Porsche Boxster with Siemens and DMOC

[Isaac96]

Really interested to hear how you get on mounting the ibooster. I'm keen to go that direction too but haven't really assessed how well it'll physically fit in the frunk.

I have two annoying things on my 986. One is the power steering pump noise and the other is the vacuum pump kicking in. Hoping to improve both this summer.

I'll be sure to let you know if I ever make progress on that front... there's quite a bit to do on the rest of the vehicle.
Agreed, vacuum pumps sound disgusting; I've got a RAV4 EV on hand at the moment and even Toyota couldn't make it quiet. Though the power steering (Astra pump methinks) sounds pretty cool...

2 week update time. My dad and I spent the last week building mounts; front, rear, and side.
Here they are installed on the motor:

And then today we lifted it into the car. After an hour or so of shoving it around and forcing bolts, it's finally mounted in place permanently (I hope).

Axles are bolted up and fitting properly, HV wiring is complete, and the LV wiring is easily completed. But the batteries need some support, a chassis part needs to be replicated (interfered with motor), and coolant hoses connected. After that is done, we'll be on the road again.

Things are moving along though! It's getting exciting!

-Isaac

[Isaac96]

Axles are bolted up and fitting properly, HV wiring is complete, and the LV wiring is easily completed. But the batteries need some support, a chassis part needs to be replicated (interfered with motor), and coolant hoses connected. After that is done, we'll be on the road again.

Added battery support, replaced chassis part (reinforcement plate support), and ran coolant lines.

Went for a drive.
WOW!
I spent the first couple of minutes using the first 10% of the pedal and it already felt wayyy faster than the Siemens system.
Then I floored it -- a burst of power and then overcurrent.
I took it up to 25, then put the pedal mostly down, and it went like stink. Absolutely phenomenal!

It's also been less than 2 months since the Tesla unit arrived. Not bad for a complete drivetrain change!

A few driveability issues I noted...
A lot of oscillation at lower pedal settings at low speeds; that being between about 5 and 15 mph. More pedal smoothed it out, as did higher speeds. But it's quite difficult to drive smoothly at those low speeds, which will be nasty in traffic.

I am using the parameter set which came preloaded, I believe it's from the Porsche 944 with SDU. Has anyone else dealt with this sort of oscillation? It could just be interference in the throttle pedal lines, or even feedback from acceleration pushing me back in the seat. Maybe I can set up CAN throttle to fix that.

Anyways, I'm super stoked about the performance! With a little more work this car will be an absolute monster. I'm afraid to think about what it would be like with an LDU. (To be fair, I usually drive a minivan...)

-Isaac

[Isaac96]

A little more driving gave some interesting results.
First, I need to set tripmode to autoresume because that's helpful, had a couple of OC events.

Second: the situations in which vibration occurs are now much more defined.
Beginning from a standstill, vibration begins to appear as the car reaches ~10mph. It very quickly increases, and positively throws the car around, not fun to drive; I always let off the gas at that point so I don't know what could happen.
If one accelerates more quickly, the vibrations don't appear to have time to 'catch' and form some sort of feedback loop, so they are minimal.
And if decelerating, if I get on the gas at 14 or even at 12mph there is no vibration.
It seems to be a bad feedback loop which hits at ~10mph.
Maybe changing fslipmin will help...

[arber333]

A few driveability issues I noted...
A lot of oscillation at lower pedal settings at low speeds; that being between about 5 and 15 mph. More pedal smoothed it out, as did higher speeds. But it's quite difficult to drive smoothly at those low speeds, which will be nasty in traffic.

I am using the parameter set which came preloaded, I believe it's from the Porsche 944 with SDU. Has anyone else dealt with this sort of oscillation? It could just be interference in the throttle pedal lines, or even feedback from acceleration pushing me back in the seat. Maybe I can set up CAN throttle to fix that.

Anyways, I'm super stoked about the performance! With a little more work this car will be an absolute monster. I'm afraid to think about what it would be like with an LDU. (To be fair, I usually drive a minivan...)

-Isaac

Yes i had the same problem in my Mazda and my high torque ACIM motor. Motor delivers more than 300Nm and for that it requires high slip at start. 3.0Hz for minslip and 21pts for ampmin to driveoff smoothly.
Now if you start from 0 and accelerate it is really good, but if you need to reduce speed quickly and then add some poweer to keep low speed in traffic vibrations will quickly progredd to when you need to use the brake and regen to keep the motor steady.

I think this is because Johannes uses only P control in ACIM code and vehicles mass is trying to dampen the torque control, which is the job of I and D part of the control loop. Since D part of control is most certainly absent you are not damping those small mass/torque excursons which happen at low speeds where torque is higher than inertia. It wrecked the hell out of my clutch... 3 times! Now i use lovejoy coupler witht the hardest rubber available.

There was an old code with some hypebolic math when J experimented with damping and stuff. It worked for me a lot better but it also had some different quirks. I dont know if it will work for tesla, but i can find it if you want to try...

[johu]

There are no control loops in slip control, except for derating and cruise control. Throttle command is just translated to slip and amplitude. Of course there are still feedback loops happening: the 36 pulse encoder.
What happens if you decrease fslipmin (by 0.1 Hz decrements) and/or boost (by 50-100 dig decrements)? Or increase fweak (by 5 Hz increments)?

[Isaac96]

I will bring more information later today, in the meantime I'll take the car on the highway to school, hope it works.
Will be sure to test slight changes in the settings; fslipmin of 2 seems pretty darn high....

[arber333]

I will bring more information later today, in the meantime I'll take the car on the highway to school, hope it works.
Will be sure to test slight changes in the settings; fslipmin of 2 seems pretty darn high....

Like i said i used fslipmin 3hz and motor feel was quite soft through most of the rpm. Only at keeping throttle pedal around 0 position at slow speed i had issues with torque jumping.

[mdrobnak]

Congrats on getting it running. I wonder if it is an issue with the throttle differences should be less than linear at low throttle pedal input? (Like driver torque demand in most OEM ICE cars)

-Matt

[Isaac96]

Like i said i used fslipmin 3hz and motor feel was quite soft through most of the rpm. Only at keeping throttle pedal around 0 position at slow speed i had issues with torque jumping.

I will certainly try that. Another interesting factoid - Damien's parameters have fslipmin of 1 and fslipmax of 3.5, interesting! Looks like there's a good range of usable parameters.

Congrats on getting it running. I wonder if it is an issue with the throttle differences should be less than linear at low throttle pedal input? (Like driver torque demand in most OEM ICE cars)

-Matt

Thanks!
Hmm, that could be happening, maybe an offset cam on the potentiometer or something. But I feel like they'd just use a software mapping for that, easier to manufacture and then to modify.

I'll be doing testing later today, just hooking up a Pi to the CAN bus so I can get info from the inverter, BMS, DCDC, and ISA. Which reminds me I need to set up all the CAN mappings again... oh dear.
On the plus side I can gather data on how well idc and reported ISA current line up, that will be fun!
And I got one overcurrent event when I floored it, that seems to keep happening, I'll just turn fslipmax down a bit (or possibly raise fweak a tad). Very noticeable torque knee at ~52mph.

It's awesome to finally be on the road again! The car feels very solid otherwise and all systems are working as they were. It was warm today, about 75 degrees, and the motor hit 52 while the inverter hit 34. This was after a 16 mile drive at 65-70. So that's certainly working, there is room for improvement too.

-Isaac

[Isaac96]

Okay, I did a lot of tuning today (just driving around the block over and over again).
My starting parameters were the defaults from Johannes.
boost 2200
fslipmin 2
fslipmax 6
fweak 220
udcnom 360

And the starting situation: Strong oscillations at 7-12 mph which could be avoided by powering through that area quickly, not giving them time to build. Overcurrents with full throttle at anything below 30mph. Huge power at speed (I checked the ISA when I hit 80mph, it said 270KW!!!!!)
I didn't try full throttle from a standing start, but that would probably have resulted in OC as well.

I spent some time playing with fslipmin; values from 1.8 to 2.2 had no discernible effect on the jitter.
fweak and boost were more promising. Increased fweak seemed to decrease jitter and allow full throttle without OC, also appeared to move the jitter to a higher RPM (but that might be pure placebo).
Increased boost smoothed the jitter still more and allowed full throttle starts - which are very jumpy for half a second or so and then move to smooth power (half throttle for starting works nicely though).

So now I've got boost at 2500 and fweak at 250 with udcnom at 370. The results are excellent smoothness and performance up to field weakening zones; I can induce the old jitter if I try to but it's rare, and full throttle starts still jump but smooth out quickly. No burnouts yet
However, the transition from V/F to field weakening causes overcurrents about half the time. I do have fconst set up, fconst is 600 and fslipconstmax is 6 (matching fslipmax).

I'm feeling great about how this system is shaping up! With another couple of hours of work I feel confident that this system will be perfect; it's already great.

-Isaac

[johu]

Sounds good. I think I will replace the current parameter defaults by yours as soon as you're happy with them.

Overcurrent at fweak is weird, never saw that before. Not sure how to cure.

[Isaac96]

Sounds good. I think I will replace the current parameter defaults by yours as soon as you're happy with them.

Overcurrent at fweak is weird, never saw that before. Not sure how to cure.

Sounds like a plan.
I'm really not sure if it's specifically at fweak, that just seems to be the general area. I may try playing with fslipconstmax and fslipmax and see if it calms down a bit, might also raise fweak a bit more.

[Isaac96]

I got a new inverter, car is back on the road. For those who missed it, I managed to destroy a SDU inverter by a large number of desat events.

Also I can confirm that a FDU inverter will fit in a SRDU. However, wire extensions will be needed. (About 2" total). I made them out of 3/4 * 1/16 copper bar, doubled up to 1/8 * 3/4. This is the same area as the factory cables.

I dropped fslipmax from 6 to 5, will be testing that change in the next week or so.

[arber333]

I got a new inverter, car is back on the road. For those who missed it, I managed to destroy a SDU inverter by a large number of desat events.

Also I can confirm that a FDU inverter will fit in a SRDU. However, wire extensions will be needed. (About 2" total). I made them out of 3/4 * 1/16 copper bar, doubled up to 1/8 * 3/4. This is the same area as the factory cables.

I dropped fslipmax from 6 to 5, will be testing that change in the next week or so.

That is good news for SDU users!
Is there any reason you could not try it in reverse? Putting RDU module into FDU inverter?

[Isaac96]

Is there any reason you could not try it in reverse? Putting RDU module into FDU inverter?

That should work about the same. The wires are too long, need to shorten them. At the inverter end they are ultrasonic welded or whatever straight to the IGBT bus bar so you can't just swap, and there's not a ton of space to get a crimper in there. Maybe a sharp S-bend?