Kevin, the manual says to "lightly tighten the side cover using two nuts opposite each other", but it can also be done using additional nuts.
The important thing is to have the cover sit evenly, with the least possible amount of pressure.

Here is another method of determining the differential bearing preload, from the factory Turbo Carrera manual.
Item B in the illustration is the magnetic disc from the VW385 set, VW385/17.

http://forums.pelicanparts.com/uploa...1519153009.jpg

http://forums.pelicanparts.com/uploa...1519153059.jpg

http://forums.pelicanparts.com/uploa...1519153096.jpg

Jon, this made my day, possibly my whole week. I did this test multiple times, and measured 0.33 to 0.40mm around the cover. there was one small spot where I could just barely get a 0.45mm gauge in, but it was .40 on either side of that. To be sure I was getting good measurements, I also torqued the entire cover, then loosened all the nuts, then smacked the case with a rubber mallet a few times to get the lid to release in an amount equal to the preload. Same results each test. I'm done worrying about preload now. Thanks!


Thanks Kevin. I think I've done this to death now.


Hi Peter! I'm not blaming you for anything, I hope I didn't come across as such. I took "clean and dry" very seriously on this build. Anyway, now I'm straightened out. Thanks!


It is possible I'm just a little bit obsessive. And crazy!

Last update for tonight: I'm getting backlash measurements now between 0.17 and 0.20mm at 4 different points around the ring gear, 90 degrees from each other. I then rotated 45 degrees and did 4 more 90 degree tests. The reading I'm getting most consistently is 0.18mm. This is with shims of 3.1mm/3.3mm on speedo/ring gear sides, respectively. Previously with 3.05mm on the speedo side I was down at 0.15mm in backlash. My ring gear is inscribed with 0.16mm as some may recall.

I'm thinking I'll measure backlash again in all positions tomorrow and make absolutely sure that my numbers are correct. I will also run a pattern with prussian blue again.

I'm feeling good about the pinion depth and preload numbers at this point, not sure how much more I can do with backlash from here. As always, all input welcome.

Oh, no, Evan! No offense taken, I just figured that I should explain my thought process regarding "clean & dry!" You've been a perfect gentleman thru-out this repair, and if it doesn't make you just a little bit crazy, then we do have a problem! Carry on, sir!

Hi Jon, yup, and this method is the one I start with, that is, use to get an approximate value to start with. Not a final value, I always use turning torque as the real and final measure. With the bearings unshimmed, the method in the picture is rough because it's impossible to get the upper bearing (with the transmission on its side, as depicted) firmly and repeatedly pressed inner against outer race.

This is what works for this baker! You and others may have techniques that work better for you, of course. And I enjoy learning from your experiences. Gracias!

Tapered roller bearings require preload for various reasons. The separate bearing pieces need to be forced together, beyond the point of contact, to maintain alignment.
In this situation, you're dealing with various directional forces, thermal expansion of the housing and other components, and compensation for long-term bearing wear.

Unfortunately there is no simple, precise way to measure differential bearing preload in a 915.
Bearing drag is not preload, but rather an effect of preload that can be measured and correlated to preload.
Even so, bearing drag measurements vary between manufacturers, and it's not clear whether those numbers from the 1980s are still applicable for bearings produced today.

In this situation, you achieve bearing preload by over-shimming the differential, which creates force by distorting the cast alloy housing, and the side cover in particular.
How much distortion is acceptable? The factory gave a nominal range of 0.30 to 0.40 mm, and these alloy housings are now 30 to 45 years old.
There are other considerations too. In Evan's case, I don't think the operating conditions will be extreme for this transmission, but maybe he'll tell us otherwise.

So with all things considered, I think that a preload in the 0.40 mm range will be sufficient for Evan's bearings, even if the bearing drag results are not quite within 1980s spec.

I agree.

OK, so I did a marathon check of backlash this evening, checking 4 points 90 degrees from each other, then rotating 45 degrees and doing 4 new points 90 degrees apart, and then doing that all over again a couple times. Why so many tests? I want to be sure I'm getting a good picture of the range of backlash in my diff, and was hoping to weed out any oddball results. Lowest reading .15, highest .20, but these are the extremes. Average of 24 (!) tests is .18mm, if I throw out the highest and lowest extremes and take the average of 17 tests, I get .17mm.

Somewhere in there is the truth, and this is as close as I'm going to get to it. The factory manual allows 0.05mm in deviation, and I'm within that. If we assume that the extremes are more a product of testing inaccuracies then anything else, I figure I'm pretty well in the right place for a diff with 165K miles on it. I await the expert evaluations...

I'm headed back out to garage to run a pattern now, more on that later. SmileWavy

And the pattern, in a lovely shade of Prussian Blue. I went with a thin film at first and couldn't really discern a pattern, so I applied a bit more and started over.

http://forums.pelicanparts.com/uploa...1519249983.JPG
http://forums.pelicanparts.com/uploa...1519249983.JPG
http://forums.pelicanparts.com/uploa...1519249983.JPG
http://forums.pelicanparts.com/uploa...1519249983.JPG

Evan, if all your measurements and adjustments to the pinion, ring gear and differential are accurate, then you should be good to go.
You can make further adjustments if you still want to increase the bearing drag, but if your preload is already in the 0.35 to 0.45 mm range, you should be fine.

I still cannot see a clear pattern on your ring gear, and still think that you're using too much blue. If the pinion was clean and dry for this last test, then what pattern did it leave?
If you haven't been doing so, then try to apply some force on the pinion, and against the differential and ring gear.

I've agonized over every measurement, and done my best-- with lots of help from everyone here. If something is off now, then it isn't for lack of trying!


The ring and pinion were absolutely squeaky clean. I'm struggling to get good readable results, but will try again with some counter force on the diff flanges when I test again this evening.

EDIT: these photos are from yesterday's test. Regarding patterns, it looks to me that I'm getting contact nearer the toe on the drive side and nearer the heel on the coast side. Any pattern is poorly defined, to be sure, especially on the coast side. Back out to the garage...


http://forums.pelicanparts.com/uploa...1519318992.jpg
http://forums.pelicanparts.com/uploa...1519318992.jpg
http://forums.pelicanparts.com/uploa...1519319351.JPG

subscribed

I've decided that I don't like Prussian Blue. Dark color on dark ring gear. Clearly it isn't idiot proof, or I'm too big of an idiot. :D

However, oil-based yellow house paint, now this stuff I like! See photos, I think it is safe to say we have a pattern now.

http://forums.pelicanparts.com/uploa...1519333909.JPG
http://forums.pelicanparts.com/uploa...1519333909.JPG
http://forums.pelicanparts.com/uploa...1519333909.JPG
http://forums.pelicanparts.com/uploa...1519333909.JPG
http://forums.pelicanparts.com/uploa...1519333909.JPG

Yes, the results in yellow are much clearer :-)
Was this done while applying opposing force to the ring and pinion? I meant was the test done without blue on the pinion. It only needs to be applied to the ring gear.
You appear to be doing a very thorough and commendable job, but I can't be sure of everything you're doing without standing there and seeing for myself.

Here is the illustration from the 70s BMW 2500-3.0CSi manual again, with additional instructions for Gleason gears.
I could tell from a previous photo that your ring gear has a Gleason tooth profile.

http://forums.pelicanparts.com/uploa...1519340721.jpg

http://forums.pelicanparts.com/uploa...1519340792.jpg

Thanks Jon, I do realize there is only so much one can comment on without seeing the subject in person.

-The pinion gear was clean and dry on the last two tests, so the only paint on the pinion gear was transfer from the ring gear during the test.

-I jigged up a drag on the diff flange to get enough opposing force and be sure the pinion was pushing firmly into the ring gear during the test. I used a ratchet on the pinion shaft collared nut to turn the shaft and be sure of good even pressure.

-I applied the yellow paint to just half of the ring gear, so I could also see what pattern transferred to the clean section of the ring gear, as you see in the photos.

-So if I am reading this correctly, the pattern I am seeing is just slightly too close to the toe (inside of the ring gear) which would indicate too little backlash. The pattern on the coast side is also not aligned with that on the drive side. My numbers, by way of review:

-My pinion depth "r" spec from the factory is 0.27mm. I achieved 0.24mm with 0.55mm of total pinion depth shims.

-My carrier bearing preload, measured as gap between side cover and diff housing, is between 0.33mm and 0.40mm.

-My backlash measures an average of 0.18mm with highest readings at 0.20 and lowest at 0.15. The vast majority of my readings are 0.16-0.19mm, putting me at the high end og Porsche spec. My ring gear is inscribed with 0.16mm.

Considering that I have repeated my tests, especially backlash, so many times, I am comfortable with the numbers I am getting. As was commented earlier in the thread, the tools and real numbers are what we use, while the pattern is another, complimentary tool in the box. However, I'm not sure what to think, if the pattern says backlash should be increased, while my measurements say not. I think most normal people would just reassemble at this point, though clearly I have demonstrated I'm not normal. SmileWavy


evan

Evan, if you want to go further with this, then add 0.05 mm to the S2 (speedo ring) shim total, and see how that affects the pattern.
It might require another adjustment to the pinion, to bring the backlash down again.
Also keep in mind that as more load is applied, the contact pattern on the ring gear will move outward.

Where are you measuring the backlash, in relation to the ring gear teeth?
For a proper pattern on the coast side, you should rotate the ring gear and differential forward, while applying a counter force to the pinion.

The downside to using oil-based yellow house paint is that it might not be too much fun to clean everything up :-(

Yeah, the house paint is a real pain to clean up, but I can't beat the color!

Do I want to go further with this? If the pattern says something still isn't right then maybe I need to play with it. But I have to admit I'd like to move on soon, if things are where they should be.

I'm measuring backlash at 76.5mm out from the center of the ring gear, see photo.

http://forums.pelicanparts.com/uploa...1519381104.JPG

It's ironic that you used yellow paint. There actually is a very specific gear marking compound that is yellow. Put "yellow gear marking compound" in your favorite search engine.

I've really enjoyed your project so far. Keep up the good work.

Thanks!

Yes, I know there is yellow compound, but after buying a tube of blue from Amazon for $35 that isn't helping me, I decided to just go for something I already had, yellow paint! SmileWavy

used gears...
 

For what it is worth, I stumbled on this drawing, which shows a nice contact pattern on the inside 1/3rd of the ring gear for new gears, but something much more like mine for used gears. I don't know if the BMW drawing for the Gleason R&P takes used gears into account, or if this has much effect? (I don't know any better, so I'm asking...)

http://forums.pelicanparts.com/uploa...1519402419.jpg