Here we remove the steering box from a 1997 bmw 840ci and replace about two dozen seals.
The motivation was to confirm a steering failure diagnosis that documented in another thread.
Earlier boxes and 850's will differ slightly.
This small task was made into a big challenge simply by the refusal by original manufacturer (ZF) to supply rebuild kits directly to end users.
The seals used herein are now precisely documented and in theory could be sourced independent of the oem.
The procedure shown below is not a complete rebuild because...
- My seal kit and documents did not cover the servotronic control.
- My tools, skill and kit did not cover the removal of roller bearings on the shafts.
- My documents did not reveal how to service the torsion control valve section of the input shaft.
Vocabulary
Remove perimeter bolts on output shaft cover. Manage/track the washers buried by paint.
Remove locknut (17mm w/o paint) from lash adjuster on top of the cover.
Break paint seal (say with box cutter blades) and turn the lash adjuster (5mm hex) to pull the cover away from the housing.
Inside you will find...
- approx 1-1/2 cup (350ml) hydraulic fluid.
- the output shaft
- the rack
Remove the orientation cap on the input shaft.
Here opposing chisels unstick the cap from hardened gunk.
Unbolt the input shaft housing.
Break paint seal at either the input housing and the intermediate housing.
Remove the input shaft and rack.
The control section of the shaft contains about 1/2 cup (100ml) of hydraulic fluid.
Separate the rack off the input shaft by rotating the input shaft (left-handed action, CW to remove).
Listen as the 23 balls fall and collect at the bottom of the rack.
**** Seal replacement ****
Time: 20 min for pro with organized inventory. 2hrs for the confused beginner.
A seal kit was obtained, definitely without any thanks to the oem.
The part numbers shown in pics below were assigned arbitrarily as I inventoried the kit and wrote out a spreadsheet.
The output cover, using magic tweezers.
- Oval seal, inside #7
- O-ring, outside #11
The roller bearing on the cover prevented replacement (by me) of...
- Yellow seal, #6
Housing seals at outshaft lower end.
- Oval seal, inside, #7.
The roller bearing on the housing prevent replacement (by me) of...
- yellow seal, #6
- blue seal, #22
Input shaft.
- Bands, #3
- O-rings, #9
- Band, #4
Original bands, layered over o-rings.
Check valve cover
- O-ring, #15
Input housing.
This spring seal was not replaced due to tooling and priority.
- Spring seal, #23 (not 22 as per typo in manifest).
Input shaft cap, #24.
The pointer is an extension of the tick on the shaft, covered by the U-joint during install. Shown not seated yet.
Intermediate housing, control side
- Band, #2
- O-ring, #8
- O-ring, #10
- O-ring, #16
Intermediate housing, rack side
- O-ring, #10
- O-ring, #16
Rack
- Band, #1
- O-ring #8
Rack recirc cover
- Band, #5
**** Assembly notes *****
Position the input shaft in the rack with approximate 6cm separation.
Insert qty 16 balls into hole A and rotate the input shaft until they appear at hole B.
Be alert for jamming, else start over.
Fill the recirc pipe with the remaining 7 balls and insert from A to B.
Fyi: The top shell has been withheld just for this pic, but this sequencing actually won't work. The balls interfer with the insertion of the half-shell.
The successful technique is to assemble and load the the recirc pipe,
then keep the rack and pipe almost level so that no balls fall out on either side,
then bring the pipe and rack together.
The hi-side hydraulic line must be installed blind. There is no access after box installation.
It has been demonstated that coming staight out to the side provides the correct orientation to mate with the regulator.
Shopcraft while we are here: This note was written six weeks prior, telling myself about the state of the project.
**** Install notes ****
Both the pitman arm and steering column U-joint have an rotation alignment according to tick marks on the shaft ends.
The convenient proxy on the pitman arm is the flat on the shaft where the pinch bolt passes.
Then raise the subframe, trailing end first until the box clears the V8 mount.
The column U-joint aligns with the pointer on the input shaft cap.
Last edited by Hyper; 06-27-2018 at 03:43 PM. Reason: Removed pull preamble
Excellent, thank for writing this up. Bookmarked.
Great information! Thank you.
Does this rebuild improve steering feel or reduce gearbox slop at all, or does it just seal up fluid leaks?
Thanks a lot !!!
What a marvellous write-up, really detailed and excellent photography, Thanks!
Timm..2007 E64 650i Individual Sport..1999 E31 840ci Individual Sport..ex owner of 2000 E38 740..1999 E38 740i V8 M62..1998 E38 735i V8..1993 E32 730i V8..1988 E28 518i
My BMW Repair YouTube Channel
My Current 840ci Sport Individual
My Current 650ci Sport Individual
My E31 Repair and Information Website
My E38 Repair and Information Website
My E63/E64 Repair and Information Website
Chase - Heroes to a generation
Great job
Outstanding write up! Weird you needed the rebuild because the box seems to be in very good clean shape on the outside. But of course, it's what's on the inside that matters.
CB42366 - 1991 850i 6-speed. Brilliantrot & Black Nappa Leather
CD00144 - 1994 850CSi. Hellrot & Black Nappa Leather
My other projects:
Supercharged Tbirds, V8 Tbirds, V8 Mustang Convertible, Audi V8 Quattro & Audi S8
That's my aggravation.
I could not and can not find any to measurement or observation that says this box (or seal) is good or bad.
Just now I got back from my first test drive following the swap out of a used working temp (#2). The rebuilt original (#1) now works perfectly.
Before, it consistently failed to provide assist and seemed to bypass so much fluid that the brake assist also went away. I have a theory, but that's typical.
Still waiting on a 2K psi sensor, to be continued on the original thread.
Nice job. Thks
95 840Ci, Calypso Red/Silver, 83k - 99 740i, Black/Black, 185k, Alpina Mods - 01 740i, Titanium/Silver, 40k, Sport Pkg
Complete Turnkey Diagnostic/Programming systems available...PM for details.
http://www.8Coupe.com - Please register, spot, and help the project.
Thanks!
Very nice write-up. Never had any idea they had so many seals.
CSi #18 - Car & Driver Magazine 1994 actual test car
-- Hellrot/Black-Gray (1 of 1 NA CSi color combination)
BMWCCA E31 Chapter International Clubs Liaison
North America Representative, 8er.or Board of Directors
Seals fall into only 3 function categories.
- Piston
- Control chambers
- Housing & shaft ends
[professor-style]
Recirculating ball steering povides assist by pushing the rack (a piston with teeth on the far side)
along the worm gear, attempting to minimize torsion between the worm portion of the shaft and the outer input shaft.
A thinner/weaker torsion bar puts less burden on the driver before the control valves apply assisting force on the piston.
In the pic below the letters A->F indicate 6 pressure zones created by the seals with zone G being outside/atmospheric.
Applying pressure to the piston in zone B above that of zone A will assist a right turn, and applying pressure to A above B assists to the left.
A band-over-ring on the piston separates zone A and B and notibly the band around the recirc cover does also.
If this seal were to fail then not only would there be no assist but presumably fluid would flow directly from source (zone C) to drain (zone F).
The chambers C->E allow control pressures to be passed from the rotating shaft to-and-from the stationary housing.
For instance the Servotronic servo could create an intentional leak between control chambers.
Finally, ordinary o-rings and shaft seals sorta keep hydraulic fluid from leaking to the outside world.
[/professor-style]
Last edited by Hyper; 07-12-2017 at 06:16 PM. Reason: Pic link repair, re PB affair
Hat's off to a great write-up, thanks. Were either of those kits by themselves completely able to do the job? If so, which one?
If I'd been a little more alert, I could have sent you an inside puller to get those bearings out. A cheapie Harbor Freight job ought to work OK.
You can also check the recirculator ball diameter and, to a lesser extent, the wear on the ball track. The balls start out at 7mm. If both the balls and the track are a bit worn, you can try replacing them with 9/32" balls, which is nominally 7.14m. There's old forum chat about this being a way to tighten up some of the slack.
How come the middle half of any project always takes the most time?
Yes Mr Lumens, thanks for contributing all the little goodies that made this project happen.
I ended up using or at least identifying a home for 27 bits, thus leaving me with 5 and 9 mystery parts in the two kits.
The mystery parts were sufficiently large that they could only go into the torsion valve section which I (literally) failed to open up.
The other mismatch occurred when I dived into the servotronic switch and found lack of support in the kit. Not surprising as not all boxes have it.
I hope you push further on kit development for the great masses of people who will now follow.
To that end, I now have box #2 on the bench and I'm willing to tear it down to confirm this or that.
I would need docs or at least a good tip on the torsion valve section and also seals for the servo.
*****
Box #1 was never leaking to the outside, so I wasn't concerned with replacing housing and shaft seals.
But as for the shaft seal, my mind was actually fixing on the roller bearing being staked into place.
Also, if I pull the roller bearing to replace shaft seals, where is my replacement roller bearing?
So I took a pass on shaft seals because it didn't meet a basic risk/reward calculus.
Could this repair a worn box. I have a huge slack in mine. Huge one
What a stunning write up
Sveinbjörn Hrafnsson
E30 CABRIO V12 M70B50 ///
ALPINA B10 BITURBO 346 @ 507
ULTIMA Evolution
Contacts
http://alpina.123.is/pictures/
Weren't the blue and yellow seals at the bearings 100% polyurethane, ie no steel form underneath the poly and no spring gaiter? I'm just wondering if there's any way to pull them without removing the bearings.
How come the middle half of any project always takes the most time?
I don't see that happening. This contrasts with the silver oval-sections which were very pliant and, for removal, the outbound set could be stabbed into by the magic tweezers.
[professor-style]
***** Lash adjust *****
aka "tightness"
The teeth on the output (sector) shaft are beveled in the axial direction.
Pushing against the top cover, the adjuster bolt forces the output shaft down onto an opposing bevel on the rack/piston.
That by itself would adjust lash at this junction.
This would force the piston to rotate in the bore if not constrained by the housing.
The piston may (!) be forced towards the input shaft, taking up slack against the recirculating balls.
System level drag increases as we clamp down, requiring the hydraulics to work harder.
[/professor-style]
Last edited by Hyper; 07-12-2017 at 06:22 PM. Reason: Pic link repair, re PB affair
You might be surprised. You can double fold a poly seal and work it into some improbable bores. I had poly u-cup seals made for the input shaft on the brake boosters, and you could stuff them into a 22mm bore. I don't have any more of the steering box seals at the moment so I can't compare.
How come the middle half of any project always takes the most time?
Now that my front end is rebuilt, I can feel bit of play in my steering wheel. I think you (Hyper) even made a sly comment about it when I saw the freshly rebuilt gear on your bench and I noted I was glad I didn't have to dive into mine because mine was fine!
You said:
"The teeth on the output (sector) shaft are beveled in the axial direction. Pushing against the top cover, the adjuster bolt forces the output shaft down onto an opposing bevel on the rack/piston.
That by itself would adjust lash at this junction."
So, if I am understanding you, if you screw the allen bolt "down into" the box you can remove some of the play, downside being increased friction?
This all depends, I suppose, as to whether a standard or reverse thread is used on the allen bolt, which clarifies screws "down into" as being a clockwise or counterclockwise turn as viewed from above.
Is there any official spec for the gear lash adjustment?
1995 BMW 840Ci
1999 Porsche 996 Cabriolet, 6 speed
1974 VW Westfalia P27 Deluxe camper
1974 Mercedes 280C
This writeup does not attempt to cover testing or adjusting.
You would have to hit up someone like Mr Lumen, who has spent much more time with these boxes and suppliers.
However, I recall him saying that special tools would be needed to verify factory specs.
I merely put my box back together the way I found it.
Between the road and steering wheel there might be a dozen sources of "play", and I can quickly think of three inside the box.
For instance while the assist was out, I got quite familiar over a 1500 mile stretch, with the 2 degrees of steering wheel slop before the torsion bar hit the stops and I could move the rack.
The adjustment bolt takes care of the rack-outshaft junction for sure, but may (?) also tighten up on the recirc balls. I just won't guarantee it.
Yes, the adjusting bolt has standard right hand threads.
By design, the outshaft can move up and down, over a range of say 15mm, and turning the bolt CW as viewed from above would push the outshaft down onto the rack teeth.
Hyper. If my box is worn and has a slack can I fix that?
Also, I have a RHD CSI box and it is supposed to have different ratio. Could I swap the internals?
No and no.
You're a student and don't have time.
So the RHD is useless for me?
Time is all I dont have during the winters, but maybe I could find some during the summer
Does anybody know if E32 box fits E31? I know that ratio is probably different, but are they physicaly the same?
Bookmarks