Tesla Large Drive Unit (LDU) Motor Teardown and maintenance

[howardc64]

DELETED

[asavage]

I used a wiring harness sleeve to pull the temp sensor wires through the tunnel, worked very well.

How did you prevent coolant from traveling through those sensor wires in future?

How do you guys purge the cooling system? Is it necessary to fill the motor before connecting the cooling system in the car? Or will it purge itself while driving?

I intend to fill mine -- when I have to DIY -- using an air evacuation tool that I use for all my other vehicles. It uses shop compressed air to generate a vacuum on the cooling system.

viewtopic.php?p=44254#p44254

I used to use an Airlift II, but I recently obtained a Snap-On RADKITPLUSA Cooling System Vacuum Filler set, thinking it might be an upgrade.

I've also got (for my Toyota RAV4 EV) a cracked install of Tesla Powertrain Diagnostics, which will command the pumps to run and valve to cycle, running the pumps for 50 minutes or something, which is the official method for my vehicle that came with a factory-installed customized version of the Tesla LDU (runs backwards, has electric Park mode, etc.).

Anecdotally, another RAV4 EV owner did not use either of those methods, and shortly after reinstalling his LDU, he received an overtemp warning at the dash, and found the reservoir low. I assume that running it around for a few minutes and topping up -- lather, rinse, repeat -- may get you to a stable situation, but I do not know if there are high spots in the Model S cooling system that can trap air. That's the advantage (and the whole purpose) of the vacuum tool: no air pockets are left, unless a valve is closed during evacuation of the air.

[WimV]

I don't have a vacuum tool, so I'm just going to try to let it purge itself. There are temp sensors everywhere, so I don't think anything can go wrong when there's an air lock.

I drained about 2 litres, so I’ll be driving and charging slowly till I’ve topped the reservoir with 2 litres.

[WimV]

Immediately after connecting the 12 V battery, the coolant pumps started running and the coolant level was lowering. I added about a litre before I started driving. I drove 20 minutes, the level was lowered again. Added again about a litre, so level should be the same as before.

Until now, everything looks fine

[WimV]

How did you prevent coolant from traveling through those sensor wires in future?

It looked like the white sleeve is embedded in the stator, so I don't expect water to leak in the sleeve.

If my seal is going to leak again in the future, I'm thinking about opening up the reluctor housing. There should be room enough to mill some slots and holes. Is there any reason why not to do this? The reluctor wheel and sensor are robust enough, they don't care about some dust.
If it would leak, it would just leak to the outside.

[howardc64]

DELETED

[Johan]

It looked like the white sleeve is embedded in the stator, so I don't expect water to leak in the sleeve.

Yes, the brown stator potting resin encapsulates the white sleeve in my LDU too.

[asavage]

Thanks for clarifying that stray coolant should not be able to travel along with the temp sensing harness.

[WimV]

Worried reluctor chamber drain tap at bottom not enough? With that drain, won't get a pool of coolant for the reluctor wheel to agitate and aerosolize? I suppose no pool is great step with drain tap. Your thought if we can get to no collateral damage from leak so no need to repair leaky seal. Even better.

Without knowing how and when the leak occurs. Kind of hard to judge what happens next. I was also curious about reluctor wheel chamber air circulation pattern while running. What happens when minute drop seeps out at say top of the seal during high speed forward rotation (seems like it could get aerosolized right away before getting to the drain), or slow speed reverse, or stationary parked. And if the minute leak sticks to the walls and not vacated by gravity right away during parking/reverse. What happens when reluctor wheel starts spinning up at high speeds with these small droplets on the walls? These thoughts were also motivated by.. just let it leak if no collateral damage (if possible)

I suppose more opening exposes the sealed hybrid bearing and excluder lip as well. But that lip might be making PTFE residue anyways?

Yeah, it probably won't form a puddle, but I do think it will be wet inside when the seal leaks.

I’ve had and tractor engine with a cylinder block rusted stuck because there was a little bit of water in the exhaust damper after standing still for a few days. During the day, the exhaust warmed faster than the cylinder block. Causing water to evaporate in the exhaust, travel through the exhaust and valves to condense in the cylinder block.

I wouldn’t be surprised if some moisture still manages to travel trough the stator.

[howardc64]

DELETED

[asavage]

Yes, I mentioned water pump cartridge seals as being in use for automotive water pumps forever (upthread? one of the other threads?).

At one time, I had to rebuild a Hillman 1390cc water pump (see my avatar), around 40 years ago, and OEM parts were NLA even then. I sourced a cartridge ceramic face seal and re-faced the back of the impeller to fit. Fitted a new WP bearing cartridge (a separate unit).

I can only assume the ceramic face seals won't work with the RPM requirement of the rotor for some reason, but I don't have the engineering to know. Also, they take up a lot of axial space (for the tension spring and bellows).

[howardc64]

DELETED

[howardc64]

DELETED

[mjc506]

I used to sell components for the heavy commercial and industrial refrigeration industry - massive surge vessels, valves, pumps etc for fairly nasty refrigerants like ammonia, R22 (before it was banned), Co2...

Valve spindles were sealed with two o-rings, a gap between them filled with refrigeration oil. Think thinking being that any leaking refrigerant had to push past the first o-ring, and also displace a volume of thick oil past the second o-ring. Very effective seal, and as a backup you have a valve cap over the top of the whole lot with a third o-ring - any leaks through the stem seal would become obvious as you undid the cap, exposing a hole drilled through the theads, which would make a whistling noise if any pressured gas would escape.

For higher pressures (over 40bar, for CO2) a different stem seal was needed(!) - two PTFE cones that fit inside each other, pressed together with a very strong spring, plus the pressure of the refrigerant (the spring was required for pressures below ambient). Again, the valve cap was a backup.

Despite sealing well enough that you couldn't smell ammonia through them, they weren't 100% effective (after several months, you'd still get a 'whistle' when removing the valve cap), and are of course entirely unsuitable for this application (only designed for very low turning speeds, and quite bulky)...

The pumps we sold were of two types. There was an 'open' type, where the pump and motor were separate. These needed shaft seals to seal against the refrigerant, and experienced quite high pressures. PTFE rings (similar to the waterpump layout about), springs, and actual fibrous 'packing', so an effective seal, but it was accepted that there would be some 'seepage' (similar to auto waterpumps - they all have a drain hole, and all say 'some seepage is normal').

There was of course a desire to develop a pump that was completely refrigerant tight so could be used inside buildings (you wouldn't want a pump that seeped ammonia inside the same building that stored your food... so 'traditional' solutions were to have a separate plant building and pipe the refrigerant around the site. Caused problems with insulation and energy loss, and everything was outdoors (and had to be well ventilated) so did deteriorate relatively quickly). The engineers spent years on testing different designs. Low speeds were easy, the PTFE cone design was good for up to around 100rpm, but much above that the sealing surfaces basically overheated at a very small scale - they would appear polished, but the material basically became porous through that fraction-millimeter layer (so it wasn't just a case of 'seeps while above xxx rpm', the seal would become degraded once the surface speed exceeded some threshold). The two-o-rings-with-oil-between design nearly worked (some refrigerant would get through the first o-ring, displacing a small volume of oil through the second, but became so diluted in the oil it was barely detectable) but needed frequent replacement, which then of course opened the system. I suspect it would be no good at all at the very high surface speeds experienced in the LDU - melty o-rings! These topped out around 3600rpm, and were usually ran slower.

In the end, the only way they managed it was to design a hermetic pump, where the shaft itself wasn't sealed, but the whole pump and motor were within a complete sealed case (which was disassembleable for maintenance, but static seals are relatively easy!). The pump was expensive, as the motor had to be custom made to cope with various refrigerants moving through it, and actually used the refrigerant to cool the rotor, stator and bearings. But it didn't leak refrigerant! Sold very well...

That's a lot of text to say I think you'll always get some seepage through any moving seal, it's just a case of finding a method of dealing with it to some workable level. The drains would hopefully help, but probably won't drain away single droplets (surface tension stronger than gravity). Hopefully a single droplet won't do any damage, and that'll be 'good enough'. Eventually that shaft seal will wear enough that it's letting through more than one droplet, at which points the drains will really start working and give time to notice and repair the problem. The only 'forever' solution I can think of is finding some coolant that won't hurt the rotor/stator/inverter and just flood the motor (or sealing the path between inverter and motor, and just flooding the motor) but this will be chemically difficult, and a rotor spinning in a liquid won't be as happy as one spinning in gas...

A possible alternative, however... Rather than trying to seal completely, why not try making it leak in the other direction? Pressurise the 'dry' side to some low amount about the coolant loop pressure, and the worst you'll have to do with is small bubbles in the coolant, which should eventually end up in a reservior (or more likely, just get absorbed into the coolant) This will require the motor/inverter to be air tight, and a 'vent' to be installed somewhere (probably near the seal), but that's probably less difficult than trying to keep that rotating seal perfect.

[howardc64]

DELETED

[asavage]

Good story, MJC.

O-rings are limited to low shaft RPM (ft/min.) in most applications; they'll neither seal well nor wear well. Decent service life for low-movement assemblies.

---

Perhaps maintain a very slight vacuum on the entire cooling system? I wonder how leaky the system as a whole is, under a slight vacuum. I wonder if I have any decent small vacuum pumps left around here (besides my Vacu-Vin, which I won't sacrifice for Science).

[howardc64]

DELETED

[howardc64]

DELETED

[beastman]

Howard, are you thinking the debris are causing the seal failures at this point? It would seem that the triple lip might hold back some of the debris floating around the du, including the dirt and grease. even though i dont appear to have coolant failure, this grease is quite thick from the bearings and i would think this grease would prevent the seal from doing its job. Makes me think the drains once again would allow us to get contaminants out of the system and prevent those small pores on the seal from getting clogged and misshape the seal. maybe all we need is a way to clean or flush contaminants annuallly and clean the speed sensor too as it catches some grease... i very much appreciate your efforts on this and think drains are the solution w a detector as the triple lip just gives us some failure redundancy for a while. I still also believe that any high voltage transients may use the coolant seal path as a discharge point further electrolysing the coolant and possible damage to the lip over time since hybrid ceramic bearings are no longer going to be the preferred discharge path.

[howardc64]

DELETED

[howardc64]

DELETED

[beastman]

those last few pics looks like elecyrolyzed coolant to me. the minerals appear to make paths and lots of pitting like water spots. if you can get a good multimeter, you could test some spots for conductivity under the scope. the dry side vs wet side really tells a story. i would think an aegis ring may help prevent this as well as using non conductive coolant if we can get one that still works w some contaminants.

[howardc64]

DELETED

[howardc64]

DELETED

[asavage]

That EKK vid is fun. The seal's moveable face floats away from the fixed face at higher RPM to achieve frictionless sealing. Neat trick, and not at all how the legacy water pump seals I worked with 40 years ago worked; they had two smooth ceramic faces and AFAICT never separated, regardless of RPM.