Tesla Large Drive Unit (LDU) Motor Teardown and maintenance

[howardc64]

DELETED

[asavage]

Shipping an LDU rotor will require a custom WOOD packing case be made. IIRC, it's ~90 lbs, and during shipment the axial load should be absorbed by the wood case, on the copper-colored rotor end that's around 9" dia, and not on the end of the 30mm shaft where the seal seats.

Even I could knock out a shipping container from scrap wood in a couple of hours, and I'm miserably slow at this kind of thing, but it needs to be done. You can't just throw this thing in a cardboard box with bubble wrap.

[howardc64]

DELETED

[Johan]

My earlier statement that my single lip Tesla seal was "worn out" was probably incorrect. Instead, it probably already had a smaller thickness from the start (in the reagion where it contacts the shaft) due strains induced by manufacturing and installation. I explain it in more detail on Youtube:

[jsimonkeller]

An update on rotor repair option

As a side note, they asked how repairable is the inverter in case the coolant leak stops the car (this will be the highest volume scenario) This brings up another major challenge to jump start LDU rebuild capability worldwide. Inverter contains Tesla IP control and IGBT boards which someone else probably can't legally make. Harness, bus bars, plastic/metal structural parts are probably not a problem.

Just as a side note on my INVERTER repair, in my case it turned out solely to be a HARNESS issue. As you may recall, my coolant leak made it into the inverter and here are some photos of the condition of the harness before the repairs took place.

I purchased a new 23 pin harness from ZERO EV, since they make one, but no one sells the harness that attaches to the gate drive board on the three phase blocks, so I thoroughly cleaned that harness and when things still did not work, I assumed my damage was in the DU card initially. I replaced it with a new one and that did not solve my issue.

I then moved onto the phase blocks and assumed that was issue based on these scary looking RED LEDs.

With the help of the amazing Boxster EV, I was able to obtain an untested replacement inverter from him quickly. THANK YOU AGAIN! When it arrived, I disassembled it into the three separate phase blocks and brought them to shop. Without HV connected and just 12V power and the car off, I unplugged the harness from the alleged bad gate drive board and plugged it into each of the three replacement phase blocks and in each case, those RED LEDs lit up again. In order to troubleshoot further, I unplugged the harness from one of the other two phase blocks without RED LEDs and when I plugged those into the phase block that had been lighting it up, it did NOT and the same went for the replacement blocks.

At this point, I surmised that the two red LEDs were supposed to be lit up when the car was off with 12V supplied power (sorry if this is obvious to all of you experienced members). When the car starts up, those RED LEDs disappear. Therefore, it turns out that the coolant exposure damaged the phase block harness and two of the blocks were not getting any power.

Luckily, the replacement inverter from Boxster EV came with a harness and when I put that harness into play, all three of the existing phase blocks lit up with 12V and car off and all went off with 12V and car on. At this point, I tested the inverter and it went into RUN mode for the first time in 2 months. I drove the car home from the shop and everything seems to be operating as it should now.

All of this to say that they PCB cards seem to be heartier than we might think and that the coolant exposure to the wiring harness did not short anything out on my inverter and just kept it from powering on. It might be a project at some point in time to fabricate these harnesses since they are not that complicated to replicate and this could be the most cost efficient solution for a coolant leak into the inverter and a first line of testing things after a coolant leak and inverter failure.

[howardc64]

DELETED

[howardc64]

DELETED

[muehlpower]

As I have already written, I am in negotiations with a seal manufacturer in Germany. He sent me an interesting draft. I have pointed out that a Speedi Sleeve may be used. He also takes into account the slightly larger diameter and the limited length.
He also has reservations about using Loctite because of the poor heat dissipation. I did some googling and came across this glue.
https://www.sepa-europe.com/wp-content/ ... 000010.pdf

[Johan]

@muehlpower: I don't have time for this yet I can't let go : Very interesting design! It looks like the lip is CNC machined and definitely not formed using a mandrel. Some questions and remarks in random order (I may be dead wrong of course):

1) Does GFD have experience using this design (and @10k RPM with coolant) or is it highly experimental?

2) The left "primary" lip is too close to the shaft end (at least for my LDU which I measured - see my drawing below). Now it would contact the "shaft end chamfer" and probably leak! It must be moved at least 1mm away from the shaft end (relative to the current position in the drawing) in order to have at least 1mm of clearance from that "shaft end chamfer". Which is maybe still not conservative enough. Simultaneously, the excluder lip is allowed to be positioned a maximum roughly 1mm extra away from the shaft end (relative again), so that it has a (reduced) 2mm clearance from the "shaft reluctor wheel chamfer". Why 2mm and not also 1mm? Because my drawing shows the motor in the "cold" configuration. Once heated up, differential thermal expansion can move the shaft roughly 0.5mm deeper into the manifold because the rotor can get hotter than the stator+housing, and also lateral accellerations can "bounce" the rotor roughly 0.5 to 1mm in axial direction (because the rotor is kept in place by a spring and floating bearing on the manifold side), which reduces the clearance to maybe only 0.5 to 1mm. This design is very tight! Seems like Tesla had very little available space ("bauraum" in german)

2) The garter spring seems to be able to "slip off" the lip easily, so maybe add some more material to "cradle" it (keep in place). Especially because I think that the spring pressure must be relatively small (compared to the classic elastomer lips) because too high pressure results in too much frictional forces and thus heat development.

3) Does the steel ring have an interference fit? It shows 55mm nominal only and no tolerance.

4) The excluder lip should ideally maybe have nearly zero pressure on the shaft, maybe even a very tiny gap like the SKF FKM seals, so maybe reduce the interference.

5) What price range?

6) Maybe increase the primary lip contact area (width) to 2mm in order to reduce contact pressure. This is also a benefit from some formed lips (like the Ceimin lips). Just an observation based on Stuttgart PhD work on PTFE seals.

[howardc64]

DELETED

[WimV]

Just found these on AliExpress:
https://a.aliexpress.com/_EvS09oh

Single, double and tripple lip (CSL, CDL and CTL).

Seller does't seem to have a lot of knowledge. Says that seals are for 5.000 rpm. According to the discription surface speed is 25 m/s (+- 16.000 rpm). He answers in Chinese (automatic translated), so that doesn't make it any easier.

[howardc64]

DELETED

[asavage]

Salesmen and politicians only lie when their lips move.

[muehlpower]

I received the final proposal from XXX. It should fit the original shaft and also work with the speedy sleeve. I also have prices, depend very much on the quantity purchased, but will be at least €80. Do you think there are enough buyers who are willing to do this?

[asavage]

For estimating purposes, I'd purchase one.

This would be a "for testing purposes"/"guinea pig" use, and I'd be willing to invest <=€180. If it had a proven track record of being more durable than other options, I'd be willing to pay more.

[WimV]

I'm also interested in buying 2 or 3 for testing.

[howardc64]

DELETED

[howardc64]

DELETED

[asavage]

Step2 is possible with DIY LDU rebuilders. Custom EVs such as @BoxsterEV is highly valuable with more accessible LDUs compared to the Tesla Model S LDU rebuilders. However, close and precise test monitoring is challenging.

If a goal is to be a provider of a tested-and-guaranteed LDU "reseal kit" or service, then -- maybe -- a USD$10k investment is warranted. For DIY-ers (ie not a repair service or a "reseal kit" reseller), having a seal with some positive track record is sufficient. To achieve that point, some pioneers may have to suffer a few arrows in their backs (if you're not familiar with my tortured version of the aphorism, read here).

DIY-ers will gladly play pioneer, if the alternatives are sufficiently awful . . . or untrustworthy (see below).

Installation Difficulty

PTFE rotary seal is much more challenging to install properly than elastomer seals (NBR, FKM)

[ . . . ]

GFD's latest design has no excluder so this shouldn't be an issue for this seal.

Installation will be uncomplicated without the excluder, because the direction of shaft install naturally deflects the lip in the normal direction.

Competition

AliExpress Ceimin seal

Western users naturally distrust this seal. It is from China via AliExpress.

The Ceimin seal might be fine. As more people install them and rack up kms/miles, some people may report success and newer DIY-ers might be more comfortable using them.

But.

A significant problem that I've personally run afoul of is that a "Seal Part No. 123456-01" that you receive today is not the same seal you received three months ago, and this seems to be a problem predominantly with China-(etc)-sources parts. IOW, you cannot trust that a seal you buy today that has the same part No. as a seal someone installed last year, will be the same seal (or made in the same plant, or by workers with the same experience, or same quality controll process, etc.).

That's why I will trust a seal sourced from eg a European mfgr over an eg Asian source: repeatability of results.

PTFE rotary seal only has 3 design parameters.

#1 is PTFE blend. Need to be as hard as possible (for longevity) but softer than the shaft (avoid shaft damage)

I'm not certain that shaft wear is really an issue. Depending on the compliance of the PTFE material (or its substrate), either the natural seal characteristics, or the garter spring-aided design, may adequately compensate for, say, 0.3mm shaft wear (or sleeve wear).

Put another way, if we assume re-sleeving at every seal change, do we care if the sleeve is worn through completely by the time the seal is due to be replaced?

#3 is how to reduce leakage. I conclude micro structure pumping aid is unavailable (and maybe ineffective after coolant silicate crystallization clogging the structures) so I think the only solution is # of lips. @asavage distrust this kind of design. However, SKF, Parker, Saint-Gobain all show simple 2x media lip + excluder lip PTFE rotary seal designs in their design catalog . . .

By definition, the excluder lip is not a liquid seal: it excludes particulates and is not expected to seal against liquid, which means the seal's main lip has at least incidental if not total wetting.

In a three-lip design, the "last" lip is deprived of that wetting, so it has to operate at ~10k RPM with only whatever material characteristics it has to prevent it from accelerated wear, and that's what I mistrust: a seal that is both supposed to prevent liquid from getting past it, and also operates dry for some very significant portion of its lifecycle.

There's a reason why you can't run a typical water pump (a type that uses a shaft seal) dry: that quickly destroys the seal. And that's what I see in the triple-lip LDU seal: a seal that runs dry (or, perhaps, with some installation "lubricant") until the main/middle lip leaks, and then that last lip is expected to be a backstop seal? Yeah, I mistrust it.

But I am not any kind of a seal engineer; I've merely had to deal with a lot of failed seals in my life (and I did work in an Engineering department in the 90's, though not as an Engineer).

How to make longevity history

I guess without $10k investment, its only possible to start with LDU rebuilders. I guess initial quantity 10 for €100 each can probably sell out quickly. However, seal must be installed and used for valuable testing. If just collected and uninstalled, there is no longevity data benefit. But even if just purchased and collected, it helps initial pilot manufacturing quantity and therefore price.

Agreed. I will be in that camp: buying now to aid in getting some into hands at a more-reasonable group-buy-like price, but not installing until I have to, which might be next week or four years hence.

15-20km data might arrive in 1-2 years. Any failed seal (or removed for inspection) need to be carefully extracted from the manifold and examined in detail.

Let's not forget that some seals seem to fail when the LDU is not being exercised. Data from infrequent drivers will be just as valuable as raw kms/miles driven on a seal.

[howardc64]

DELETED

[howardc64]

DELETED

[SuperV8]

Academic papers and seal manufacturers say all seals leak. The key is the acceptable leak rate. Perhaps any seal we put in will leak at some rate. Anyway, Ceimin seal at 2k miles has no coolant drops on speed sensor.

This is key:

At some point any rotary seal will leak.
There are no rotary seals in any ICE I can think of where if the seal leaks if causes total engine failure.

In my humble any LDU seal fix needs to incorporate seal leakage mitigation - so when it leaks again the coolant is safely diverted away from the electronics into a catch can or back into the cooling system.

On an side note - but related to LDU: a work friend recently had his LDU replaced (Model S P85D - I think 2016my?) - luckily under warranty (the garage who sold him the car covered it!) - Tesla said it was due to water ingress - apparently from the large power cable's insulation/seal braking down and allowing water in the inverter. Something else to ensure is well sealed and the cable's are in good condition.

[muehlpower]

I agree 100% with SuperV8. I would definitely drill a drain hole, on the one hand to notice a leak and on the other hand to prevent major damage. One of my motors also had water ingress on the inverter side, clearly due to corrosion on the data connector.

[howardc64]

DELETED

[howardc64]

DELETED