Tesla Model S drivetrain hacked!

It is done. The Tesla motor is happily spinning next to me, as I write this text. You can see the proof here:

This is actually quite an achievement for me, as this is the most advanced electric car drivetrain in production. And now, anyone can grab a wrecked Tesla Model S, Mercedes B electric, or Toyota RAV4 electric, as they share the same motor, and reuse these motors in any car. Another interesting thing is that they all have different ratings on a paper – Tesla Model S is obviously more powerful than Mercedes B, but it all boils down to the same drivetrain inside – all that changes is a firmware. Unfortunately the CAN protocol changes, so the interface is not the same.


The original Tesla motor controller

But this doesn’t really matter when you replace the motor controller from Tesla with a new one, where you have the full control of all the parameters. You can even put two such motors in one car, and they will not collide with each other. Do you want to squeeze all the possible power from your drivetrain, and put it in some kick-ass car? With the UMC Drive motor controller you can πŸ™‚

UMC Drive with a Tesla inverter

Short summary of the UMC Drive:
– Possibility to drive any three phase motor (induction, BLDC, weird hybrid designs etc)
– Both sensor as well as sensorless capabilities
– Drive by wire dual output accelerator support
– CAN interface
– Isolated USB interface
– Very robust input protection
– Compatible with all Toyota / Lexus stuff, Chevrolet Volt inverters, Tesla inverters and several others
– Self-learning of new motors (you can use this with motors I have never owned)

UMC Drive 2.0 assembled and ready for testing

When everything started working, and I could spin any motor here at my place with Tesla inverter, it was time to get somehow that huge Tesla motor in my tiny apartment. Not an easy task! But thanks to a friend of mine, we managed.

2015-06-09 18.28.21

Loaded up and ready to go

2015-06-12 00.23.52

And it’s there!


This entry was posted in Electronics, Tesla hacking. Bookmark the permalink.

30 Responses to Tesla Model S drivetrain hacked!

  1. Magnus says:

    Congratulations Michal, well done!

  2. Memed says:

    WELL DONE! Big congrats Michal! It’s the beginning of a new era!

  3. felix says:

    Awesome work!

  4. kennybobby says:

    Most Excellent Success, Michal, you have persevered thru a difficult task and reached your goal. You make all of us who have reverse-engineered impossible hardware proud, and we stand in awe at your skill and speed to sort it out and solve the problems.

    • Akash says:

      But how much will it cost? I’m worried that these cars will be so udoaffrnable, and the infrastructure so lacking, that nobody will buy them and industry will develop a conventional notion that electric cars are a bad investment. What is being done behind the scenes to make sure this doesn’t happen?

  5. ernes says:

    Great work!
    I hope you will publish more about your UMC Drive motor controller board.
    It’s definitely a EV hacking time.

  6. dohnalik says:

    Very interesting,
    will you realese some detailed photos of PCBs, IGBTs etc. ? Im very interested. Also, what kind of procesor are you using in UMC Drive?
    Thank you!

  7. Great work!!! Looking forward to hearing more… Keep it coming

  8. Paul Guzyk says:

    Thanks for the education! When will your controller be available for purchase and do you think it will be ok for driving a Highlander Hybrid REAR motor (Q211) ? I have a small Electravan (similar to http://www.evalbum.com/662) that I’d like to convert to Lithium batteries with a Highlander Q211. Q211’s are very affordable, only around $300 or so at US junkyards.

  9. Cosmacelf says:

    You are right, the Tesla Model S does not have a parking pawl – it just has an electric parking brake. However when Tesla did the drivetrain for the Toyota RAV4, the Toyota engineers were appalled that the drivetrain did not have a pawl and made Tesla add one on. Presumably Mercedes thought the same thing.

  10. Joep says:

    Very nice to see people do this! I plan to build my own electric car as soon as the price of the batteries go down after Musk’s gigafactory is done. I’m curious, how much did you pay for that motor? It can’t be cheap..

  11. c704710 says:

    So the P85D has two motors. The front motor is smaller and not as powerful. Presumably the same engine motor up front would be way to inefficient for the performance it could provide plus the added weight for that tiny bit of performance. ICE all-wheel drive cars are designed to deliver less power to the front as well, as little as 25%. In an ICE car, 50/50 is very bad for efficiency as the power must go through an efficiency destroying transfer case, unlike the back wheels which are more directly connected. Putting 50% into the front wheels robs power leaving the whole system with less. Even the Ferrari FF, which doesn’t use a transfer case and can therefore efficiently deliver 50% power to the front wheels, doesn’t. So I’m guessing there is very little performance advantage for having equal power up front (on roads at least). But, if one was only concerned about takeoff speed and having just a bit more power to overcome wind and other resistances that take away a negligible amount of top speed, could you replace the front motor of P85D with a rear one and use the UMC to translate this added power to maximum speed and acceleration?

    • Michal says:

      Hi, in theory you could do that. But I’m afraid that it wouldn’t help you in any way – the P85D is already limited by how much peak power the battery can deliver. Motors might be able to give even more than that, but you would need a bigger battery to push more power from the car. Also there is a limitation how much power you can actually transfer on the road (tire slip).

  12. Michael says:

    Congrats! What a cool idea to replace the signal generation and make use of the original power electronics! Looks like you’re about to enable a whole new tuning world for EVs πŸ™‚

  13. Ted says:

    What an achievement. If I’d live near to you, I’d gladly offer to do a frame, suspension and everything for such a “kit” car. In my current project I’m using petrol engine, but my real dream is to go electric one day. I will follow you, and I’m expecting very much from your work!

    How much ~~ does the whole powertrain weight?

    Best of luck.

    • Michal says:

      Hey, thanks!
      The whole drivetrain (gearbox, diff, motor, inverter) weights 135kg. Not bad! πŸ™‚

      • Ted says:

        Yeah, not bad! What kind of gearbox there is? Single speed (a simple transmission 1 : x ratio) or something more sophisticated? IIRC from your video, there you said motor could do up to 16k rmp, and at top car speed seems it actually hits that.. impressive.

  14. Rodger says:

    Very cool.
    Any thoughts on grabbing a Smart Electric Drive and reverse engineering its controller?
    (The Smart ED uses a permanent magnet rotor.)

    • Michal says:

      As a matter of fact, that is indeed what I will be doing in approx two months. I’m just discussing a possibility of getting some Smart drivetrains.

  15. Doug Dority says:

    Brilliant work. You are the first person I have seen to actually do an educated hack of a very sophisticated drive train, and you have convinced me that the era of EV hot-rodding has begun.
    I drive a P85D (soon to be a P85DL) and eventually will want a “P110Q”. If Tesla doesn’t make such a car (Q stands for Quad motor) in the next 4 years, I will. I expect Tesla may be planning this since the Tesla drive unit positioning is set up to be perfectly symmetric through the case both front and back. It would be possible to create Siamese dual-drive units for both the front and back of the car, creating true independent traction control on each wheel and sufficient traction to become the quickest car in the world.. Your UMC boards would make this possibility much closer to reality for DIY hot rodders with money, engineering experience, CNC machine tools such as myself, just lack spare time…..
    Keep blogging and I will keep following. Thanks for the inspiring engineering and classic ‘Hacking’ effort.

    • Michal says:

      Hi Doug,
      Thank you! I agree, the era of EV experiments is here – and I’m very happy to be in it. There will certainly be more updates in the near future, and you will see quite a few cars using Tesla motors, Chevy Volt inverters and similar equipment in the next year. The Tesla drive unit is perfectly symmetrical thanks to it’s inverter placed in the same axis as the drive motor – very cool and efficient placement, if you ask me. I’m not sure if they do a quad motor anytime soon, because that would force them to redesign a good deal of the drivetrain. Plus if Tesla decides to do quad motor, we are bound to see another step in the EV evolution – it would not just be slightly modified Model S, but most likely a completely new approach. Maybe they will finally solve the hub motor plague? (Nobody yet managed to make a good hub motor car).


      • Doug Dority says:

        Hi Michal,
        I agree about the drivetrain re-engineering; there is much beauty in Tesla’s inverter packaging design, and even more in that it can be unrolled. I would still rigidly mount it to the drive unit assembly (to avoid buss and cooling circuit flexure), and use separate cooling circuits for the motors vs inverters. (My 30 year engineering career has been all about electromechanical packaging, so I love this stuff!)
        The Tesla front drive unit packaging is not nearly as pretty as the rear unit, and the 85D apparently uses the same motors front and back. (I would base the quad on an 85D, not a P85D). I don’t know if the 85D rear drive unit is simply a de-rated rear unit, or based on the front drive unit. (I suspect the former) .
        The reason I think Tesla may attempt a Quad is that Elon Musk said the RoadsterII would have ‘maximum plaid mode’, which I think indicates a blank-sheet design approach, and will include true independent traction control and torque vectoring. Tesla typically ‘beta tests’ by trying their designs on existing platforms- the Roadster on the Elise, the AWD for the X as the SXXD variants, and I suspect they will try to jack Quadmotor Siamese drive units into the S platform as a test bed for the RoadsterII. No matter what, lowest possible CG, mechanically integrated inverters, low radial mass, are common design parameters for any drive unit; the Siamese unit by definition would fit in the same place as the existing units-less drunk, no rear footwell for the P110Q….
        I’m curious to see how (Honda) Acura NSX front Siamese drive unit will be packaged.
        I had looked at hub motor designs years ago and was curious if anyone has tried to build an external rotor AC induction motor (non PM) using machined aluminum/composite squirrel cage style rims. I still would expect the stator laminations to be fairly heavy though….
        Light driveshafts with inboard motors have been the simplest solution so far and pretty hard to beat. The unsprung weight problem of wheel motors may never see a good solution.
        I’m curious to see if you have been linked through the TMC or Tesla forums yet. Your stuff is goood.

  16. Robert Grann says:

    Hi Michael and congrats to this great achievement!
    I’m confident we will see a big new market for this, ElectricTuning πŸ™‚

  17. D says:

    Your UMC sounds very useful! Looking forward to details of its construction.
    Is its control signal a target speed, target torque, target motor-current or something else?
    I’m wondering how easily it can be made to do traction-control for snowy hill-starts or for slow-speed in the mud, or torque-vectoring on fast corners.

    I was just reading about another self-learning universal motor controller: at http://ackinetics.com/ they claim their software gives tighter speed-control and better motor efficiency than some (unnamed) competitors. See the “Test Hardware” and “Results” pages, though I don’t know whether or not the common alternatives are as bad as the ones they tested there. In any case, it could be interesting to try some measurements and see how your controller compares. If theirs is better, you have a target to aim for. If yours is better, rejoice!

  18. Sarunas says:

    To be honest, using other controller is not hacking.
    Have you tried to figure out what protocol it uses, to run it only via CAN?

    • Michal says:

      So reverse-engineering the HW interface and developing system that can exploit it is not hacking, while just logging the CAN bus and replaying it to the CAN bus is? πŸ™‚
      Sure, you can log the data and reply it. But I’m not too interested in it.


  19. Matej says:

    Any progress on UMC?
    I will be designs power stage for electric scooter and I would like to use your UMC as a brain. πŸ™‚

    • Michal says:

      Plenty of progress! There is a new revision coming with fully isolated HV phase measurement and much smaller package (single PCB solution). Stay tuned for details.
      PS: Nice gmail name..

  20. Elvis says:

    Hej Michal,

    Where did you buy the drive train from a Mercedes b class.
    I am working on a project where I need such an drive train
    is there a website you can recommend for search for it?

    by the way, it’s an interesting video you have made whit the (Tesla motor hacked!)

    Best Regards Elvis

    • Michal says:

      Hi Elvis,

      Thanks! I don’t think that there is a particular website where you can search for this. Maybe except for crashed cars auctions.
      It takes a lot of asking around and good old phone calls to get such things. It will get better in few years.

Leave a Reply

Your email address will not be published. Required fields are marked *