DIY Electric Car Forums banner

Fun with Tesla motor

133K views 142 replies 32 participants last post by  matchke 
#1 · (Edited)
So last week I've became a proud owner of a Tesla drivetrain:



On the way home, we stopped by a Tesla superchager - the owner of the brand new white Tesla was quite surprised, to see a Peugeot with a Tesla drivetrain :)



By the looks of it, this motor indeed is the same one as in Tesla Model S. I have no idea if different battery in Model S means different motor, but most likely they are all the same (as power limiting factor is the battery).

The history of this motor is quite complicated. It came from Mercedes, where they were using it on a testbench. Most likely this motor never even seen a car :) The reason why Mercedes had it, is their new Mercedes B electric car. I presume that they couldn't be bothered with downsizing their licensed Tesla parts (they bought everything from Tesla, including the battery and charger), so Mercedes B motors are (most likely) Model S motors. If someone has the dimensions of Tesla Model S motor, please write them here. I would love to compare them.

I've bought the whole drivetrain - motor, single speed gearbox and the inverter. Since this is a factory test motor, most likely the firmware will not be compatible with Model S (it might be compatible with Mercedes B). In the best case it just spits bit more diagnostic information, but most likely having a Tesla capture will not spin it. This is okay, as my plan is to use my own motor controller to replace the "brain" of this beast. Over the past year, I've started to enjoy performing lobotomy to different inverters :) (link in my signature).

I'll be using this thread to give some information about this motor, and how to use it in your conversion.
 
See less See more
2
  • Like
Reactions: JRP3
#4 ·
WOW!
What's you plan? What vehicle are you going to put it into?

It makes my latest acquisition, Beryl the EV, look a bit, ummm, rustic! ;)
Well, I bought it just because it was cheap (unfortunately I promised the seller not to give the number, but I payed much less than EVTV did).
There are several candidates, but for now main ones are old Porsche or Lotus Esprit Mk1. Anyway, it will be an interesting conversion to do.

If numbers match Tesla Model S, we could be looking at 310kW 600Nm.
 
#6 ·
So it Seems that kennybobby was right :)
This motor indeed is the high-end Tesla Model S motor.

I started with the inverter. The idea was to find a way to get out all the electronics and power parts to judge on what kind of power levels could I expect, and how difficult it will be to do the hardware hack of the Tesla inverter.



Main DC cables go off quite easily, just one screw on the orange plastic protector, hiding two massive bus bar screws (DC from the battery pack).



Now is the time to remove the whole aluminum cylinder protecting the inverter. You just have to go around and release ~8 screws. Some warranty had to be voided :)



Now just slide it off and observe the beauty (more like a beast, this is a 300kW 3 phase inverter!)



The nice triangular design is clearly visible (each side is one phase). The PCBs on top are IGBT drivers. You have to remove the top to access millions of screws that joins the inverter to the gearbox.



Orange cap on the gearbox assembly hides three phase bus bars that routes the power directly to the motor. They have to be removed as well.



Unplug the temperature probes and other wires that goes to the gearbox and motor assembly. Now nothing is preventing you from removing the whole inverter :)



Pay attention to the clever cooling cutouts, don't damage the sealing as you pull the inverter.

I've done a basic analysis of the circuits and interconnections between modules, checked the power parts and I'm pretty certain, that I can make it compatible with my UMC Drive controller. More on that some other day.
 
#8 ·
Thanks! I don't have an exact figure, but I would place this somewhere around 140kg (motor+inverter+gearbox). The power per kg ratio in Tesla is amazing.
 
#10 ·
Its like watching you take apart a UFO. Love it!:cool:
 
#11 ·
Great photos, thanks for sharing. Looks like you have a fun project.

Would you be able to measure the resistance and inductance at the motor lugs, phase to phase? Maybe at DC, 60 or 120 Hz, 400 Hz, 1000 Hz?

Why is this important or of any interest? Because these electrical characteristics are used in the calculations made by the inverter to drive the motor using a Direct Torque Control (DTC) scheme, described in TM patents.

For a 3-phase, 4-pole wye motor with the poles wired 2s2p, and with 4 Turns per pole using 16 AWG copper wire 12-in-hand, i've done a rough calculation:

phase inductance ~ 493 nH
phase resistance ~ 5.3 mR

This gives an L/R time constant of 93 usec. If we use 3 tc to reach full current, then the inverter would need to generate a current waveform at about 895 Hz and the motor max speed ~ 26,858 rpm. With a 9.7:1 gearbox the theoretical max speed would be ~228 mph, but realistically the available motor power at that rpm wouldn't be enough to propel the car that fast.
 
#12 ·
Great, I'm glad there are people seriously looking into the design of that motor as well :) I'll be more than happy to take those measurements for you. It will be next week, as the motor is in my friend's garage and I took home only the inverter. (I already have more than enough motors stored in my flat..)
 
#14 · (Edited)
There seems to be an encoder mounted on the side of the motor - not a resolver or anything like that. I still have to look into that section of the control, but there are no resolver digitizers onboard, so it is safe to presume Tesla is using standard optical/magnetic encoders.

kennybobby, the Direct Torque Control actually sounds about right, as I found very hardcore phase current measurement system in the inverter. Every other manufacturer uses hall effect sensors to measure the phase current (two or three), but Tesla designed a bus bar that has a small section made from less pure / more resistive material, and soldered a voltage drop sensor on top of it.



This configuration will be very precise (definitely better than 1% accuracy), but has to be quite difficult to manufacture just right, and every sensor like this is must be compensated for resistance error. You can see different coloration in the material, I would presume that this is due to laser trimming - they put the bus bar under test and cut in it using high power laser, to get the resistance just right. They went for this most likely because the DTC have high requirement on current measurement accuracy.

Someone might ask, why to go through all this trouble with laser trimming of the shunt, when we can easily compensate for it in software? That would perfectly fine in one-off equipment, but we are talking here about large scale automotive manufacturing. When you receive a batch of current sensors, someone would have to pick a bus bar, write it's calibration parameters in the firmware and upload that in the inverter. This creates delays, adds complexity and can lead to errors and problems during regular firmware updates pushed by Tesla.
 
#16 ·
Looks like they are changing the resistance by over heating the bus. I had a fire in the engine compartment burning the insulation off the alternator power wire, so I taped it up but got a charging failure light. I was told to replace the over heated wire ,I did and it worked fine.
 
#28 ·
Great stuff! Thanks everyone for sharing your findings. I'll try to slowly answer all the questions, as I get the chance to measure/test stuff. I will not be taking the motor apart anytime soon (if ever) - exactly due to the cooling system. There is a good chance that as I pull out the rotor, some of the coolant will contaminate the gearbox oil, or in the worst case I damage the actual seal.

I have just put the inverter on a scale - without coolant and outer cylindrical case it weights 12kg. The case itself will be around 1kg. That would make this 300kW/13kg = 23kW/kg. Not bad!
 
#36 ·
The same way as I did with the Chevy Volt - by following circuits, searching for buffers to get clues about what are inputs and outputs, tracing op amps and their gain, voltage levels on power supplies, current sensors' scaling.. Basically you need to know everything about every wire in the inverter, to be able to replace the controller :)

The traditionally annoying things are: hard layer of conformal coating and 4-6 layered PCBs. But at the end fortunately I don't really need to see the inverter working as a whole, passive analysis is enough to be able to interface it. Then you still have to figure out which control signal is doing what (half bridge PWMs, driver enable and so on).



It will take me a couple of weeks to finish that (many other projects - finishing my car, assembling and testing the UMC Drive with other inverters..), but from what I know already, it looks quite promising.

Before someone asks this question - yes, it would indeed be better to decipher the CAN communication for this thing, and not to do any modifications at all. But getting the bare CAN minimum to know only how to command some torque, it wouldn't really be my inverter. I like doing things the hard way. Plus there is a good chance that my inverter is a hybrid between Tesla Model S and Mercedes B electric. No idea at which state of the project did Mercedes flash my firmware, or if it is even remotely similar to the production version of any of those cars.
 
#37 ·
Hi guys,
Time for a small update.

Motor specs:
The motor has a maximum rotor speed of 16000 rpm.
Base: 415 Nm of torque and 270kW for a motor shaft speed between 6,500 rpm and 9,000 rpm
Performance: 590 Nm of torque and 310kW for a motor shaft speed between 5,500 rpm and 9,000 rpm

The drive inverter delivers 900A rms to the motor in a Base drive unit, and 1200A rms
in a Performance drive unit.
The gearbox is a single speed 9.73:1 reduction.

As for the encoder - it is an hall-effect encoder, with two output signals, 90 deg out of phase. It is serviceable, you can remove the sensor from the motor (it is the black plastic part on the motor side)

The rotor cannot be removed from the gearbox assembly. So reusing just the Tesla motor without the gearbox is not possible.

I'm just days away from being able to spin the Tesla motor using the original inverter and my motor controller solution. The game is on, EVTV.
 
#38 ·
Awesome! I hope you do get it going.

The EVTV crew is usually not the first to do something. Being second can still be kind of cool. ;) It's actually kind of neat that the two of us are taking different roads to get there. There are benefits to each technique and I'm happy to hear that you've had reasonably good luck with your approach so far.

Take video. Lots of video.
 
#39 ·
Thanks, Colin - I mean it in a friendly way ;) I like the work you guys are doing (especially now with the heater, that was a tough cookie to crack).

I'll make videos and pictures :) I think that the plan for this weekend is clear.

BTW, did you observe any differences between your drivetrains? If I recall correctly, you have one performance and one from a 65kWh unit? If my findings are correct, there are no real differences. It will be limited in the firmware of the motor controller (makes sense, as manufacturing two different rotors/inverters would not bring any advantage to Tesla).
 
#42 ·
...It's neat that they make use of the high speed capability of brushless motors instead of just throwing torque at it...
Both I would say. Tractive effort is what accelerates the vehicle, which is proportional to motor torque for a given wheel diameter and gear ratio. But to maintain that high tractive effort at higher vehicle speeds requires high power. They have both. I assume the 6500 and 5500 are base speeds, and from those to 9k is the approximately constant power region?

The Chevy Spark has higher high peak torque to vehicle mass ratio, but low base speed. Similar issue to HPEVS/Curtis stuff, acceleration rate decreases at higher vehicle speeds. The Tesla S can sustain it.
 
#45 ·
The motor "face" and bearings are a part of the gearbox assembly, plus you have a water jacket through the rotor. So if you somehow hack apart the gearbox cast, you will find a splined shaft entering the gearbox.
 
#46 ·
Would you even want to hook this up to another gearbox? I believe that you will just be destroying these boxes.

The lash in a standard gearbox is way to much for an electric torque monster.

A few questions;
What kind of voltages do the gate drives operate at?

Also how are the igbt's paralleled, multi gate drives or one driver for all or all?

Any form of torque damper to be observed in the gearbox? (spider clutch/tuned shafts(a shaft in a shaft)
 
#47 ·
I'll write a report on the electronics eventually, but I don't want to split this information too much.

Plus some of it I don't know yet, or I'm not 100% sure. Also I had some totally shameless "representatives" from major industrial power electronic companies stalking this thread or even trying to contact me with questions about particularities of the Tesla design. I'm saying shameless because they of course didn't even want to offer anything in exchange for that information. If there would be a way to make sure they do not get this while you guys do, I would do it. If you want to discuss this, or tell me your opinion on this matter, you can write me a PM.

I agree that you don't want to use this motor with a different gearbox anyway. Tesla wanted to have a two speed gearbox in Model S, but they didn't manage to get/design one that could survive (that's the rumor).

For the torque dampeners - from what I've seen, there is no such thing in there. The fact is, that you can make a reasonably accurate predictor that limits your torque slope. Or even takes the backslash in the gear set into consideration - that occurs when changing the direction of current (regen to drive transition). At least that's what I would do. Can't talk for Tesla though :)
 
#48 ·
The main problem is the torque ramp up. Especially from "stand still" or coasting (no load). Gears do not enjoy or handle shocks well. Seeing that electric motors can deliver instant torque and with out a clutch that can slip, the teeth take a beating every time you change your throttle position.

I believe tesla had to ramp their increase torque/second down a whole lot to keep the car together. Have seen some "simpler" converted cars destroy axles gearboxes and spline couplers.
 
#56 ·
I alluded to the welded up housing in post 29 partly in jest and partly because the housing looked to be welded up in an earlier cut away photo I'd seen. Good to get conformation. I suppose this could be to deter other uses of the motor. I've cut through many a weld and casting to get at parts, so it wouldn't stop me.

The catalytic converter thieves could really clean up stealing these copper rich motors if they were, instead, just bolt on items.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top