to figure out the rest.

Let me preface all of this in that while I would love to help here more directly, I have a commercial product making use of this information as well, and so I don't want to directly shoot myself in the foot here.

I'll give some hints, and definitely tell you if you're correct, however.

If you can find a Funktionsrahmen for the ME7.1 you should be able to find what you're looking for.
I'm sure you've found RPM on 0x280.
Fuel Consumption is on 0x4e0. Look for something that looks like a counter. It isn't. That's all the hint I'm give on that one for the second.
The Kombi (Instrument cluster) data is being transmitted on 0x520 on the 996. Fuel level is in liters, and it's a single byte. It's a very simple calculation for this one.
The oil pressure input is linear, and it's fast responding, if you want to repurpose it for something else.

So let's just say neither works.
-Isaac

Sounds like a trial by fire! Because of the interplay between the control board / EEPROM for mileage storage, I'm sure that it's probably not easy to just swap control boards out...


-Matt

[johu]

RPM on 0x280,seen that before
https://openinverter.org/wiki/Touran_Conversion#CAN_bus

[mdrobnak]

RPM on 0x280,seen that before
https://openinverter.org/wiki/Touran_Conversion#CAN_bus

This makes sense given both are part of VW/Audi Group.
That particular piece of information is exactly the same.
Consumption is in a different place. It is an interesting way to express that particular value on that wiki page, too.
Edit: Good job on finding the 'visit workshop' bit

[Isaac96]

Sounds like a trial by fire! Because of the interplay between the control board / EEPROM for mileage storage, I'm sure that it's probably not easy to just swap control boards out...


-Matt

Yeah it requires a EEPROM reprogramming of some sort. We'll see how that goes. In the meantime I'm just controlling the tach with Arduino and L293, so CAN is actually possible.

-Isaac

[Isaac96]

Got home again yesterday, took the Boxster out for some tuning. Results:


Burnouts are now easy in first gear. Acceleration is strong in general. I'm getting some resonance at certain speeds but hopefully that will tune out nicely.

I was getting some pretty random overcurrent events, sometimes as soon as I turned the car on. However those sorta went away
So now it's time to do a little more tuning (dealing with boost and fslipmax mostly) and some range testing/high speed testing. Also need to put some air in the tires, it's been sitting for a while now and they're a little low.

Temperatures are a little high, need a bigger radiator for the inverter or maybe just some fans. I've been only driving around town (so under 30mph) and hopefully the higher speed will improve airflow.

A Raspberry Pi is going in pretty soon to integrate inverter configuration, BMS info, and charging info into a single webserver (probably just some Python and HTML).

More pictures and documentation will follow; I'll be updating my website with that once I get some more free time.

Really happy to be at this stage! Thanks to all!

-Isaac

[mdrobnak]

Congrats. Nice job!

Integration. Not as simple as some think.

-Matt

[Jack Bauer]

You shouldn't be getting excessive heating like that. I ran a siemens and dmoc for years in the E31 with a little tiny oil cooler rad. Check your not getting shoot through or slow switching speeds.

[Isaac96]

You shouldn't be getting excessive heating like that. I ran a siemens and dmoc for years in the E31 with a little tiny oil cooler rad. Check your not getting shoot through or slow switching speeds.

Possible... Although the Volt IGBT drivers should be sufficient. I have deadtime at 185. Would not be surprised if there are air bubbles somewhere, I'll be combining the 2 cooling circuits so that will be a chance to check and fix.

Congrats. Nice job!

Integration. Not as simple as some think.

-Matt

Agreed. Thus the Pi for multiple tasks, or at least ease of programming and display.

I did 30 miles today, pack voltage went from 398 (4.15 per cell) to 358 (3.7 per cell). That was hard driving though, half of it with a terrible tune.
Anyone know what the discharge profile is like for Pacifica modules? Safe discharge voltage? I've got HVC at 4.2, charger is set to 4.15 per cell.

-Isaac

[arber333]

Yes, congradulations!

I think Volt inverter has really low thermal mass. I got overtemp within minute of running it dry. If you have a good coolant flow however it should be fine. I use 185 deadtime with Volt inverter for almost 2 years now. I found out the highest temp events were slow drive and stops in the city.
I used a small 2 fan server cooler at first, but i added another small cooler in front. That was Mazda, in Pug i use a long and narrow cooler from one Fiat Uno car which is perfect because it has a small tank on the side through which i can release bubbbles.

What i did with Mazda and here. I put the car "head down" on an incline so that the complete system was definitely lower than the coolant tank. Then i filled it through the tank and i observed the bubbles and the sound my pump was making. To deair everythink i also squeezed the return hose repetitively because air wants to escape through water. By squeezing hose it is easier to keep back water and air wants to jump ahead of it.