openinverter forum

Setting up a thread to document the reverse engineering of the Model 3 rear drive unit. The specimen is from a European 2019 car with just over 20k km on the clock.

Initial observation is the motor exhibits cogging on initial turn but very little while turning via the drive shafts. The plan will be in two parts :

1)Remove inverter and run the motor with a prius inverter so we can learn how to control it with the Openinverter system.

2)Reverse engineering the Tesla inverter to allow use with an open source logic board.

Some useful background videos of the rear motor

From the inverter video description :
Here is the list of major components:
SiC FETs: ST GK026 (24)
Gate Drivers: STGAP1AS (6)
MCU: TI TMS320F28377DPTPQ
Resolver Amp: ON Semi TCA0372BDW
Temp Amp: TI LMV844
EEPROM: Microchip 25LC256E
CAN Transceiver: TI SN65HVD1040A (2)
LIN Transceiver: NXP TJA1021
HV Sense: Broadcom ACPL-C87BT-000E
DC-DC Power: Infineon TLF35584QVVS2
DC-DC Transformer: TDK VGT22EPC-222S6A12

As the Tesla wiring diagrams spread out the connector info over several pages I have put together the oem pinout for the rear du.

https://github.com/damienmaguire/Tesla- ... Pinout.pdf

Hi Damien,
can you do me a favour and make a raw drawing/video (L x W x H x D) of the dimensions of the 2 Model 3 DriveUnits.
Ony for a rough placement check.
Thank You very much!
Peter

JeanVapeur wrote: ↑Sun Mar 01, 2020 7:00 pm can you do me a favour and make a raw drawing/video (L x W x H x D) of the dimensions of the 2 Model 3 DriveUnits.

You might be interested in this crowdfunding for a 3D scan of both motors... would save Damien or myself from doing the work and provide a much more accurate result

viewtopic.php?f=10&t=184&p=7280#p7280

Been looking at the component specs and discussing with a very learned MC designer here. Given how integrated it is and that all the gear is about as good as you can find, is it worth trying to erase the micro and reprogram it? If that isn't possible then resolder another one onto the board. I believe the TI dev tools come with a vast array of motor control libraries which would make a lot of the initial setup fairly straightforward. I'm not a code person really so have no experience with that.

The SiC switching devices appear to be a custom package. They obviously set out to sell a lot! Also there will be no replacing them other than by replacing the whole module so test carefully.

57 Chevy wrote: ↑Mon Mar 02, 2020 8:33 am is it worth trying to erase the micro and reprogram it? If that isn't possible then resolder another one onto the board.

Lots options are being considered

Kevin Sharpe wrote: ↑Sun Mar 01, 2020 7:06 pm

JeanVapeur wrote: ↑Sun Mar 01, 2020 7:00 pm can you do me a favour and make a raw drawing/video (L x W x H x D) of the dimensions of the 2 Model 3 DriveUnits.

You might be interested in this crowdfunding for a 3D scan of both motors... would save Damien or myself from doing the work and provide a much more accurate result

3D scan will be nice to have, but as long as I dont have a CAD drawing or 3D scan of the underbody of my car, is this for me not realy helpful.
I try to keep it simple.
With the dimensions of the DU, I lie under the car with my pocket rule and can imagine if it fits or not.
Because actual are there exhaust pipes, differential housing and the fuel tank...

So the dimension of the hight and the distance from the driveshaft to the front edge, will help me further.

JeanVapeur wrote: ↑Mon Mar 02, 2020 3:46 pm 3D scan will be nice to have, but as long as I dont have a CAD drawing or 3D scan of the underbody of my car, is this for me not realy helpful.

With a 3D scan of the motor you can measure anything you want including;

JeanVapeur wrote: ↑Mon Mar 02, 2020 3:46 pm the dimension of the hight and the distance from the driveshaft to the front edge

Weigh in for the rear drive unit.

Awww yisss.

Every time a new project's motor spins, an angel gets his wings.

That first wiggle of movement where you pump your fist in the air, nod to your buddies, and say "Fuck yeah!"

So, first power up of the M3 rdu inverter on the bench. Elon flashed some nice coloured leds at me and pumped out massive amounts of CAN on TWO busses. I took a sample for posterity. Up on the github repo for the can hackers
https://github.com/damienmaguire/Tesla- ... Drive-Unit

Now we can start the real hacking. Phase one : eliminate the musk, eliminate the problem.

Pin 124 of the TMS main processor is the reset pin called XRS. Pulling that to ground puts Elon asleep and leaves everything else up and running. Easy pickings for the upcoming HubenAtor XM-3000 modchip:)

"time to get some sleep Elon... night, night"

https://twitter.com/EvBmw/status/123850 ... 38432?s=20

Hi Damian,
maybe this pictures helps you a little bit in your exploration...

Great to see the underside. Many thanks.

JeanVapeur wrote: ↑Sat Mar 14, 2020 6:50 pm maybe this pictures helps you a little bit in your exploration...

Do you have any photo's of the power stage?

Kevin Sharpe wrote: ↑Sun Mar 15, 2020 10:11 am

JeanVapeur wrote: ↑Sat Mar 14, 2020 6:50 pm maybe this pictures helps you a little bit in your exploration...

Do you have any photo's of the power stage?

Some useful info in this article;

ABOUT THE SIC MOSFETS MODULES IN TESLA MODEL 3

Thats all I found, possible this can help...

Anyone find a high res shot of the rear of the board? I'm getting a feeling old Elon is using bootstrap gate drivers:)

Yep. D421,D422and D423. High side bootstrap diodes:)

So a little more investigation reveals a Viper16A based power supply section used to generate the low side iso voltages. Not 100% sure as yet but looks to be derived from the hv. Can't say I like this. They're building cars not cheap mobile phone chargers. A multi output dcdc wouldn't have put them out of more than a few $...