Removing Carrier Bearings from a Guard LSD Carrier

[burgermeister]

The title says it. How the heck does that bearing come off? The other side had the 2 notches to fit a puller, no problem. I believe the factory carrier has notches on both sides. The Guard Transmission carrier, on the ring gear side, nothing.
I suppose one could destroy the bearing and then pull the race, but how does one set up a new R&P without being able to pull the bearing off? Doesn't *seem* like Guard to purposefully make life impossible.

[gled49]

The diff has the register for the side cover oil pump drive gear, which makes the S1 side even tighter. Hopefully they used the 996 bearing shim which has 2 reliefs, at least you can pull evenly. You need regular pullers or special bearing pullers.

[burgermeister]

Ah. Unfortunately the spacer only has once cutout, so no luck there. Rancho put the thing together back in 2017 - I guess machining a second flat was too time consuming for them.
Judging by the other side, the bearing is a crazy tight fit - it seemed about the limit of my 2 jaw puller before it budged.
I'm debating making a tool similar to what I used to pull the pinion shaft bearing off - something to register against the rollers (might have to do the cage on this one) but that also envelopes the rollers so they can't go sideways and pop out of their cage.

[HaroldMHedge]

Maybe try this style of bearing puller.

[Zuffenwerker]

I have gotten those off with careful heat and 2 prybars. Not ideal but works. Obviously if you heat the bearing too hot it’s ruined.

- Daniel

[dannobee]

Use the last bearing puller kit in gled's pic; the one with the yellow ring in the box. You'll be throwing rocks at the other pullers and it won't destroy the bearings.

[burgermeister]

The generic Harbor Freight bearing puller does not fit, nor would it leave the bearing intact.

Tried prying - there is room for a .300 x .150" tool to fit into 2 holes partially covered by the bearing race. I am not the best heat treater, propane torch and all, but my 2 pieces of hardened and tempered oil hardening tool steel were not up to the task. They broke without budging the bearing. Needless to say annealed steel just bent.

The yellow tool suggested by gled and dannobee seems intriguing. I found a "Dana & Ford" carrier bearing puller that looks a lot like that on Amazon.

Seems like it might be too large - is there a smaller version somewhere, or does that one fit?

Thanks!

[dannobee]

Worst case scenario is to cut the cage off then use a die grinder to cut the inner race. You'll be cutting at an angle and go to almost the thickness of the race. Then get out the hardened chisel and whack it in the cut groove until it cracks. Then it'll fall off.

I think one of those puller kits has three or four different sized pieces to cover the vast majority of differential bearings.

[kevingross]

For the 90 mm bearing, there is a collet-style puller made by the same company that makes the VAG1582 set for the VW/Audi/Porsche tools program, which set includes an 80 mm collet. The 90 mm collet and the VAG1582 are absolutely indispensable when you do this kind of work every day. The collet works by putting a "ledge" over the bottom of the bearing's rollers, which you can see in the second photo.

I have an extra one of these tools for sale. It belonged to a good friend who passed away earlier this year, and we have been liquidating his shop. Original cost was about $900 for the "gripper" plus the size-specific collet. Will consider offers.

[burgermeister]

I am familiar with the grinder / chisel approach - but I figured that for setting up the R&P I would need to remove and reinstall the bearing a number of times with different shims to set preload and backlash. I assumed that the same bearing should be used ...
I ordered one of the Dana / Ford pullers - still in the 2 figure range on Amazon. I don't plan on doing this a lot. I wasn't planning on it at all actually, all I wanted to do was replace the 2nd gear synchro band / teeth / sleeve...
But the thankless car had other ideas. R&P was shot, and of course the bearing races were loose in the drive housing, despite the smaller 55mm mainshaft bearing.

[burgermeister]

Well, the Dana/Ford carrier bearing puller arrived this weekend ... and failed. The bearing race lip had insufficient strength to overcome the press fit. Destructive removal methods worked exceptionally well - after cutting the inner race about 50% through, it just cracked on its own.

Post mortem measurements suggest that some things are amiss. As reference I have an old 914 diff carrier (used to make a tool for setting pinion depth). It's had some bearings pressed on and off, so no idea as to it's original dimensions. It measures 1.96935" and 1.96870" for its stubs.
The Guard LSD measures 1.96965" and 1.97035".
The SKF bearings that came off the Guard measure 1.96835.
Measured with Mitutoyo digimatic micrometer and .0001" reading bore gage.

So on a stock carrier, interference fit is .0005 to 0.001", which seems about right to me. On the Guard interference was .0013" (the bearing that came off) and 0.0020" (the bearing that did not come off). The latter seems way too tight. Also evident is serious galling of the carrier stub, not present on the side where the bearing came off nondestructively. I should mention that the Guard carrier won't cut with a file, and the ring gear flange measures about HRC 60. So not exactly easy to smear and gall.

As an added bonus, the new SKF bearings measure .0005 and .0006 tighter than the old ones. So I foresee trouble. I ordered a couple of FAG ones, will see how those measure. I always thought bearings were made to within a couple of .0001" of their spec.

I'm also looking for ideas on how to hold the inner race when attempting to hone a bearing for a slip fit. With the purpose of using these honed bearings for setting up the preload & backlash, and then hopefully only pressing the new bearings on once.
And also, maybe, for honing the new, apparently undersized, bearings for a more appropriate interference fit. And also for appropriate ways to hone bearing races in the first place - those things do not cut well with silicon carbide and even less well with aluminum oxide.

Failed pull

Galled stub

Other Side Stub

Galling polished off. If the two holes marked were slots, bearing removal (assuming a reasonable interference fit) would be a lot simpler. I cannot understand why Matt wouldn't have done that. Given the hardness of the carrier, machining such slots now would eat quite a few carbide end mills, and still might not succeed.

[Magyar Kiwi]

Suggest you have a look at this thread: "A guy in his garage attempts to rebuild a G50" Page 10 of 11, Post 184 It describes the use of a pair of sacrificial honed out new bearings for the set up process.
Best regards
https://forums.pelicanparts.com/porsche-911-technical-forum/1157509-guy-his-garage-attempts-rebuild-g50-10.html

[Tom F2]

I thought that preload had to be set with new bearings. Not true?

Also, not having Kevin's impressive array of tools, I found that the only way to remove the carrier bearing on a factory limited slip cleanly was to separate the housing and initially press the bearing off from behind by means of two decapitated bolts slipped through the holes from the inside of the carrier, the ones circled in red in the photo above. On the factory carrier there was just enough of the inner bearing race exposed through the holes to get a purchase and move the inner race a smidge. This initial movement enabled me to use a regular puller to complete the removal.

[dannobee]

Preload spec depends on if the bearings are new or used. ~15 in lbs for used, ~40 in lbs for new. I outlined the method to measure in the other thread mentioned above. The preload spec is not specifically given in the workshop manual and instead they use a 0.4mm "constant." I would venture to say that they're assuming an "average" technician is going to be working on it and not someone whose specialty is transmissions and differentials. And their procedures have changed over the years and seem to reflect that mindset.

If you find that your original bearings have little to no preload, something is worn and needs to be corrected.

For reference I just measured one of my G50 carriers and got 1.9697" and 9.9690"

[Matt Monson]

I have stayed out of this thread initially because I had been talking directly with the OP. However, when someone decides to call me out by name, well, I feel like that warrants a response.

When I bought GT 15 years ago, this was a pre-existing product. While not an engineer, Paul Guard was a machinist, and I consider this the best 915/930 LSD in existence. It is designed to OEM specs, and is intended for installation and service according to the factory manuals. I have sold over 1000 of these LSDs. They are as intended. People sometimes consider me elitist when I tell them why I’m a wholesale business. These are not DIY parts. The people with the proper tools and knowledge to work on them is limited. On a scale of 1 to 5 in difficulty, setting up a 915 LSD/ring and pinion is a 5. I avoid a lot of brain damage by not selling these parts to the home mechanic. I see zero reason to redesign the product.

Tom F2 is correct. The holes are there so that one can disassemble the LSD and press the bearing off from the backside with a couple pins, during service.

As for the numbers? For starters, these are metric parts. The spec is 50mm. We manufacture the LSDs to 50.01mm +/-005mm. The bearing removed is clearly out of spec and tighter than 50mm. You can do your own conversion…

[burgermeister]

Sorry Matt, I did not mean offense. The LSD is fantastic in use, and I love it, and I do tell that to everyone I talk to about it. That doesn't mean it can't be made easier to service, but that's your call to make. I'd have made that change, but I didn't design it and I'm not selling it. And I only see one end of the problem, though I do find that end really irritating!

Your posts always contain useful information, and are always appreciated. No need to stay out of it. I have a thick skin.

As to DIY - I realize I am a moron, as I own a car I cannot afford to pay to have serviced at $200+/hr. Poor career choices. So DIY it is. Used to be more common, but is getting much less so these days. Sorry if that offends. I am an engineer with decent hobbyist machining skills & tools, and the pinion setup does not look too difficult to me. So far the most vexing thing is that carrier bearing.

Yes, obviously metric - but my hobbyist machinist knowledge is in inches, as are my tools, and my bore gage...

One of my friends has a bearing spec book, and the ID of a FAG carrier bearing has a tolerance 0.012mm, so +-0.006mm. Which is bigger than I had imagined! Your carrier does better at +-0.005". Still, that is quite a variation of interference fit stackup! Not saying it's anyone's fault, or that other parts wouldn't have the same problem - just saying that it seems like a lot.

The new FAG bearings also measured .0005" tighter than the old (presumably original) SKF bearings, and about the same as the new SKF bearings.

I will figure it out eventually.

[Matt Monson]

No worries. Saying “called out” makes it sounds more aggressive than it really is. I’m more like Beetlejuice. I will say that I only speak metric. If you flip back and forth between inches and metric, or use inches, I’ll just scroll on by. I usually check in on my phone over my coffee, and pulling out a calculator and the like seems like work. I come to Pelican to chill out.

DIY doesn’t offend in any way. It’s how I got my start and did so much of my learning. As a business owner of a very specialized product, catering to DIY doesn’t make sense. I don’t run a tech support line. Askholes are a daily nuisance. The number of random gearbox questions that have zero likelihood of me selling something is likely 1/3 of my inbox. I have to make a choice of where to direct my limited resources, and it’s not on unpaid tech support. Do NOT take this comment in any way as personal or directed at you. It’s just business.

With the damage, you can no longer trust the numbers on my diff as being true.

On the bearings, I’m left wondering if SKF is manufacturing them tight on purpose because the original diffs are getting old and worn out, and are getting loose. 9 times out of ten what wears out a differential is the bearings start spinning on the flange.

I suggest you order some bearings from FAG and see how they measure out.

[burgermeister]

I have the carrier bearings all sorted out. I took the LSD stubs down to 50.019mm (bit above the spec Matt provided) with some 600 grit paper. Interference fit should be reasonable at .033mm / 0.0013"

Opening up the set-up bearings was not a happy task. Grape hone or cylinder hone was driving me bananas, for a looooong time, with minimal results and increasing bellmouthing, until I gave up. I ended up turning an aluminum cylinder with a very short recess that fit the back lip of the inner race of the bearing with a press fit - there is about 1/16" of metal to work with before the cage gets in the way. Chucked it in a 4-jaw, very carefully dialed it in, and then used a tool post grinder to open up the inner race. That also took a long time but was much more satisfying and maybe more accurate.

The next question is of course how much to overthink (or not) the pinion depth. At operating temperature (190F as a SWAG) the pinion depth should increase by about .09mm (+.14 from the case, and -0.05 from the crown wheel, very roughly). Does it make sense to account for that when setting the new one? Porsche, being crazy Germans, may or may not have thought about it and accounted for this in their markings. I am guessing Albins (the replacement R&P) did not, since their instructions specify biasing the setup & pattern to at least acknowledge the effects of operating loads/temperatures.

FWIW, the old one measured at 66.37 with the existing shim but is marked N25 (so 66.55 desired depth). I don't have lots of confidence that the shop that did the setup & rebuild did a very thorough job. Or perhaps they set it up by mesh pattern and not the depth spec - pattern seemed reasonable to my untrained eye.

FAG bearings measured the same as the SKFs; they became the set-up bearings.

[burgermeister]

The LSD came apart and reassembled with zero difficulty. So as to bearing removal when swapping R&P, that path would seem to be a no brainer.

The crown wheel is another matter. The albins crown wheel has 0.046mm more interference than the factory crown wheel, and as such was a bear to get onto the LSD housing. I found that odd because the pinion shaft had exactly the same dimensions as the factory piece, at least to what I can measure.
Using all 12 bolts going round and round by 1/8 turn ended up getting the crown wheel seated. I expect it won't come off easily.

[gled49]

I don’t know about others but I heat the ring gear to install if needed. All late G50 types require heating to install.