LDU rotor cooling bypass ("Coolant Delete") modification

Toyota Rav4 EV Forum

Help Support Toyota Rav4 EV Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

asavage

Well-known member
Supporting Member
Joined
Aug 7, 2017
Messages
619
Location
Oak Harbor, Wash.
Summary: I had an aluminum plug machined, modified the OEM LDU coolant manifold by cutting off the tube that runs into the hollow rotor shaft, milled a bypass channel, and pressed in a plug. I anticipate no more coolant leaks in my LDU.

Background:
The Tesla Large Drive Units ("LDU") of the original Model S (and derivatives such as the GEN2 RAV4 EV and the MB original Electric Drive (later: B250e)) went through many, many internal revisions from its inception in Jun2012. Many of the early issues were ironed out over the course of the first four years, culminating in a relatively reliable "Revision Q" in 2016.

However, there remains one issue with the LDU that hasn't been adequately addressed. The rotor is cooled via a water-based glycol mix coolant, via its hollow shaft. Johan ("DIY EV Guy" on YouTube), a RAV4 EV owner, has done us a service by making a set of vids that go into aspects of the LDU's "water" cooling system; if you're interested in the details I can recommend watching a couple of them, esp. this one.

The Tesla design for the rotor cooling involves a lip seal on one end of the hollow rotor's shaft. That seal has had at least three revisions, and none have proven reliable. In addition, Tesla provided no "weep hole", a standard engineering practice for water pump shaft seal designs, which Tesla design mirrors.

The result of this rotor seal leaking is that "water" leaks into a sealed chamber, works its way through the outer rotor bearing (eventually leading to its destruction . . . if run long enough), then to the stator cavity where it compromises the electrical isolation resistance of the stator. If the LDU is driven further (if the isolation resistance issue doesn't trigger a BMS fault which prevents the HV contactors from activating), more coolant makes its way through a "tunnel" at the bottom of the gearcase, where the three HV phase leads run, and makes its way to the inverter electronics. If you're lucky, it fouls some wiring harness connectors and the monitoring system prevents operation, triggering many -- often spurious -- Alert codes. If you're unlucky, it wipes out the inverter electronics and renders the LDU salvage (for parts only).

This rotor seal leak issue has been well known for many, many years. If you're an owner of an LDU, until recently your options were limited, among them:

  • You could check for blue coolant leaking into the speed sensor chamber on a regular basis, perhaps once or twice a year.
  • You could add a weep hole so when coolant leaks it would exit the LDU via the hole to the outside, rather than forcing its way through the rotor bearing.
  • You could install a water monitor instead/in addition to the weep hole, to light a lamp when coolant was detected.

But, you couldn't actually obtain a reliable sealing situation. LDUs remanufactured by Tesla subcontractors have had a very spotty record of longevity, ranging from hundreds of miles to ??? but they always leak eventually.

This has become a significant issue for Model S, MB B250e, and RAV4 EV owners, now that they have some time on them. The LDUs are time bombs. They can leak if you drive them a lot; they can leak if they stay in your garage.

Tesla has apparently given up on keeping these units sealed, as they had been shipping their own version of "coolant delete" on their remans. There is an cast aluminum manifold bolted to the very end of the rotor case cap, and this "coolant manifold" routes coolant three places:
  • Down to the stator case (further: cools the inverter electronics)
  • Through the hollow rotor (and back)
  • To a "flyover tube" external to the LDU, which routes to a gearcase heat exchanger
Below, Tesla's new coolant manifold. Coolant enters at the red plastic plug, and travels both down to the stator (& inverter) and up to the flyover tube:

1714775117376.png


Because Tesla is Tesla, and they sell no parts to the LDU, you & I cannot buy that newer coolant manifold :(

QC Charge sells a bespoke replacement coolant manifold for $770 (with tax).

1715375861983.png

Revolt is now selling their version of what QC Charge proposed last year; theirs also deletes the flyover tube to the gearcase heat exchanger (opinions vary as to the necessity of that heat exchanger), and theirs is supposedly around $550 (after tax).

276-fb2acab060851bfbface89a7827c2df3.data


So, there appears to -- finally -- be a more permanent solution to rotor seal leaks: don't cool the rotor with liquid!

Other ideas that had been floated:
  • Swap out the water-based glycol coolant for a non-water-based one (ie "waterless coolant" or a light oil)
  • Pull a slight vacuum -- and maintain it on the coolant system, to prevent coolant from leaking, via pressure differential

Naturally, there are DIY approaches to modifying the old coolant manifold to bypass rotor cooling. There's 14-page thread over at teslamotorsclub.com with a lot of implementations (it's a long read). The approaches boil down to:
  • Drill a hole from the flyover tube down to below the reluctor chamber and run a metal tube from top to bottom.
  • Weld up the internal passage to the rotor, either retaining or abandoning the flyover tube flow
  • Same as above, using epoxy
  • Replace the rotor seal with a "cup" that presses in place of the seal, preventing flow to/from rotor but maintaining flyover tube flow
    • Aluminum cup
    • Formed SS cup
    • [20240509: after I was done doing the following steps I outline, some came up with this new idea] Cut the manifold "bottom" off that feeds the stator, replace the lower section with a hose barb and adapter plate, cap the upper half where you cut it (no flyover tube feed to gearbox).
    • Both solutions require cutting and milling (or merely grinding) the OEM coolant manifold to remove the rotor tube and to provide as coolant bypass path, and this is destructive and cannot be reversed . . . and Tesla sells no parts of the LDU. So you kinda have to commit.
---
It should be mentioned that -- in the RAV4, but not in the Model S -- you can R&R the coolant manifold without removing the LDU from the car! If you have a dry/non-leaking/no-symptom RAV4, you could do a "coolant delete" upgrade yourself for less than $200!

I had purchased my 2014 with 49k miles on it, in Aug2017. At 59k mi. I had the LDU replaced with a Reman unit. It now has 119,300 mi., so I got almost 50k on my Tesla remanufactured LDU before I saw the Blue Drops of Death. I had just experienced a "Check EV System" dash alert, so I expected trouble.

I removed my LDU and was surprised and lucky to have found no traces of blue coolant in the "tunnel" or the inverter side; the coolant -- and attendent corrosion -- was confined to the stator cavity & rotor area.

The assumption is that, if Tesla is doing a rotor coolant bypass modification for reman LDUs, "anybody" can. It's unknown whether there are accompanying firmware modifications to Model S vehicles to accommodate this new Tesla "Coolant Delete" manifold. Opinions vary; my thought is that my RAV4 EV does not tow, and rarely sees full power for more than a few seconds, therefore for my uses, air cooling the rotor may be OK.

Leaving aside the partial disassembly of the LDU, cleaning, testing, reasembly and sealing, below shows how I implemented the "aluminum cup" rotor cooling bypass method on my RAV4 EV's LDU coolant manifold.

View with the left plastic side panel removed, and the large underbody plastic panel off:

IMG_7629.jpg

IMG_7630.jpg

IMG_7643.jpg

IMG_7645.jpg



(cont'd next msg)
 
Last edited:
Stock, unmodified OEM coolant manifold. The tube must be removed; it's aluminum and tapered somewhat:
IMG_020180.jpg


Seal removed. The tube has a steel circlip that looks as if it holds the tube in place (it doesn't). It has to be removed as well:
IMG_020181-1.jpg


I used a hacksaw to cut most of the tube off, then did a bit of milling play. I only took off tube material and "kissed" the casting; it doesn't need to be taken down any further:
IMG_020190.jpg

IMG_020192.jpg

The installed plug comes pretty darn close to the casting when installed, hence the need for a coolant bypass channel for the flyover tube. Cutting the channel can be done with a bandfile, or even a Dremel, if you want. Looks don't count here, you just need to remove metal down to the base of where the tube exits the casting, then cut some kind of channel so that when the plug is pressed in, it doesn't completely obscure the coolant outlet.
IMG_020193.jpg

I had a cup plug made to spec (based on a concept by philip295 in Feb2024). First, I bought some 6061 RB from eBay [chuckle]:
IMG_020229-1.jpg

Then handed that slug to my lathe guy, along with a slaved-over diagram, who turned it into this (55mm + .003" oversize version shown):
IMG_7748-1.jpg


I built a pressure-test jig, using some gasket material, some 1/4" steel flatbar, some tapered rubber plugs, a cheap c-clamp that I had to buy, a very cheap bar clamp that I own, and some more-expensive brass fittings, including a Schraeder valve to admit pressurized air (RV water system blow-down adapter (GHT), adapted to NPT).

The first plug I'd had made was .003" oversize of 55mm. I tested it, fitted dry, ~200 lbs press force.
IMG_7777-1.jpgIMG_7778d.jpg

(cont'd next msg.)
 

Attachments

  • IMG_7780d.jpg
    IMG_7780d.jpg
    171.2 KB · Views: 2
Last edited:
IMG_7783.jpg

IMG_7782.jpg

IMG_7780.jpg

The .003" oversize plug leaked badly at 20 PSI. I wanted to try dry fit first, because testing. I'm certain it would be fine if I had used any kind of sealant, but I also didn't like the interference fit at .003" oversize, so I had a second plug made, .005" oversize, and that one, fitted with Loctite 640, pressed in place and left to cure 14 hours, passed a test at 29+ PSI for two hours. It's now installed in my car; I have only ~25 mi. on it so far.

I paid ~$190 to have the two plugs machined (two different setups, costs more). I bought three 6061 slugs from ebay, about $32.

Here's a rendering from eMachineshop.com's free CAD software. If I wanted to have a batch of (50) made, the cost would be under $25 per, using eMachineshop.com . But I'm not really interested in dealing as a distributor, I think.

LDU_Coolant_Manifold_Plug_03-1b.png

LDU_Coolant_Manifold_Plug_Sketch_20240408a.png

I don't have the corner radii or chamfer specs on this sketch (though I did spec them in software), but you can pick "any" radii and chamfer you like; it's not critical.

My lathe guy won't do production runs, so the two I had made are pretty much all I can get via him, but any shop that handles small parts can do this. I contemplated buying a used lathe and making them myself, but I can ill afford the space for a bulky tool that only gets used a couple of times/year, and while I waffled a bit, I decided to let someone else make these for me instead.

I bought 2.25" 6061 round bar stock, 2" lengths, via eBay:
Screenshot from 2024-05-04 08-59-53.png
 
Last edited:
MrMoparMan has come up with perhaps an even easier "coolant delete" method, if you're willing to forgo the flyover tube coolant feed to the top of the gearbox heat exchanger. It's four pieces, all custom, but there's a rumor that he'll sell them. Its chief advantage is that (on the Model S) it doesn't required dropping the entire rear subframe and extracting the LDU, in order to do it this way. And it requires only some dexterity and a hacksaw (no machining):

MrMoparMan said:
. . . eliminated cooling to the rotor and diff by cutting off the original cooling manifold and just gave full flow to the stator. I removed the coolant tube and capped off either end . . . the 3D printed caps work perfectly . . . I thought about routing coolant back to the tube, but I cannot come up with any benefit to bleeding off some coolant to the top of the diff heat exchanger.

Custom parts required:
  • Welded: Alum hose barb welded to a plate: an adapter to route coolant to the stator (and inverter);
  • 3D-printed: A rectangular plastic cap to keep debris from traveling up the old coolant manifold and to the rotor;
  • 3D-printed: two block-offs for both ends of the flyover tube, which is removed.

1715398716373.png

1715398941059.png

1715398960413.png

1715398986660.png

1715399018337.png

1715399036421.png

If you have a Model S/X and want to reduce your labor costs substantially, and are willing to forgo coolant flow to the gearbox heat exchanger, this is a very easy way to prevent future rotor seal leaks.

If you've already got coolant in your stator cavity, you'll have to pull the LDU anyway, so I believe one of the other solutions would be a better call (IMO).
 
View attachment 382

View attachment 384

View attachment 383

The .003" oversize plug leaked badly at 20 PSI. I wanted to try dry fit first, because testing. I'm certain it would be fine if I had used any kind of sealant, but I also didn't like the interference fit at .003" oversize, so I had a second plug made, .005" oversize, and that one, fitted with Loctite 640, pressed in place and left to cure 14 hours, passed a test at 29+ PSI for two hours. It's now installed in my car; I have only ~25 mi. on it so far.

I paid ~$190 to have the two plugs machined (two different setups, costs more). I bought three 6061 slugs from ebay, about $32.

Here's a rendering from eMachineshop.com's free CAD software. If I wanted to have a batch of (50) made, the cost would be under $25 per, using eMachineshop.com . But I'm not really interested in dealing as a distributor, I think.

View attachment 341

View attachment 342

I don't have the corner radii or chamfer specs on this sketch (though I did spec them in software), but you can pick "any" radii and chamfer you like; it's not critical.

My lathe guy won't do production runs, so the two I had made are pretty much all I can get via him, but any shop that handles small parts can do this. I contemplated buying a used lathe and making them myself, but I can ill afford the space for a bulky tool that only gets used a couple of times/year, and while I waffled a bit, I decided to let someone else make these for me instead.

I bought 2.25" 6061 round bar stock, 2" lengths, via eBay:
View attachment 343
Your napkin sketch, are these the final dimension you used?
10mm + 8.3mm outer diameter 50.18mm ? (I found another drawing with 8mm deep instead of 10mm and now got worried that might interfere with rotor?)
I now have a guy working on my part so just would like to make sure we make it correct. :)
Thank you for your awesome work.
 
1) No, 2.170" is the final oversize of .005", which = 55.12mm. Another person used 55.18mm which is an awful lot of interference fit on an aluminum casting of this size, but I was convinced to try 55.07mm first (.003") and that was not enough for a truly tight fit IMO, and then I had 55.12mm (.005") fabricated, which is plenty of interference fit, IMO. I still feel that 55.18 (.0075") is too much.

2) 10mm deep, as shown, yes. Early sketches that Howard & I batted about show 8mm, but that was before we took more careful measurements, and took into account removing the cut-off tube completely flush to the casting (and cutting the coolant channel), which allowed more rotor clearance.
 
Thank you very much.
My workshop was into milling down the tube but keep a little left to keep the lock ring in place.
But it sounds as if this might be to tight when having 10mm instead of 8mm :)
I will inform him about this.
I suppose that part is pressed in from outside and sits tight enough not to "pop out" without the lock ring.
 
The tube has been pressed out and AFAIK nobody has an explanation for that circlip: why is it there, what did it do? Howard considered that pressing the tube out and installing a second plug there would be simpler for home DIYers but after seeing the galling that occurs when it's pressed out, he rejected that as a reliable DIY solution.

I, and others, have removed the circlip and ground/milled/scraped off the tube down to the casting surface. This buys much-needed clearance.

Howard and Mr. Hyde's SS sheet metal cup plug is thinner material; IDK what instructions they will provide for installation, and perhaps a coolant channel cut won't be necessary. I elected to fab the aluminum plug(s) instead, because I like it over the formed SS cup solution, and I've got a way to cut a coolant channel and I don't need the "easiest" way to do this. Howard seems to want to make the cheapest/easiest DIY mod, and those aren't my goals.
 
Kevin has his in-situ coolant delete kit for sale, $150 + tax. On a Model S/X that has no current signs of a leaking rotor seal, this method saves a lot of labor, but it has no practical advantage for the RAV4 EV. YT video:

Website for sales:
https://113a8f-4f.myshopify.com/products/tesla-ldu-coolant-delete-kit

Brian is selling the SS formed caps (without Loctite) for $44 (with tax) via eBay:
https://www.ebay.com/itm/166811720663

Paul is selling the SS formed caps + Loctite for $50:
https://armstrong-mobile-forklift-repair.square.site/shop/tesla-sales/OFZJ2ZCVNNQRWQANTM2DMIU6
 
Is that video the best how-to youve seen for making this mod? If the LDU is right behind the left front wheel, I might actually attempt this if I can do it in a weekend.
 
Kevin's video I linked to above? I didn't watch the whole thing, because I personally wouldn't consider doing the mod that way, because you don't get the benefit of coolant flow through the flyover tube anymore, plus you hacksaw the coolant manifold in half while it's still in the car.

This makes sense if you have a completely dry Model S LDU, but makes no sense for the RAV4 EV, because it's so easy for us to remove our manifold to install a plug or cup; it's a lot more work to do on a model S or model X.

For any of the several DIY methods, you have to round up the parts in advance.

For Kevin's method, doing it on a Model S in the car, you have to buy his plug kit and manifold adapter.

For any of the other bolt-on manifolds, you have to buy one and have it shipped, and they are all expensive.

For the plug, you need to have one machined.

For the stainless steel cup, again you need to order it.

Don't take your car apart until you have the DIY part(s) lined up, because you don't want to leave it apart very long, you'll forget how it goes back together.
 
Is there a video for the stainless cup mod? Is that method the one you'd recommend for Rav4s?
 
Of the available DIY methods for performing "coolant delete", the SS cup method is probably the best balance of cost and ease of installation. Using one of the $$$ CNC coolant manifold replacements is the easiest, because no cutting/grinding/presswork is involved (and it's reversible in the case of QCC's). If you need the absolute easiest method, the QCC is likely a good choice.

I'm not aware of a video start-to-finish for either the plug or SS cup.

The car has to be raised enough to remove the RF tire.
An L-shaped plastic panel must be removed. This is shown in Alex' video on speed sensor checking at around 0:30 :


I had my under-panel off as well, because I was removing my LDU due to coolant having leaked into the stator cavity; I do not know if it's necessary to remove the bottom panel, but it'll be a lot easier if you do. Unfortunately, this is not as much fun to R&R, as it's got both 10mm head bolts and two different sizes of push clips holding it in place, and it's large (though not particularly heavy). So it's fussy but not difficult. There are "always" a couple of the push clips missing or lost or broken during removal, so plan on having to go back and install a few more later, after you're completely done, unless you can obtain a 10-pack before hand . . . and, again, two different sizes.

One the two plastic shields are out of the way, place a drain pan under the 19mm hose, unbolt the barb and let it hang; drain coolant.

Remove the speed sensor (one bolt) to prevent it from being damaged later. It's over $300.

Remove one vertical 10mm head bolt on flyover tube clamp (up on top/left as you are staring at the LDU end). See pics here.

Remove one vertical 10mm head bolt that fixes the flyover tube to the top of the manifold. Pull the flyover tube straight up to disengage it from the manifold.

Remove (5) 10mm head bolts that fix the manifold to the LDU rotor cap. Pry the manifold loose from the rotor cap, prying between the lower and upper half of the manifold where there's a large obvious place to put a big screwdriver or whatever. You can't hurt anything. You have to break loose the RTV in two places, so this may take some effort, but no beating on anything is required.

The coolant manifold can be pulled straight toward you. The bench work begins.

Cut off the taper tube, grind (or machine) the remaining taper tube back to the casting, grind/machine a bypass groove in the casting, push in the cup. Howard's pics of his are here, but he used RTV and I recommend anaerobic sleeve/bearing mount compound instead.

Then it's just reassemble, add coolant, and try to get the air out of the coolant loop (run coolant purge procedure via TPD x2, which didn't work for me; use vacuum purge method; or just drive/add coolant until the dash warnings dissipate).

That's the basics, and I don't see a video being of much use.
 
grind/machine a bypass groove in the casting
In the video above posted by @ToyTesla it looks like with the steel cup mod, no bypass groove is needed. Is that the case? That's the part I'd be least confident in taking care of myself.
 
Last edited:
In the video above posted by @ToyTesla it looks like with the steel cup mod, no bypass groove is needed. Is that the case? That's the part I'd be least confident in taking care of myself.
The steel cup appears to let coolant by since it doesn’t bottom out. At least that is my understanding. Might want to check with seller (Armstrong) to verify. I ordered one yesterday and will attempt a refit in the coming weeks. No issues with my vehicle, just preventative since i love driving this car.
 
Some think that the SS cup is thin enough that it can both clear the rotor and leave room for enough coolant to flow out of the casting, make a 90° turn, and continue to the flyover.

I'm not convinced. But that's what the sellers of the SS cups are saying. Thing is, you can't check: we've measured that coolant manifold up and down, but tolerance stack-up between different manifolds means that we're measuring from different points, and if you use the modeling clay check method, you burn a cup and you still don't know if your measurement applies to the next manifold; their castings vary in some dimensions and that particular surface is not machined the same way every time, from the samples I've seen.

We don't know the function of the flyover tube coolant circuit to the heat exchanger (stalagmites in the casting) over at the differential. And we don't know the implications of not having coolant flow there. Is it important? Tesla thought so.

I cut a groove for coolant bypass in mine, and I advise grinding one (or more, shallower grooves) when I advise people doing a DIY mod, but only time will tell if not providing any or adequate coolant flow to the heat exchanger downstream has any noticeable effect. Three minutes with a Dremel is all it takes to do a ghetto groove, and you have to grind the taper tube back to the casting anyway, so you've already got some kind of grinding tool in your hand when you're working on it.

So, it's a no-brainer to me to know for certain that there's adequate coolant flow to the flyover on your manifold, by providing a defined passage; regardless of how little or much clearance between the cup's bottom and the casting, it'll still flow.

IMG_7748.jpg

The groove does not need to be fancy; any gouged-out groove will work. It doesn't show, and there's not a huge flow so smooth isn't much better than rough. I just happened to have mine on a milling table when I was working with mine, so I mounted an end mill and made mine fancy, but that's not necessary.
 
Last edited:
Permatex 64000 seals the cap to the manifold. Does any sort of Permatex or other seal need to be applied to the manifold junction with the LDU/motor (flat areas lined with red below):mani.jpg
 
Last edited:
RTV sealant on the large left side, nothing at all on the small right side of your picture.

That right side is sealed by a formed O-ring, there's no point in adding any sealant to that. Unless you have lost that o-ring. You can't buy it, it's one of those Tesla parts. In that case, you goop it up with our TV and hope for the best.

None of this is under any serious pressure, so there's a lot of tolerance for errors.
 
Back
Top