OK folks - this is gonna get fiddly, so if you are not a geek, or don't plan on rebuilding your transfer case at some point (4Matic only) you'll want to move along... ;-)

When rebuilding my transfer case (including replacing the four tapered bearings) it was clear that the bearings were under a fair amount of preload tension, as they went from being easy to spin with my fingertips with the case bolts inserted but not tightened, to requiring a pretty good grip and bit of effort to spin the output shafts by hand. At the time, I had no reference to what the preload / rotational drag SHOULD be, but guessed (incorrectly?) that it would be OK.

Fast forward to now, and I've learned a good bit about the subject, but really can't find any definitive specs for the actual preload specs for the bearings used in our MB 722.9 transfer cases. Hence, this thread...

The bearings involved:
Timken NP604623 (three)
Timken NP925485 (one)

The Timken (OEM of the bearings in the 722.9 transmission) website had some info (generic) on the relationship between preload, endplay and bearing life. There's no specific numbers for a specific bearing, but it's clear that once you get into the "preload range", you can get too tight pretty quickly. OTOH, it looks like you can get away with a good bit of endplay without reducing the lifespan of the bearing dramatically. Here are two charts from Timken...




Based on a 50mph average, that 5,000 hour line should correspond to something like 250,000 miles (probably more than anyone rebuilding a transfer case would ever drive their repaired GLK). ;-) I show the two charts because the first one shows more detail on the preload dimensions, and the second because it shows lifespan. That (wide) span is MUCH wider than the "optimum setting" range, and (to me) seems to indicate that these bearings will last a long, long time if the preload / endplay is at least reasonably close to the "optimum setting".

Basically, these types of single-row tapered roller bearings are set up using some method of adjusting the preload, and the results are calculated by measuring the rotational force to spin the bearings (after "break-away", and at a rotational rate of around 4 RPM). In the case of the 722.9 transmission / transfer case, the bearing preload is "set" based on the thickness of flat spacers under the bearing housings (I believe around 2.5mm thickness). Based on similar bearings being used on (big) differential pinion gears, I'd guess (SWAG?) that the target rotational drag would be in the 20 inch-pound range, for new bearings, based on this chart:


It's interesting that the rotational drag for a new bearing is about twice that of a "worn-in" bearing. I suspect not many of us would be rebuilding a transfer case reusing the old bearings, but if so - you'll want to adjust accordingly (maybe use 10 inch-pounds as a goal?).

I also found a chart on the Timken (OEM of the bearings in the transfer case) that shows how to calculate the relationship between rotational drag of the bearing, and the preload. This is helpful because I assume you can calculate the "perfect preload" with the following process:
1) Measure the rotational drag of each of the two transfer case shafts (rear and front gear clusters in the intermediate housing), each installed separately, with the intermediate housing bolts torqued to spec.
2) Look those drag numbers up on the following chart
3) Calculate the DIFFERENCE in shim thickness by subtracting the shim thickness of the "ideal" (20 inch-pounds in my case) with the shim thickness corresponding to the ACTUAL rotational drag. For example, if your goal is a rotational drag of 20 inch-pounds, and you measured 50 inch-pounds, you'd subtract 0.28 from 0.53 to get 0.25mm, which would be be the amount you'd want to REDUCE the thickness of the shim behind the bearing race (or add elsewhere - more on that later). The Timken chart only went up to 38 inch-pounds, so I did my best to extrapolate the "extension" of the chart to higher rotational drag numbers (for those really tight transfer cases). My extrapolation is shown in purple in the chart below. FWIW, Timken repeated some of the shim thicknesses in their chart, but I suppose that just reflects availability of various shim stock sizes.



To adjust the preload, there are two options (I believe)...
1) Remove material from the existing spacers under the bearing races (requiring an accurate micrometer and some care)
2) Add material in the form of a gasket between the transmission housing and the transfer case intermediate housing (the one that holds the two gear clusters with the bearings). According to a MB tech (who took exception to my how-to transfer case rebuild video's lack of preload info), it seems that they generally use 20 thousandths of an inch gasket (aka 0.5mm) material to make a gasket, which will effectively reduce the preload. The end result might be slightly less due to gasket compression, and due to at least a tiny bit of thickness with the original sealant that isn't used with the gasket, but it's still moving the preload result dramatically. But based on the "orange" lifespan chart above, that 0.5mm move would be the entire width of the curve where it lives above the 5,000 hour line. That makes sense, because if the bearings DID fail, it's probably because they were living "to the right" of the optimum zone (to the right of where the lifespan curve line crosses the 5,000 hour line, heading down).

My thoughts are:
1) Test each "gear cluster" individually (leaving the other one out), for rotational drag (and therefore, preload / endplay). I'll use 20 inch-pounds as a goal for each of the gear clusters.
2) Leave the sealant on the housing until it's time to do the final assembly (to account for the actual thickness of the sealant)
3) Don't replace the front output shaft seal until all the testing has been done (to minimize the chances of damaging it - it's VERY fragile)
4) Drop the front driveshaft out to do the testing / installation. Based on my experience, the best way to do this will be to drop the right side exhaust system (cat back to mid-pipe), which will provide the necessary clearance to get the front driveshaft (aka propshaft) in and out. Otherwise, you end up about 5mm short of being able to do so - at least on my 2011 GLK 350.
5) Install the front output shaft seal and assemble the transfer case before reinstalling the front driveshaft. This will allow for checking that the front seal wasn't damaged (it's easy to displace the tiny spring loop during front output shaft installation), and prevent having to force anything to get the front driveshaft and rubber seal aligned / pressed into place.

A couple other notes:
1) The rotational drag will be measured using a very small, quarter-inch beam-style torque wrench. Using a click-style torque wrench would be difficult. I'll couple the torque wrench to the output shafts using a large deep-well socket over the shaft, with a shop rag in between to take up the slack (as well as the appropriate drive adapters from 1/4" up to 1/2"). It just needs to be snug enough to provide the very light rotational torque.
2) My torque wrench is calibrated in inch-pounds. If you want to use Newton - meters, divide the inch-pound numbers by 8.85.

Sorry for a long post, but I wanted to pass this by the folks here who may have done this job and/or are interested in doing so in the future, and to get any feedback / corrections / comments (I'm making some assumptions and am willing to listen to any / all opinions about this plan).

 

A wealth of info, thank you for posting. You'll have to keep us updated on how the rebuilt T-case is doing.

 

A couple of comments. Note, these are my opinions, not presented as indisputable facts.

Mercedes W126 rear wheel bearings are a similar application. They are set up with 0.04-0.06mm end float. You check it with a dial gauge and a pry bar. I suspect the reason for play rather than preload is due to the fact that they are carrying a heavy radial load which tends to increase the angular contact.

Every rear axle pinion I've ever worked on specified preload. One way to set preload is with shims. You can start with a known (measurable) amount of end float and add or subtract shims (or gaskets) to arrive at the desired preload.

I'm leery of using gaskets in this application due to the variability in compressiblity. I think sealant is a much better choice. I haven't had hands on this actual transmission, but maybe Hylomar? If you use a gasket, don't use more than one as the results are difficult to predict.

If you need to reduce the length of a sleeve to set the preload a surface grinder may be the best way to do it.

If you are assembling it with used bearings (anything with more than an hour or two of use) you definitely want to set it up with a little less preload (or more end float) than if the bearings are new.

 

I read that too much end float on these transfer case bearings can cause damage as the rollers "bounce around" (not the way it was described by Timken, but you get the idea). It looks like there's a pretty wide (relatively speaking) range of end play gaps that will still result in a long, long bearing life (longer than necessary).

There's also the fact that the bearings and races are steel, and the case that determines the preload is aluminum, which does change the preload at temperature. OTOH, I really don't think that the transfer case gets all that hot, so the effect is minimal, but has to be taken into account (and IIRC, the heat increases preload, leading me to want to cheat a little bit toward the end play end of the spectrum).

I've never seen different width shims for our transfer case bearings - I suppose someone out there might make them, but probably a "shop-only" vendor. I'm still vacillating about whether to reduce the preload by "machining down" the existing 2.5mm (IIRC) spacer, or by adding the gasket (knowing that it's going to produce a less than precise result). It'll be a fair amount of extra work to machine down the spacer, but (big but...) if my calculations are correct, AND the existing races aren't very damaged (if at all - who knows?) I should be able to measure each assembly with the old mid-case races installed (figuring they're still dimensionally the same as the new ones), then pull the old ones out, adjust the shims, and reinstall the mid-case races (new ones) and now-thinner shims.

I understand what you're saying about grinding the "crush sleeve" that's used to set the preload in many pinion applications. On my 1996 Jeep Cherokee (with an 8.25" Chrysler rear end), nothing I was doing was crushing that sleeve. I ended up using a REALLY stout 1" drive breaker bar and "turned" it with my floor jack. By the time the sleeve STARTED to compress, it had literally lifted the back of the Jeep off the jackstands. Doing the math, that would equate to something like 2,000 foot/pounds of torque. Don't want to have to do THAT again!

And yeah, with used bearings it looks like the general consensus is to cut the target rotational force by half.

I would have done this job a month ago, but was a little under the weather just before I had to leave for a few weeks, and since then I've had lots o' family in the house. All good, but delaying progress - and worse - allowing the triple digit temperatures to creep up on me (fortunately, I'll be doing this in my garage, which does have a small A/C unit).

 

Haha! Same as the W126 wheel bearing! The difference between .04 and .06mm is about 20 degrees on the hub nut, hard to control when you're bending a 6 foot pipe to turn it.

 

Well, THAT was fun! ;-) I finally got the time and opportunity to re-rebuild the transfer case... replacing both the munged front output shaft seal, and setting up the bearings with the proper preload. In the end, I wasn't able to work out any realistic way to measure the actual rotational drag (which correlates to preload) but was able to work out a proxy that allowed me to get the preload dialed in to the point I'm happy with it (and confident that the bearings will last a good long while).

I know that the preload on the bearings from the original rebuild is much higher than ideal, but after 2,000 miles, the bearings and races looked fine - clearly you can see where the rollers were running on the races, but all surfaces are smooth, so I didn't swap 'em out (anyone need a bearing set?). ;-)

I did a video on the process (what a surprise, huh?) concentrating on the parts that weren't in the original video (dropping and reinstalling the front propshaft - ugh), and put in my best attempt at explaining preload and rotational drag in the middle of the video.

I hope this adds to the knowledge base on the subject, and helps those with transfer case issues keep their vehicles on the road for a long, long time...