Porsche Boxster with Siemens and DMOC

[Boxster EV]

I'm not familair with your cooling configuration, but just a note to say that I had a nightmare bleeding the system on my Boxster (although it is rather complex with three battery banks, drive unit and charger). Every time I thought the air was gone I was wrong. Went on for about three weeks. All good now

[Isaac96]

I'm not familiar with your cooling configuration, but just a note to say that I had a nightmare bleeding the system on my Boxster (although it is rather complex with three battery banks, drive unit and charger). Every time I thought the air was gone I was wrong. Went on for about three weeks. All good now

If you're using the original radiators I can imagine how bad that would be....

I've got a pair of small radiators (Boxster center and heater core) mounted on the back of the transaxle. Motor and inverter on separate loops.

I'm rather sure the airflow is too low, so it's time for some fans. Air is a huge pain in this system, since the hoses are rather short and there isn't much slack to move components.

-Isaac

[BLWNHR]

need a bigger radiator for the inverter or maybe just some fans.

I'm rather sure the airflow is too low, so it's time for some fans.

I would look at larger radiators and improving airflow. Fans = current draw = reduced range.

[Boxster EV]

I'm not familiar with your cooling configuration, but just a note to say that I had a nightmare bleeding the system on my Boxster (although it is rather complex with three battery banks, drive unit and charger). Every time I thought the air was gone I was wrong. Went on for about three weeks. All good now

If you're using the original radiators I can imagine how bad that would be....

I've got a pair of small radiators (Boxster center and heater core) mounted on the back of the transaxle. Motor and inverter on separate loops.

I'm rather sure the airflow is too low, so it's time for some fans. Air is a huge pain in this system, since the hoses are rather short and there isn't much slack to move components.

-Isaac

I'm using just one of the original radiators in the front which seems more than sufficient (inverter temp never really gets much above 30 degrees). I suspect you'll struggle to get decent air flow in the back - perhaps some air ducts could be channeled from the side vents...?

[johu]

I would look at larger radiators and improving airflow. Fans = current draw = reduced range.

Not sure if it's that simple. A large fan consumes maybe 50-100W. Aerodynamic drag from a larger cooler will reduce range more in my opinion. That said I would also shy away from fans for the extra complexity. Maybe if you want to compete on Pikes Peak but for regular street driving I don't see the need for that much cooling.

[Isaac96]

I'm using just one of the original radiators in the front which seems more than sufficient (inverter temp never really gets much above 30 degrees). I suspect you'll struggle to get decent air flow in the back - perhaps some air ducts could be channeled from the side vents...?

Unfortunately the batteries are all in front of the side vents (very very tight fit) so those are useless. I'll probably make some ductwork under the rear end which will direct the airflow better.

In other news I drove 88 miles yesterday (mix of highway and city, with hard acceleration) and used 30.8KWh. So about 350 Wh/mile. I'm pretty pleased with that efficiency

I'm getting rather random OC events which are not accompanied by any current draw - generally occurs when releasing the brake pedal, almost as if high regen gets commanded? I'll look more into it. Regen is currently disabled but I will add some and see what happens.

I threw together some HTML and Python and now have a Pi running a webserver which displays lots of information from inverter, DCDC, and BMS. Update speeds are a little slow but that can be improved.


-Isaac

[Isaac96]

Went for another drive, temps hit 70 (motor) and 50 (inverter). This was around the hills at pretty high speeds, so just about maximum load.

I'm using the clutch pedal for shifting since regen is enabled (clutch disables forward input). But the 'dirchrpm' parameter will not go above 2000.

Johannes, is there any chance you could change the maximum for dirchrpm? Maybe up to 10000?

Thanks!
-Isaac

[arber333]

I think OCs are occuring because of encoder filter issues. Can you tell which firmware you are using?
I had similar issues, but now since i use latest ACIM code with filter behind inputs i dont have any issues at all anymore, Mazda runs great and i am at 150000Km electric!

[Isaac96]

I think OCs are occuring because of encoder filter issues. Can you tell which firmware you are using?
I had similar issues, but now since i use latest ACIM code with filter behind inputs i dont have any issues at all anymore, Mazda runs great and i am at 150000Km electric!

I hope that's the problem! I'm on 4.76 (sine of course). Only at 200 miles so far, maybe I will catch up someday!

-Isaac

[johu]

Johannes, is there any chance you could change the maximum for dirchrpm? Maybe up to 10000?

I see what you're doing. Yes I can change that.

[Isaac96]

Johannes, is there any chance you could change the maximum for dirchrpm? Maybe up to 10000?

I see what you're doing. Yes I can change that.

Awesome, thank you!

-Isaac

[Isaac96]

Went for another drive with my new gauges. They are displayed on a cheap Fire tablet, from a WiFi access point and CAN decoding running on a Raspberry Pi.
The general text information:

And the nicely formatted gauges:

I've got code of course if anyone wants it. This can be expanded and modified fairly easily if you have any knowledge of HTML and Python (probably even without it).

-Isaac

[Isaac96]

Been a while since an update...

The drive system is still giving me headaches. I'm getting fairly random overcurrents, often when easing off the gas and often when starting in 3rd gear (even happened in 2nd). And there are nasty noises coming from the motor coupler -- chances are that 200SX clutch plate isn't the best fit, so I may look for something better.

I'm still suspecting encoder issues. Since the motor is coming out soon I will disassemble the rear end and look at the encoder itself, also will replace the wiring with shielded twisted pair 'CAT7' which I already ordered.

The coolant bottles were leaking and cooling was overall pretty wimpy, so I did a few things:
-Welded up a very custom stainless steel coolant tank - it doesn't leak yet
-Relocated radiator to the original front position where airflow is better
-Rerouted a hose to get air out of the motor
So that should improve the thermal situation -- I was getting high motor temps after only 2 or 3 hard pulls, hopefully that will change.

I was driving on the highway recently and the motor suddenly shut off and the gauges began displaying 300C - turned out one of the IDC connections wasn't great, so poking that wire got it working again. I checked over all the others and they seem fine.

Now for some good news:
90 mile drive went well, I was scared of overheating but it ended up working fine when ambient temperatures dropped below 100F. Put 31Kwh back into the pack -- these Pacifica batteries are great!
BMS works very well (surprisingly). Gauge system is very good, Pi is going to take on a more central role once I get back home (controlling DC-DC, handling all CAN interface duties).

And I got a free Prius inverter! I was at Thunderstruck working on some gauge systems for Curtis controllers, and was offered a Gen2 Prius system that was just sitting around. I've already torn it down to what I need and will be running the lightbulb tests (am without STM32 right now, but can at least test the power stage functionality). This may go in my car for a few reasons:

I actually suspect the long deadtime (3.1us) of the Volt inverter of causing some of the oscillations I am seeing; apparently deadtime can have a significant effect on harmonics and torque oscillations, which would probably explain some of what I am hearing and feeling.
Damien is having good results with a Gen3 and the same Siemens, it may simply be my incompetence but I'd like to try this system.

Does anyone know what deadtime the Gen2 inverter incorporates? It is the unipolar system. I suppose I can scope the gate driver outputs.

The Gen2 can handle significant current according to Damien and the wiki; 350A for MG2 and 250A for MG1. These are peak values, so RMS currents will be quite a bit lower. Which leads to the next question:
Is it safe to parallel MG1 and MG2?
And if I do that, do I even need current sensors? I'm running an induction motor so there's no need for FOC, and the toyota system can shut itself down easily. I'll just keep pushing more slip til it trips, then back down a notch or 3 -- and do this for each gear separately, probably with a Volk-style CAN parameter modifier. Software time again!

I'm also thinking about installing a small STM32 board in the inverter and having a separate one control it via CAN - dealing with all the auxiliary I/O. That's just growing out of laziness though, or maybe I like CAN too much. The board in the inverter would only deal with the 3 outputs, the encoder inputs, and a couple of temperature sensors - nice and simple unless I'm forgetting something.

I'd really appreciate any input on these weird ideas; especially concerned with my motor coupler, anyone have a good system for dealing with Siemens motors?

Thanks!
-Isaac

[Jack Bauer]

The Gen2 inverter generates its own deadtime to set it to zero. If using a Gen3 set to 130. My Gen3 and siemens are getting on very well together. I might add in the MG1 powerstage (running mg2 now) as I can get it to trip out if I get really agressive. MG2 trips at a little over 420A. Love the gauge display.

[arber333]

Hi

My Volt inverter in Peugeot running with Leaf motor has no problems with 3us. It seems that only ACIM motors come under some resonance with longer deadtime. I use my DIY inverter in Mazda with 3.1us deadtime and if i used something less i would get a desat or a kind of voltage spike event on takeoff. This applies only for the first couple of RPM when motor seems like it would want to stretch rubber. Then when it is rotating there is no other problem besides resonance at lower RPM when transitioning from regen to drive and VS.

You can safely try with 2.8us or 2.6us. I think you are still safe there. I choose 3uS because we toasted one Volt inverter just by running it under 1.8us. When i went and recalculated all the timing from data available to me i got 2.5us and i like to use some overhead.

[mdrobnak]

I will also be driving a PMSM motor, so I hope the Volt inverter works well for me.

If you're not looking for 900A of current, I think the Prius inverter is potentially more bulletproof. As far as I know you can parallel the outputs to get a respectable amount, and the advantage of it is allowing for higher voltage, although that's not a factor in your case.

I think people underappreciate the amount of magic going on behind the scenes with respect to throttle control in your standard vehicle - the throttle pedal to output is anything but linear. The result of this is, if your pedal -> torque mapping is linear in the OpenInverter side of things, and its also straight up linear in any VCUs in front, then the car is going to be difficult to control under low pedal position scenarios especially.

Definitely a fan of keeping sensitive analog signals to the minimum...

How do you have the batteries wired up? I remember reading you're supposed to have diodes to prevent eddy currents, but I've yet to see a practical implementation of this. How deeply did you discharge the batteries to achieve your 90 mile drive?

Glad to see your car is coming along well!

-Matt

[Isaac96]

The Gen2 inverter generates its own deadtime to set it to zero. If using a Gen3 set to 130. My Gen3 and siemens are getting on very well together. I might add in the MG1 powerstage (running mg2 now) as I can get it to trip out if I get really agressive. MG2 trips at a little over 420A. Love the gauge display.

Thanks! Yes, I've been reading through the threads and learned about the internal deadtime. Just wondering if anyone had scoped the internal signals to see what the actual lag was.
420A sounds great, I remember in your video you couldn't get MG2 to trip even with that giant battery! I'll start with one stage and see how it feels.

Hi

My Volt inverter in Peugeot running with Leaf motor has no problems with 3us. It seems that only ACIM motors come under some resonance with longer deadtime. I use my DIY inverter in Mazda with 3.1us deadtime and if i used something less i would get a desat or a kind of voltage spike event on takeoff. This applies only for the first couple of RPM when motor seems like it would want to stretch rubber. Then when it is rotating there is no other problem besides resonance at lower RPM when transitioning from regen to drive and VS.

You can safely try with 2.8us or 2.6us. I think you are still safe there. I choose 3uS because we toasted one Volt inverter just by running it under 1.8us. When i went and recalculated all the timing from data available to me i got 2.5us and i like to use some overhead.

I saw the same - it was only induction systems that had issues.
Specifically, what I am seeing is huge oscillations when rolling at a couple MPH for parking, this moves the entire car -- and also a nasty harmonic at 1400 RPM which shakes the car up as well.
I will drop deadtime a little and see if it performs better - I think I remember dropping it too much and the thing heating up a bunch, same happened when I set PWMFRQ=17.6 (twice the heat generation). If it dies it dies, there's always the other half of the inverter

I will also be driving a PMSM motor, so I hope the Volt inverter works well for me.

If you're not looking for 900A of current, I think the Prius inverter is potentially more bulletproof. As far as I know you can parallel the outputs to get a respectable amount, and the advantage of it is allowing for higher voltage, although that's not a factor in your case.

I think people underappreciate the amount of magic going on behind the scenes with respect to throttle control in your standard vehicle - the throttle pedal to output is anything but linear. The result of this is, if your pedal -> torque mapping is linear in the OpenInverter side of things, and its also straight up linear in any VCUs in front, then the car is going to be difficult to control under low pedal position scenarios especially.

Definitely a fan of keeping sensitive analog signals to the minimum...

How do you have the batteries wired up? I remember reading you're supposed to have diodes to prevent eddy currents, but I've yet to see a practical implementation of this. How deeply did you discharge the batteries to achieve your 90 mile drive?

Glad to see your car is coming along well!

-Matt

I do like the self-protective instincts of the Toyota system, it seems like it'll be hard for me to kill it. 900A would be nice but that would just break the transmission.

Also wouldn't be surprised if there is some interference going on with the pedal readings - I'll plot one of those next time I do testing. I guess I could throw an Arduino up front and translate the signal to CAN as soon as possible, it is about a 8ft cable run for those signals.

My batteries are paralleled on a module level -- every 16 cells there's a fat busbar (okay only 1/16 by 1") and I haven't seen any issues, it's possible that it could cause problems in the long term but the cells are still at 9mv differences overall. I think the battery went as low as 330V, so 3.43 per cell, but it still pushed plenty of power even that low. I would say 3.35 is 0% SOC when at rest, I have amp-hour counting now so no more relying on voltage and mileage readings.

Thanks so much for all the advice! I will have lots of things to check out when I get back to the car.

-Isaac

[arber333]

Hm... Regarding general motor jumpiness. Johannes solved that for me with a parameter that seamlessly adjusts fweak and boost range VS the battery voltage.
To tell it plain... My motor was wound for 185Vrms. I use it with 380Vdc which is clearly much and motor had really sharp response.
With this mod I would set some 320Vrms into parameter and new fweak and boost were calculated. Motor sees less immediate voltage.
My motor ran much smoother and it developed more power despite voltage drop at empty. Parameter is called udcnom.
Maybe your motor was wound for lower voltage and has too sharp response to throttle change.

Fconst parameter is also used to limit constant power region if your cells can't handle motor power.

[Isaac96]

I will try setting it lower - the motor's nameplate voltage is 215V, there are test graphs for 300V, and I'm running at 360. So I'll take a look at what a lower udcnom does.
Thanks!
-Isaac

[Isaac96]

It's been a while since an update, I've made some progress so this is a good time.

Today I got the new motor-transmission coupler - it's an EVWest interference fit on the motor side, and a Porsche 996 clutch plate for the transmission. I've been working at Thunderstruck so they built it up for me -- about 2/1000" runout. Hopefully this will solve all the noise issues.

A few weeks ago I made a new encoder cable with some shielded CAT5; it's able to spin the motor normally, we'll see how it runs on the road.

And one more specific thing I've been working on. I have rather different behavior in each gear as it stands with a static parameter setup. So I've written some software and built some hardware that will let me set different parameters for each gear, using hall sensors on the stick and CAN straight to the inverter. At this point I'm only setting boost, fweak, fslipmin, and fslipmax. It's working so far, more testing will occur once I get the drivetrain back together.

CHAdeMO testing is a priority as soon as the car is rolling again.

-Isaac

[mdrobnak]

Very clever.

You can say you have a "boost per gear setup" to hardcore gas guys now.

Nice progress.

-Matt

[Isaac96]

You can say you have a "boost per gear setup" to hardcore gas guys now.

Hmm, yes, I have heard of those. Now all I have to do is go brag about my 160Kw (or so) to the pavement and I'll have plenty of gearhead cred.

[Boxster EV]

Good work. As a fellow Porsche guy, it would be good to see some pictures of the build if you get a chance.

[Isaac96]

Pictures, pictures, pictures. I always try but then they don't end up being uploaded - plus working alone it's difficult. I've got some sitting around that I can upload.

The last few days have been spent installing the new motor coupler. It is now done and the driveshafts are back in - a massive improvement in noise.
The new encoder cable seems to be improving no-load behavior as well, no more random overcurrents yet.

My 'boost-per-gear' is going in later today once I find the right glue.

And today I got a new firmware file for my TSM2500 charger and upgraded it to 500Kbaud CAN; now all the CAN devices are at the same speed on the same bus, no more dead 12V battery!

One issue is presenting itself so far.
With the car in run mode and foot off gas accelerator, some power is constantly being applied - ampnom is 162 and a whining noise and heat present themselves.
Potnom reads zero, both throttle pots are calibrated. And I believe all my regen settings are correct.
I think I have a modified version of 4.76, or else it's 4.87 -- version is '4.00' in the settings.

Anyone have an idea? Should I upgrade to 4.96 before doing anything else?

Thanks!
-Isaac

Code: Select all

boost           3300.00
fweak           150.00
fconst          150.00
udcnom          390.00
fslipmin                1.68
fslipmax                3.50
fslipconstmax           3.50
polepairs               2.00
respolepairs            1.00
encmode         1.00
fmax            250.00
numimp          64.00
dirchrpm                6000.00
dirmode         1.00
snsm            13.00
pwmfrq          1.00
pwmpol          1.00
deadtime                155.00
ocurlim         800.00
minpulse                1000.00
il1gain         2.37
il2gain         2.37
udcgain         1.90
udcofs          149.00
udclim          1000.00
snshs           2.00
bmslimhigh              50.00
bmslimlow               -10.00
udcmin          280.00
udcmax          500.00
iacmax          800.00
idcmax          5000.00
idcmin          -5000.00
tmphsmax                85.00
tmpmmax         150.00
throtmax                100.00
throtmin                -100.00
ifltrise                10.00
ifltfall                3.00
chargemode              0.00
chargecur               0.00
chargekp                80.00
chargeki                10.00
chargeflt               8.00
chargemax               90.00
potmin          475.00
potmax          1730.00
pot2min         475.00
pot2max         1790.00
potmode         1.00
throtramp               10.00
throtramprpm            20000.00
ampmin          5.00
slipstart               50.00
brknompedal             0.00
regenramp               100.00
brknom          0.00
brkmax          0.00
brkcruise               -30.00
brkrampstr              10.00
brkhistr                1000.00
brkhistp                1000.00
brkout          -50.00
idlespeed               -100.00
idlethrotlim            50.00
idlemode                1.00
speedkp         0.25
speedflt                5.00
cruisemode              0.00
udcsw           300.00
udcswbuck               540.00
tripmode                0.00
pwmfunc         0.00
pwmgain         100.00
pwmofs          0.00
canspeed                1.00
canperiod               0.00
fslipspnt               1.21
ampnom          4.53
version         4.00
hwver           1.00
opmode          1.00
lasterr         12.00
status          0.00
udc             380.84
idc             0.00
il1             -1.25
il2             -0.40
ilmax           0.81
uac             1.12
il1rms          2.96
il2rms          2.68
boostcalc               3300.00
fweakcalc               150.00
fstat           1.21
speed           0.00
cruisespeed             -1.00
turns           0.00
amp             162.00
angle           265.56
pot             455.00
pot2            450.00
potnom          0.00
dir             1.00
tmphs           43.75
tmpm            34.68
uaux            13.81
pwmio           59136.00
canio           0.00
din_cruise              0.00
din_start               0.00
din_brake               0.00
din_mprot               1.00
din_forward             1.00
din_reverse             0.00
din_emcystop            1.00
din_ocur                1.00
din_desat               1.00
din_bms         0.00
cpuload         24.78

[Isaac96]

Upgraded firmware to 4.96, filled coolant system, and went for a drive.
Amazing. Quiet, smooth, powerful, cool. I thought a wheel bearing was bad but I had just forgotten to tighten the wheel lugs

A little more tuning, still need to mount those pesky gear position sensors, and the car should be perfect.
(getting OC events when hitting ~4500rpm - that's the fweak setting, where should fconst be set? Equal to fweak?)
And I'll set tripmode to autoresume so I don't have to pull the key out at 60mph on the freeway.

Temperatures stayed very nice. Ambient was 15C which helped, but after 20 miles at 60-70 inverter only hit 36 and motor read 65 (that sensor is suspect, reading too high).
This is with the single front center radiator mounted in its original spot, with 3/4 hose and a 25l/m pump.

I'm waiting on a full charge to see efficiency numbers. I can't remember what my original 0% voltage was, so I'll have to run it to zero again (or use the ISA for Ah counting!)

CHAdeMO soon - maybe later this week. I'm not sure where to mount it yet, but tossing it in the trunk should be fine for now.

I'll be adding some regen to improve shifting behavior as well.

Overall, I'm very happy with how the car is driving. A little more tuning and I'd give it to a new driver.

Pictures will come soon! They're all up on my website right now (electricboxster.com) but not exactly curated and described very well.

-Isaac