Tesla Large Drive Unit (LDU) Motor Teardown and maintenance

[howardc64]

There are some data samples that suggest early Tesla LDUs are lasting a longtime without leaking. Some people claim 150k on original. Met a 2014 owner @ 60k miles on original LDU just recently.

There is also evidence Tesla LDUs (original and reman) speed sensor are coated with grease (and some say also inside speed sensor cavity)

Here are few examples

60k+ 2014 original LDU speed sensor pulled 8/23. # lips unknown.

IMG_1404.jpg

IMG_1403.jpg

My leaked 2017 reman at 30k miles

While coolant leaked, speed sensor is coated with a base coat of grease

https://teslamotorsclub.com/tmc/posts/7011708/

Here is a 2020 (probably reman) with about 30k miles

No leak but greasy residue on sensor and its cavity

https://teslamotorsclub.com/tmc/posts/7085069/

Most likely source of this grease is applied on shaft/seal lip on install. Outer rotor bearing grease doesn’t seem to push out this much grease. Anyone have idea what type of grease this is and where applied?

In @Johan and mine rebuild process, no grease was applied to shaft or seal lips. Speed sensor checks shows it is bone dry without any grease.

I wonder if original LDU longevity comes from

1. Originally manufactured LDU with rust protection (eg plating) and polish to spec.
2. Triple lip seals installed (QCC and EU rebuilders all claim original triple lip lasted much longer) with grease packed between the lips? Maybe this acts as a grease cache?

[mrhyde]

Hello everyone! I’ve been lurking for a while and am the owner of the LDU Howard posted speed sensor pics of in the previous post. The car is a 2014 P85d with 67,000 or so miles on it. The tag confirms this is the original LDU and I am thrilled that it looks like I have a unicorn unit produced with the better early seals but likely with ceramic bearings. Howard rode in my car and confirmed it is fairly quiet. My hearing isn’t good and this is my first Tesla so I don’t have the experience/perspective to have an opinion on how it sounds. I’m thrilled the original owner didn’t give Tesla a reason to swap this out for a remanufactured unit!

At any rate, thank you for the great work here. I look forward to following along and contribution whenever I can.

[jrbe]

Oil testing companies might be able to test a sample of the grease and tell you what it is, how it's doing, contaminants, etc.
Cost + time it probably makes sense having an oil analysis company identify it.

The diy route:
Looks like normal Super Lube grease is glycol & ptfe friendly and should be readily available.

PAG greases might also work from a quick search.

I found golden syrup as an option too. A bit odd..

Might be worth narrowing it down to a few different greases that are compatible with both glycol and Teflon to buy. Then do a sniff / finger slip / tacky comparison vs what's in an original greased LDU. Might be able to match the original grease that way.

Can also mix some g48 with the different greases, heat them to almost boiling (maybe a few times) stir them around and see how they do. Freezing the 2 mixed might be a good test as well.

[howardc64]

Oil testing companies might be able to test a sample of the grease and tell you what it is, how it's doing, contaminants, etc.
Cost + time it probably makes sense having an oil analysis company identify it.

The diy route:
Looks like normal Super Lube grease is glycol & ptfe friendly and should be readily available.

PAG greases might also work from a quick search.

I found golden syrup as an option too. A bit odd..

Might be worth narrowing it down to a few different greases that are compatible with both glycol and Teflon to buy. Then do a sniff / finger slip / tacky comparison vs what's in an original greased LDU. Might be able to match the original grease that way.

Can also mix some g48 with the different greases, heat them to almost boiling (maybe a few times) stir them around and see how they do. Freezing the 2 mixed might be a good test as well.

Thanks. Interesting thoughts. I could trial a grease but challenge is each trial = PITA LDU home garage pull+install on a Tesla Also challenging to get hold of a speed sensor with ample grease residue and an owner willing to aid grease search. Finally a question of grease best with single or multi lipped PTFE?

Proper rebuild process+questions seem to be

1. arc spray repair of damaged shaft
2. plating for rust protection, polish to roughness spec
3. unfortunately it seems arc spray shop don't plate, plate shop don't spray (main business is chrome parts to restore old cars)
4. locate a PTFE seal (still impossible seeking traditional high quality brands, so custom made / AliExpress Ceimin)
5. grease? what kind? where? work better with multi-lipped seal?

A lot of variables. Tesla provides no info/parts. Rebuilders aren't disclosing at this level of detail. Just enough to market their service.

My arc spray shaft region supposed to be stainless, I guess question is that enough rust protection? If so does it cover sufficient region for an intended seal? Single lip seal probably doesn't matter. But if Multi-lip + grease in intra-lip cavity is best result, then multi lip is far wider region which may still need plating+polishing to cover the entire region.

This probably shows why coolant instead of oil cooled rotor is a bad idea with multiple failure mechanisms. #1-#3 could be where Tesla reman is failing which would make #4-#5 ineffective as shaft is half of the sealing surface.

[howardc64]

Coolant Seal Grease Update

Confirmed with LDU rebuilders they also believe Tesla use grease on shaft and coolant seal during install and its the reason for finding it on speed sensor.

However, many PTFE seal installation guidelines explicitly say clean and dry installation and note grease will ruin the seal. Confirmed with LDU rebuilders they follow these guidelines with good results. My guess is Tesla may use some grease to help reduce chance of lip damage over the leading edge of the chamfer on the shaft (accommodating tech's skill variation?). Most PTFE seal's media facing lip will be ~28mm (or even smaller) depending on profile and will be smaller than the chamfer lead-in tip diameter. It is critical the chamfer's tip is smooth, free of burrs as it is basically the blind PTFE seal install tool. SKF (and many others) provide good summary on the care required on PTFE seal installations. Summary outlines difficult (blind) install scenarios where burr free long lead-in chamfer is required ( https://www.skf.com/us/products/industr ... tions/ptfe )

As a side note, my spray metal repaired shaft has some tiny burrs on the chamfer tip. Close inspect shows the chamfer is likely spray flamed over and polished but leading edge not rounded to remove burrs. Pre spray flame repaired chamfer shows visible concentric machining rings but no sharp edges (all rounded over). These rings are all gone post spray flame repair. Will take it back to rebuilder to clean up the burrs.

IMG_0760 (1).jpg

[howardc64]

Post Metal Spray Shaft Chamfer Analysis

Here is what happens to the chamfer post metal spray

Screenshot 2023-08-19 at 11.22.37 AM.png

There are 2 problems

- Since metal spray is porous, the chamfer tip is slightly rough unless cleaned up. Ideally, the tip would be rounded off slightly.
- Chamfer is not milled down so the chamfer tip diameter increases slightly. While its small (probably < ~0.25mm) There isn't much margin as PTFE lip diameter are frequently smaller than the opening.

Chamfer tip diameter is now about 28.5mm (didn't measure original)
Ceimin's coolant facing lip ID is 26.5mm. Other seal designs can be slightly larger

Seal lips will slide against this tip during installation. Elastomer seal can handle more imperfections but PTFE's hard plastic like material will micro cut easily with lip profile an additional variable. SFK's PTFE seal installation outlines a "long" chamfer is desirable for difficult (hinting blind) installations. This would have improved the margins on the LDU.

Anyway, very high care and precision work to prepare the shaft and seal for installation. Having multiple shops involved (say metal spray + plate + mill/polish) probably makes it more difficult. Potentially a reason Tesla LDUs use some grease.

For my effort, need to find a way to get a nice smooth rounded tip to avoid potentially damage the PTFE seal's sealing surface on install.

Modifications to the Metal Spray Process

Probably a good modification if possible is to mask off the chamfer (if not damaged) to preserve the original size and finish.

[howardc64]

Microscope Closeups of Metal Flame Spray

Photo on 8-22-23 at 11.44 AM 2.jpg

Photo on 8-22-23 at 11.45 AM 2.jpg

I'm guessing

- PTFE seal lip will not survive sliding over porous chamfer tip
- pits (or filler) on the finished shaft surface (naked eye see as tiny specks on other wise smooth surface) unknown for sealing performance.

Probably chrome plating will yield better results. Or more likely use speedi sleeve with FKM elastomer seal. Repair shaft to narrow tolerances probably can't be accomplished with regularity.

UPDATE

Spoke with spray flame shop, they use a sealer to fill the porous holes after metal spray. Here is an example

http://diamantpolymersinc.com/thermal-spray-coatings/
https://www.ep-coatings.us/thermal-spra ... y-sealers/

Curing these seems to require 200C heat in above link. My rotor shows signs of baking (some color hues on metal surfaces) and remaining spline grease seems cooked/dried.

[jrbe]

I thought you may have used a linty rag to clean it from a previous pic you posted, that's disappointing. I think you'd be better off with an undersized seal area vs. how porous / rough the repair ended up.

Most plating options will tell you to get the surface as smooth as the desired Finnish before plating. Might make sense to clean it up then get some thick plating over it to bring it back up to seal spec OD. (Skip the metal flame spray) I'm not sure what plating would be ideal for the seal.

Plasma spray deposition could end up a cleaner surface finish if you want to stay on a similar path. Probably could see some repairs that were done to see if it's worth pursuing.

You could passivate the seal area / end to help keep corrosion away.

[howardc64]

I thought you may have used a linty rag to clean it from a previous pic you posted, that's disappointing. I think you'd be better off with an undersized seal area vs. how porous / rough the repair ended up.

Most plating options will tell you to get the surface as smooth as the desired Finnish before plating. Might make sense to clean it up then get some thick plating over it to bring it back up to seal spec OD. (Skip the metal flame spray) I'm not sure what plating would be ideal for the seal.

Plasma spray deposition could end up a cleaner surface finish if you want to stay on a similar path. Probably could see some repairs that were done to see if it's worth pursuing.

You could passivate the seal area / end to help keep corrosion away.

Yes disappointing indeed. Ukraine based EV repair shop (CarRepairVideos on youtube) shows flame spray repair of damaged Tesla rotor shafts. Their process is not even on a lathe. They just hand rotate the rotor periodically and spray away on the shaft. However, from my experience so far, a basic sealer+baking cure pass to cover the pores is insufficient. Maybe a heavy coat, 200C oven cure followed by polishing can achieve better results but after seeing the poor spray metal results along with over sprayed chamfer that damage its installation capability. Am quite reluctant to give the rotor to anyone who hasn't studied this application in detail. These shops probably never deals with the uncommon sealing coolant situation such as this rotor.

Anyhow, after flame spray, I'm at the same situation as a rusty shaft from a long idled salvage yard LDU. Speedi-sleeve + single lip elastomer is the best option now. @Boxster EV is is at 15k miles on this solution with a SKF FKM seal.

One might argue a Tesla reman rotor shaft is basically same situation as a rusty shaft from salvage yard since its unknown what the reman process is for the shaft surface. My 32k reman shaft surface had a coating of coolant solidified haze and a tiny scored groove outside of the narrow single lip seal (but is under the wider triple lip seal) while @Johan's 40k mile new LDU shaft surface was pristine. In general, have found other poor workmanship and damage (spun bearing, stripped bolt holes) likely from inexperienced tech beyond the seal shaft area on my remanned LDU. Very poor job.

[Boxster EV]

@Boxster EV is is at 15k miles on this solution with a SKF FKM seal.

Yes, with this seal (to be used with Speedi sleeve).

SKF CR30x55x7HMS5V Single Lip Viton Rubber Rotary Shaft

[howardc64]

Summarized LDU rebuild seal rotor shaft decision choices. Probably 2 main options

- If have pristine rotor shaft surface -> Ceimin triple lip seal without shaft surface prep
- If poor shaft surface -> Speedi-sleeve (with sharp leading edge seal installation protection) + SKF FKM / @muelhpower single lip PTFE

Also noted easy to strip inverter mounting bolts and considerations on refinishing the shaft surface.

LDU rebuld decision.pdf

[howardc64]

Some additional info on skipping coolant cooling of rotor.

Besides Tesla PN showing such manifold ( viewtopic.php?p=59851#p59851 ) Received recent contact from Model S LDU owner in Poland indicating Poland local EV repair shop is modifying manifold to bypass coolant from rotor cooling. They claim stator cooling is the key.

Would be nice to have a bench setup to and instrumented to compare rotor temperature. Anyway, there are now 2 hints at this solution possibility.

[jrbe]

I find it very hard to believe that Tesla would go through all the trouble of getting coolant in there if it didn't matter, especially with all the warranties and failures they have had with these.
My assumption is the bypass coolant manifold isn't available because of what is very likely to happen without rotor cooling. It might work for light use but the magnets will likely overheat and lose strength without cooling.

Many shops / people do the wrong thing with good intentions trying to solve a problem, share the fix without long term data, and don't follow up with the unintended consequences of that choice if it goes sideways.

If someone wants to try a coolant bypass, I'd be really interested in how it works out long term.

Installing an ir temp sensor pointed at the magnet area would be a good idea to keep an eye on them. Or some temp indicating paint on the rotor to see how hot it got at the failure point.

[muehlpower]

At least you don't have to worry about the magnets with the LDU. It does not have any.

[howardc64]

If someone wants to try a coolant bypass, I'd be really interested in how it works out long term.

I guess the question is indeed longevity. Would be really nice to have an actual induction motor design engineer to chime in. Or even a non induction motor design engineer.

I wonder if a temp sensor attached on the rotor coolant output line (there is a nice long externally accessible aluminum line to attach a temp sensor to provide an indication how much heat is removed. Wouldn't be as accurate as sensor in the coolant stream. Input coolant temp to LDU is known. Rotor coolant output flow rate can probably be estimated by the ~5mm diameter hole limit on the rotor coolant output + PSI. Combined would be able to estimate removed heat capacity.

Another interesting variable is to change the software to reduce power output limit. Something only Tesla can do. It would probably work fine driving <= 10mph without cooling rotor. How much power reduction is necessary to keep it in a safe zone for longevity? Torque doesn't seem to need much power judging by the dash panel power meter. High power need is on aggressive acceleration (eg 50kw). Then there is regen which can be 30kw on a steep downhill.

[Antti84]

I have a Tesla Model S2014 Rear wheel drive LDU renovation ongoing. I just replaced the bearings and noticed that the rotor shaft is is very worn out by the seal so the shaft is uneven. I tried with speedi sleeve but it did not work- it still leaked.
Therefore I have machined the shaft end to be good and then i have done the hard chrome treatment on the shaft.
I will then have to machine the shaft into right measurement before installing the ring for the vehicle speed sensor ring and the chenmin 3 lip PTFE Seal.

Question:what is the exact tolerances for the shaft end.It should be 30mm but does someone have the exact tolerances?
The vehicle speed sensor ring was 30,2mm in diameter.

Ptfe seal requires Ra value of 0,4. Is this true?

Thanks in advance!

[howardc64]

I have a Tesla Model S2014 Rear wheel drive LDU renovation ongoing. I just replaced the bearings and noticed that the rotor shaft is is very worn out by the seal so the shaft is uneven. I tried with speedi sleeve but it did not work- it still leaked.
Therefore I have machined the shaft end to be good and then i have done the hard chrome treatment on the shaft.
I will then have to machine the shaft into right measurement before installing the ring for the vehicle speed sensor ring and the chenmin 3 lip PTFE Seal.

Question:what is the exact tolerances for the shaft end.It should be 30mm but does someone have the exact tolerances?
The vehicle speed sensor ring was 30,2mm in diameter.

Ptfe seal requires Ra value of 0,4. Is this true?

Thanks in advance!

Speedi-Sleeve and Ceimin triple lip is incompatible

Speedi-Sleeve will not work with Ceimin triple lip since the sleeve doesn't cover surface region required by the 3 lips. Probably excluder lip will sit on the sleeve (if not folded by speedi-sleeve edge during installation) Middle coolant facing lip likely land on the sharp edge of speedi sleeve and get cut. The outer most coolant facing lip will sit on the existing shaft. If shaft surface is poor condition, it will not seal.

30mm SKF speedi sleeve require shaft measurements at 3 different points to be between 29,95 and 30,07. When I measured on original non worn region of the shaft, it is 30mm +- 0.05mm. Therefore, shaft diameter should be 30mm after any surface refinishing effort.

Shaft Surface Finish

AdvancedEMC article suggest 8-16 RMS for liquids

https://advanced-emc.com/selecting-the- ... aft-seals/

This table cross reference RMS and um Ra (8 RMS = 0.2 um Ra 16 RMS = 0.4 um Ra)

https://www.cabww.com/toolkit/surface-r ... conversion

I do see uin Ra and um Ra used often so be careful which one as they are obviously hugely different.

There are more details on what is Ra and RMS measurements on shafts. You might want to read and understand those.

Ceimin Seal Installation

This is really tricky as the outer most coolant facing lip need to stretch 3.5mm to install onto the shaft. PTFE material is hard and this stretching requires a lot of force. The small chamfer at end of shaft is the blind installation tool. Any refinish on the chamfer area require perfect smoothness to avoid cutting the PTFE seal lip during blind install. Any added material from refinish (spray metal or plating) need to carefully evaluate the potential increase diameter of the chamfer tip to see how it affects PTFE seal excluder lip installation. If the tip diameter becomes larger (even by 0.5mm) will likely increase difficulty to install the Ceimin seal without folding the excluder lip.

The Ceimin triple lip seal should be first pre-fitted onto the shaft to stretch the seal lips for awhile. Then removed, mounted into the manifold bore (with RTV) and install the manifold (with RTV) fairly quickly before PTFE seal lip return to its original shape and becomes tighter again. If Ceimin seal is installed with manifold without pre-fitting stretching. It is highly likely the excluder lip will become folded during blind installation.

The GFD seal from @muehlpower doesn't have as much difficulty as Ceimin triple lipped seal. GFD seal doesn't have an energized excluder lip pressing on the shaft. The coolant lip pressure doesn't come from stretching PTFE seal lip. Instead, the PTFE seal surface is on the end of a long mostly horizontal PTFE beam with much greater flexibility. Garter spring is the energized pressure source rather than stretching PTFE lips aggressively to energize sealing lip.

viewtopic.php?p=51663#p51663

Signature link on LDU rebuild website has links to diagrams on how to prefit + install Ceimin PTFE seal.

[howardc64]

UPDATE : Speedi Sleeve + SKF FKM seal leaked at 800 miles

400 miles - dry
800 miles - seal chamber drain tube showed coolant. Drained out ~1mL Impossible to capture everything so call it 2mL leak max.
never drove faster than 80mph on highway

Mid Inverter casing drain tap. No signs of leak
Motor end plate/cap drain/inspection port. Not at lowest point (close) but dry.

Seal chamber drain tube

- zip tied at the end so no coolant got past
- drain tube (5mm nipple and 1/4" OD silicone tubing won't drain completely with gravity as seal chamber has sufficient vacuum. Pulled the speed sensor to relieve vacuum to fully drain. Syringe suction from tube didn't capture more coolant.
- routed down along manifold towards coolant inlet hose. Can inspect via my 19" alloy wheel opening without raising the car + remove belly pan.
- probably need to fashion a higher volume leak drain reservoir

[howardc64]

The coolant delete manifold has been found. Looks like its on the latest Rev U LDU. Unfortunately, can't just buy the redesigned manifold from Tesla

https://teslamotorsclub.com/tmc/posts/7898613/

[muehlpower]

Moodifying an existing manifold should be easy. The tube is difficult to press out, but shortening it up to the retaining ring should work. Then you only need a cover in place of the sealing ring that leaves enough space for the shaft and has a gap to the shortened tube. If the gap becomes too small, a hole can be drilled into the duct below the pipe. I could imagine that omitting the rotor cooling works for lightly loaded engines, but I don't want it for my sport unit!

Plug.png (9.67 KiB) Viewed 31422 times

[howardc64]

Moodifying an existing manifold should be easy. The tube is difficult to press out, but shortening it up to the retaining ring should work. Then you only need a cover in place of the sealing ring that leaves enough space for the shaft and has a gap to the shortened tube. If the gap becomes too small, a hole can be drilled into the duct below the pipe. I could imagine that omitting the rotor cooling works for lightly loaded engines, but I don't want it for my sport unit! Plug.png

I think may also need to solve the coolant feed on top of the manifold. That feed is from a hole marked X in the diagram

Plug.png

So either seal that hole and route coolant externally or cut the external tube and feed it directly from a T junction in the coolant inlet hose.

If the coolant circuit through the 4.5mm hole just blend with the much higher volume out of stator inside the gearbox casing, then maybe this small circuit is unnecessary. But I don't know the circuit inside gearbox casing.

Would be good to have induction motor engineer evaluate the cooling rather than just assume Tesla's design mod is robust.

Also highly desirable to have additional confirmation Rev U LDU comes with coolant delete manifold for both base+sport LDU. Tesla provides no information so best to collect statistics.

[asavage]

If the coolant circuit through the 4.5mm hole just blend with the much higher volume out of stator inside the gearbox casing, then maybe this small circuit is unnecessary.

It's a reasonable guess that the purpose of that external tube is for air bleeding the circuit (connects to high point of the LDU, moves trapped gasses on), and that it doesn't contribute substantially to thermal transfer.

[muehlpower]

There is no reason to run anything externally. The water comes from below, flows around the cover and exits at the top as before. Only the space for the gap could become tight, so it makes sense to drill a hole in the channel. The hole would then be where the word "seal" is in your picture.

[howardc64]

There is no reason to run anything externally. The water comes from below, flows around the cover and exits at the top as before. Only the space for the gap could become tight, so it makes sense to drill a hole in the channel. The hole would then be where the word "seal" is in your picture.

Ok I understand your proposal. Very nice!

- Remove manifold rotor coolant tube
- Custom cap to insert into coolant seal bore to close off the chamber, cap shaped to avoid touching rotor shaft and provide enough gap to let coolant flow into the closed off chamber and exit top of the manifold.

Internal routing solution would be best as there isn't much clearance externally when removing/installing LDU into Model S rear subframe.

I wonder if any LDU software update is necessary...

[muehlpower]

In case anyone has the idea of removing the pipe. I have tried it. It seems to be pressed in and secured with Loctite 271 (high-strength). I welded mine at the end to get it tight, as the bore was very damaged.