Tesla Large Drive Unit (LDU) Motor Teardown and maintenance

[muehlpower]

Here in Germany there is a manufacturer of PTFE sealing rings. I contacted him 2 years ago and received an offer. At that time I assumed a bearing inner ring with 35mm to repair the damaged surface. The dimensions were 35x55x8 with 2 lips. He wanted 78€/piece for 10 pieces, 61.10€ for 20, 37€ for 100, and 16.70€ for 500.

[howardc64]

DELETED

[muehlpower]

That's the company I asked about: https://www.seals.de/en/solutions/shaft-seals.html

IR30x35x16-EGS 13€
IR30x35x13-EGS 11€

EGS means: "Surface ground free from spiral marks for rotary shaft seals to DIN 3760 and DIN 3761"

DIN 3760 means: Ra = 0,2µm-0,8µm, 55HRC above 4m/s

[Johan]

If you plan to use a Speedy Sleeve because the shaft surface is bad, then I would highly recommend using a thermally conductive epoxy to fill the microscopic gap between sleeve and shaft. Not doing that can (according to my heat transfer model) severely increase the seal temperature and cause early seal failure. For more background and a numerical example, see here. Not doing it is like mounting a CPU on a heat sink without thermal grease.

In case you want to try the triple lip Ceimin seal and simultaneously need a Speedy Sleeve: Shown below is a Speedy Sleeve that has been positioned such that the sleeve's "outer" chamfer edge coincides with the shaft's "inner" chamfer edge so that both lips can touch the sleeve. Now the excluder lip interferes with the sleeve flange, so, simply cut off the flange with (for example) a Dremel cutting disc (first wait for the epoxy to dry and wind protective tape around the sleeve!). After mounting the sleeve, that flange has no function anymore anyway. Shallow damage to the shaft from cutting does not matter bc it is ruined anyway, that was the reason to use the sleeve in the first place.

Double check the dimensions of your LDU first, because different revisions may have different dimensions.

1670279937729.png (52.16 KiB) Viewed 5026 times

[howardc64]

DELETED

[muehlpower]

If I summarize, I come up with a PTFE seal and a good shaft that only needs to be polished, or the use of a speedi sleeve with something for better heat transfer. The PTFE seal would then be the same. So I will ask XXX what the best material and design is. Shaft 55HRC, Ra 0.4µm, medium water/G48, speed 28m/s, pressure max 0.3 bar, temperature max 90°C. And what he thinks about a diameter of 30.5 with a sleeve instead of 30. Did I forget something?

[howardc64]

DELETED

[Johan]

If I summarize, I come up with a PTFE seal and a good shaft that only needs to be polished, or the use of a speedi sleeve with something for better heat transfer. The PTFE seal would then be the same. So I will ask GFD what the best material and design is. Shaft 55HRC, Ra 0.4µm, medium water/G48, speed 28m/s, pressure max 0.3 bar, temperature max 90°C. And what he thinks about a diameter of 30.5 with a sleeve instead of 30. Did I forget something?

Seems correct and mildly conservative, which is good. Maybe the 28 m/s (seal slip speed) is a bit high because it corresponds to 250km/h = 155mph for my wheel size. Maybe use 137km/h = 85mph -> 16m/s seal slip speed, which may open up a larger solution space for seal designs? Unless faster driving is a desired.

Also the max coolant pressure could be higher: 1.3 bar (source).

Also mention, just to be sure, that the seal should seal both statically and bidirectional dynamically (duh) and that the shaft is horizontal.

Maybe also request a dynamic service life of at least 500 hours (500 x 100km/h = 50000 kilometers)?

Also there is a bunch of dimensional constraints like the primary lip (touching the coolant) can stick out no further than 2mm from the housing because if more, it may rub the shaft chamfer edge. Also the excluder (if any) cannot stick out further than 3mm (assuming 8mm thick housing) to avoid touching the shaft chamfer edge for the reluctor wheel. Also not sure what the best interference fit size is, i.e. best outer diameter for stainless steel housing (Ceimin is 55.2mm I think).

[howardc64]

DELETED

[smathermather]

Probably any new version of MB or Rav4 EV LDU are just Tesla's latest LDU revision #s. MB/Rav4 EV volume is much smaller than Model S so likely they will track LDU rev slower.

Makes sense. With a few thousand MBs (and I think a similar number Toyotas) in North America, that's not a lot of incentive for MB to do anything beyond using essentially what Tesla has. I hope that's wrong, but suspect it is not now that you put it that way.

Sigh. Well, I guess I'll be following this thread and checking my speed sensor with some regularity.

[WimV]

For the speedi sleeve: the flange is pre-cut (you can easily pull it off after making an incision.

I do think speedi sleeve is not such a good solution because of the tight tolerances of PTFE seals (0,5 mm diameter increase is probably just to much).
There is also a “speedi sleeve gold” that is coated and more resistant to abrasive (probably better for PTFE seals). The problem is that it’s not offered for a 30 mm shaft.

I thing machining the shaft is the only good solution in case of wear on the sealing surface.

[howardc64]

DELETED

[muehlpower]

A small update. I had a long talk with an engineer from XXX today. He sees no major problem. All shaft versions are possible, i.e. good and polished, speedi sleeve and inner ring. He is relaxed at the temperature, he describes the pressure as low and the speed up to 18000 is feasible. I sent him a small drawing with the installation dimensions and some photos, also with rusty shafts. I will send him a package tomorrow with a water manifold, a destroyed original sealing ring and a Chinese ring. He evaluates all of this and then makes a suggestion. Then he makes samples that I can then test. For this I build a small test stand with which I can run speeds of up to 20000 rpm and change the pressure and temperature. In 4 weeks I can cover 80000km without ruining an LDU! I can't simulate the original tread because I don't know about it. I will use an inner ring with an outer diameter of 30mm as a tread.
I hope google translated correctly

[howardc64]

DELETED

[howardc64]

DELETED

[howardc64]

DELETED

[crasbe]

University of Stuttgart has professor and laboratory (lab) studying PTFE lip seal for 20 years. I emailed lab leader professor in English but didn't receive reply. His PhD from Stuttgart is PTFE lips with helix pattern. Now running lab researching PTFE lips. They are probably supported and closely affiliated with German automotive industry.

It is not really untypical for a Prof to forget answering a mail. You have to imagine the flood of mails arriving each day. The prof of the institute where I work explained it like this: He sees a mail and when he forgets to answer, it's so far up in the mail inbox after just a day or two, he'll never find it again.

You can either try to send the mail again or try to contact one of the research assistants/PhD students who might have a project that's in the research area. The research assistants usually reply a lot more reliably than the profs

[howardc64]

DELETED

[howardc64]

DELETED

[howardc64]

DELETED

[muehlpower]

A little tool I made to pull the gear off. Should also work to dismantle a Speedy Sleeve.

[jsimonkeller]

I wanted to share my coolant leak experience, since it is now "patched up" thanks to the help and resources of Howard. I believe he shared some photos already, but I obtained my LDU back in 2018 and it sat for a little more than a couple of years before coolant was added to the system and then maybe close to another year before the coolant was running through the system with pumps on a regular basis with testing. I got about 90 miles of driving before the erratic behavior began. I posted about that on here before I knew the issue. It started with losing power on braking and having to be towed home in shame . . . and then the overcurrent issues that I thought was in the programming . . . and finally the pot settings went awry and the temp was reading -1230 c, which is below absolute zero!!!!

I was assuming the card was bad and when we opened up the inverter to replace the card, we saw the coolant invasion.

This meant we needed to check the other side of the motor and sure enough . . .

After cleaning off the rust, it was clear the shaft was pitted, which was not unexpected by Howard.

CONTINUED IN NEXT POST

[jsimonkeller]

CONTINUED . . . .

Howard explained all of the options and since my car was in a shop with lots of cool tools and my partner on this build has a lot of experience in metal work and welding, we went for the option to undercut the shaft and then add a new outer sleeve of stainless steel at the same length as the shaft to accommodate the seal.

We closed up the breather hold and drilled out the new emergency drain and set the new seal (courtesy of Howard's stock) and re-installed everything.

We also added drain holes on the Inverter side.

[jsimonkeller]

I am reusing the DU card that was in there during the coolant leak, along with same harness (thorough cleaned) for now and it powered on fine and temp readings are correct again and pot nom is back in line with where it should be and voltage readings are accurate.

I am getting NO ERRORS at the moment BUT the OPMODE is staying in OFF and I have not been able to get it back to ON. I am trying to determine if the card just has some damage I cannot see that is causing this or if the inverter has any damage. No visible damage to the inverter.

I am not sure if there is any troubleshooting anyone would suggest at this point. I plan to swap the card out for an older one I have in a box somewhere to see if that solves the problem and answers the OPMODE not coming on and damaged DU card issue.

[LunaticScientist]

All make sense. However, the rotor's shaft locks in 30mm diameter and ~8mm seal width so these can't be changed. Only remaining design variables are material + shape + shaft surface prep.

Thinking about this aspect, what if we COULD make the sealing shaft diameter larger? I.e, a machined aluminum adapter, press fit into the end of the rotor shaft, sealed against the original shaft with RTV?

It could allow less stringent design requirements (possibly an off the shelf seal) with respect to the surface area speed and material while still being removable for service, lightweight to keep added mass and stress down and field installable without removing the whole drive unit.