Rebuilding an S4 T2 Differential The s4 diff seems pretty straightforward to set up. That said, I had a couple of strange findings in mine so I found that following the FSM verbatim yielded some strange measurements?
I started by pulling out the stub axles and taking off the cover exposing the diff. If you don?t have a factory service manual (FSM) you?ll want one and there are links at
TeamFC3S... Second Gen RX-7 Performance Site.
The backlash between ring and pinion was perfect, but I still seemed to have play in the stub axles so I decided to dig deeper and pulled the diff.
There are a series of measurement in the FSM that result in two tolerance stack ups relative to the differential internals.
-One tolerance stack up measures all the clutch plates, springs, and the two halves of the spider gear housing. This is responsible for friction. My numbers were a little on the high side of the tolerance but still acceptable so I just switched a few plates around to give things fresh friction surfaces (not that the existing ones were in seriously worn, but why not)? and called it a day.
-The other tolerance results in clearance to the spider gears. You use thrust washers to define total backlash of the internal gears. The FSM is written such that you take 94.00 mm and subtract measurement to yield your total clearance. (Note that the FSM never refers to this as backlash but as near as I can tell that?s what?s going on.) If required you can buy thrust washers in +0.1 or +0.2 mm (Mazdatrix carries everything you?ll need).
Here?s where it gets weird. My total tolerance stack up for the spider gears measured 92.40 mm. That?s 1.60 mm clear on a tolerance that?s supposed to be 0.10 to 0.40 mm. That?s a HUGE variance.
I triple checked that I measured it correctly, I checked for appreciable wear in all the pieces and found none. I even double checked the calibration on my calipers. So then I measured the total space available inside the two halves of the diff itself (in case there was a miss print in the FSM and it was supposed to be 93.00mm minus your measurement). Total space was found to 94.13mm.
Turns out I actually DID have about 1.73 mm (.068?) total clearance in the inner housing stack up. I reassembled the diff without the clutch plates so I could measure just the spider backlash gap. You can see it in the pic here and I confirmed with feeler gauges.
This is what caused the perceived slop in the half shafts. Unfortunately it wasn?t something I could solve +0.2 mm thicker than normal thrust washer on each side. Instead I took some more measurements and ordered shims from McMaster. You can buy steel shims in any thickness your heart desires using a 1 7/8? OD, 1 3/8? ID, which is close enough to the factory thrust washers that it should work fine.
There are a couple things to be aware of. First when you measure the overall clearance with feeler gauges you need to be aware that total height of the spider gear stack up changes as it rotates. With everything tight to iron at all times the stack up rises and falls by 0.025? as you turn the spiders. This means that although I measured 0.068? clear I really only had 0.043 clear.
I had bought some thin 0.004? shims with the logic being that adding one 0.004? on each side changes the total tolerance by 0.2 mm so it?s a small enough step to put me in the target 0.10-0.40 mm range. The problem is that this is a wear surface and I wasn?t sure I?d trust a thin shim in a sliding application never to bind or rumple because the surfaces aren?t fully supported. After playing with shims for a good while, I finally bought 0.015? shims and installed one on each side. McMaster sells a 0.020? shim that I debated about but decided that 0.003? was too small a clearance to try to run?especially considering I wasn?t torquing the ring gear on while taking measurements so I might lose some tolerance with the assembly fully torqued down.
By the way, the Spider Gear Outputs (the splined pieces that the Stub Axles insert into) only have ~0.042? play on each side until they bottom out. Essentially if your backlash was REALLY loose the spider gear outputs would rub on their housing instead of floating inside like they?re supposed to. My 0.068? (0.034? per side) clearance was just enough to avoid rubbing hard parts. No damage but I?m glad I tore into it to check.
On final install I put my 0.015? shims between the diff frame and the existing thrust washer on each end so they can?t slip out of the notches in the frame. (If you put them between the thrust washer and spider assembly they might.)
This was a test fit with a couple shims...
Now it was just a matter of reassembly per FSM specs. My spring plates were installed backward when I opened it up so I fixed that (shouldn?t really matter given it?s just a belleville spring but this way should have a smidge more leverage and therefore harder bite). Backlash for the ring and pinion is set with threaded collars on either side of the main diff bearing races. You may have to make yourself a custom spanner wrench tool but that?s pretty easy.
It looks like moving one tooth on the bearing adjustors (procedure: loosen one side first then tighten the opposite) changes backlash by 0.002?. I left mine at the factory 0.004?
Bearing preload is measured via the total deflection of the two bearing supports. It takes a big pair of calipers so be warned that this might require a tool purchase. (I only had a 6? set.)
My diff has never whined or had appreciable problems beyond the slop I just solved so I didn?t replace bearings or get into the pinion crush sleeve. I did however change out the front seal because I was starting to show signs of leakage. That took a little playing with a three jaw puller set and some shims to keep the jaws biting but once set up it came off easy.
The front pinion seal has a nice metal lip on it so it?s easy to tap back in place with a hammer and a socket extension (used like a blunt drift).
The little custom tool I had was handy for retorquing.
You follow the FSM procedures on the dialing in preload on the crush tube but it's pretty easy.
Diff done, and that?s one less thing to think about.
