Citroen drive shaft oil seal alternative

[910Esprit]

The original oil seals used on the drive shaft housings are non-standard insofar as the ID does not bear directly on the axle, a side lip bears axially on an 'oil flinger', I suspect as a failsafe design feature to manage potential oil leakage.  (as the rear discs are perilously close to the drive shaft housings)   The result is preposterously expensive proprietary seals (eg approx £100 for 2 delivered).   This seal has an 65mm OD and 8.5mm width    (This is a standard theme with these boxes, that are also full of 'unusual' bearings).

The dimensions of the shaft, housing & spacer width are 38x65x8.5  Standard metric NBR R23 rotary shaft seals are available in 38x65x8   (£6.80 for 2 delivered...)

From examination of the driveshaft housing and axle, I cant see any obvious reason why a standard seal could not be used.  The OD would have positive location against the outer race of the bearing and the oil flinger would still be in place to deal with any leakage, or if oil seeped directly into the housing it would be collected by the valley in the casting and leave via the 'tell tale'.

Has anyone else tried this mod, or can anyone see any reason why it is doomed to failure before I trial it? 

Edited May 18, 2022 by 910Esprit

[910Esprit]

Whilst waiting for the seal to arrive, I've turned my attention to the drive shaft bearings, which, quelle suprise, are virtually unique to Citroen and would cost near to £100 each delivered.    I did however note that SJ sell them for around £39 plus the usual extras.   How so?

The dimensions of the bearing are 30x72x23.8, which happens to coincide with a common version of a cam roller bearing (has a much  deeper outer ring than a standard automotive bearing and is typically used in industrial applications like conveyor systems).   This is clearly pictured on SJ's website and is probably a LR5206KDD equivalent  (you need to be careful with this bearing as not all versions have a parallel outer - some are convex/concave). 

It would seem that such a bearing is a satisfactory replacement for a standard double race angular contact bearing and can tolerate axial and radial loads.  Unbranded versions start from around £20.   

(Id probably be less comfortable using it on an S1 or S2 where very high axial loads will be transmitted from the suspension.

[910Esprit]

This has now morphed into a 'budget' (a very relative term on these boxes) rebuild.   As this is on a spare box, I can pretty much take it one component at a time and see what the options are.

The most expensive bearing in the box is the pinion support roller bearing.  Its a handsome bearing , 41x80x18, seems unique to Citroen and over £300 delivered from SJ.   I suspect the one I have removed accounts for some of the noise from this particular box.

I've just imported one from a Citroen spares place in Europe, which still cost a hefty £130 delivered.   And again, it looks like a creative aftermarket solution.  Its marked NJ208, which is a common bearing sized at 40x80x18.  So I assume the supplier has had the ID ground from 40 to 41mm.  If that's the case, they have done a nice job.

Another observation is that all my spare output shaft flanges (4) are not flat.  The worst was almost 20thou out of spec.  I wonder if it was salvaged from a crashed car?   Anyway, the new press makes an excellent 'vice' for holding the shaft vertical, while I 'tap' the flanges back to spec.  Definitely worth checking if your rear discs are running out.         

Still waiting for the radial driveshaft seals to arrive as first one sent was supplied incorrectly

  

[dsvitesse1]

HI here is one of those imitation pinion bearing.

 

https://www.facebook.com/harry.martens.7/posts/pfbid02ENZXABTjwqGw1bCC3AuUTgxJysz7J8R3n7XDTSdSzLbcRmwjYqw1yDZmJ6ehHJNZl?notif_id=1653557486312733&notif_t=feedback_reaction_generic&ref=notif

 

 

beware.

 

cheers,

 

Harry

[910Esprit]

Interesting...   Looks like it failed at the point it was pressed onto the pinion shaft?    

The first driveshaft housing has now been reassembled with a 30576 2RSR bearing (with the rubber seals removed).  With radial 38x65x8 oil seals used.      

   

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[910Esprit]

Put the other driveshaft hub together today.  Used a NOS genuine Lotus bearing I've had for years and a new radial seal.  Total cost of today's effort £3.42.   Not a particularly conclusive test, but the hubs are now sitting full of oil to ensure no static leakage.

Also looked at the final 2 unique stepped double row bearings that sit at the rear of the box between 4th gear and the 5th gear extension housing.  In addition to supporting the primary & pinion shafts, they also control fore and aft movement of the shafts (i.e zero).   I washed them in paraffin, examined them and held them in the press with the split centre sections pushed together, to mimic the installed condition.   Put a dial gauge on them, which confirmed zero axial & radial movement, so they will be reused.  That's over £500 saved..... 

Remaining primary shaft bearing and diff support bearings are a common size and even SKF versions can be found inexpensively.   

Next job is to clean all the remaining internals.    

[910Esprit]

On removal, I distorted the very feisty snap ring that secures 3rd gear to the primary shaft.  Turns out it is selective thickness, which controls the end float of 3rd gear.  SJ's price is over £50 delivered for this snap ring!   Supply and demand I assume, but even if Id won the Euro Millions I would not pay such an absurd price.

Anyway, 2 mins on Google has located the correct item, seemingly readily available, in Europe starting from 2.5 Euro....   (plus potage).   

So the plan is also to size up the selective washer that controls pinion height and with a bit of luck I may find both of these from the same supplier. at a sensible price.   

[910Esprit]

So, plan C was to dismantle another spare box, so I could do a direct comparison of a few bits.  However, there is no way my adapted 'professional' open ender is going to get the speedo worm/pinion nut off the pinion shaft.  I can feel it flexing as I'm leaning on a motorcycle fork tube extension handle - This is the tightest one I've come across.

I have a plan for a new version, which will start life as a very large socket - Will need to wait a few days for that to arrive. 

[910Esprit]

The 'nut' that secures the pinion shaft assembly doubles as the speedo drive worm gear requires a 36mm AF 'spanner' no more than 5mm wide.   This nut can vary between not very tight, to extremely tight.   The one that I've just removed was way beyond the capability of a viscous fan spanner or even my modified forged open end spanner, so a new approach was required.

A 46mm socket is a nice sliding fit over the worm drive.  I then created 2 hardened steel rollers of 8.25mmx5mm from a drill bit shank.  These were then placed on the 2 flats of the 'nut' and the socket slid onto them.   This then created a roller clutch to allow as much torque as I needed.   Still too tight for the rattle gun and 1/2 drive was flexing with an extension bar applied.  So used some good quality Stiltsons on the OD of the (unchromed) socket, again with a large extension used.   This did the trick and the pinion nut unscrewed with no damage.   The process will be reversible for assembly and allow a torque wrench to be used.       

Just a matter now of deciding which parts will be replaced and which box to rebuild.  

Edited June 6, 2022 by 910Esprit

[910Esprit]

Acquired a depth micrometer and slip gauges to allow a bit more precision than my previous vernier caliper pinion height measure.   I can confirm something that Georgio previously mentioned on another thread, which I'll admit I was a little sceptical of...  However, I can empirically confirm that the centreline of the diff bearings is not aligned to the machined face of the diff housing!   I have measured 2 sets of cases and in both instances the centreline is approx 15 thou nearer the pinion than the machined face.   As the factory tool and pinion depth uses the bearing centreline as its datum this is fine.   Pretty much everyone using a non-factory tool will use the machined face as their datum.   This will clearly result in a significant error if its the only measure used.   Now I've calculated the offset, I can use the machined face of the casing, but need to subtract an addition 15 thou (converted to metric of course)

[910Esprit]

Done loads of relative measuring while waiting for a few parts (using a depth micrometer and getting fully repeatable results).  here's the findings:

The centre line of the drive shaft axles (ie the centreline of the differential) is not perpendicular to the machined face of the bellhousing/gearbox joint, by 1.5 thou, when comparing one side to the other.  (seems a bit crap)  But what I cant accurately measure is if the shafts are true to the machined face or the centreline of the diff.   But I'll assume the axles and the diff are perpendicular for pinion height calculations.

The overall difference in measurement between the depth of the diff centreline bearing cutout in the gearbox casting and the depth of the diff centreline bearing cutout in the diff casting is 14thou.  (The bearing cutout being deeper on the gearbox/pinion side).  So I need to add 1/2 that value (7 thou) to my pinion height calculations to logically align the centreline of the diff with the machined joint between the gearbox and the bellhousing casting (not subtract as I previously mentioned).  Otherwise crownwheel would be 7 thou too close to pinion.         

Next step will probably to reassemble the pinion shaft only and look at the contact pattern.

 

Edited June 17, 2022 by 910Esprit

[snowrx]

I need a diagram to understand the .014" offset becoming .007". Not enough caffeine yet this morning for me🥱

[910Esprit]

Haha - yeah, you need to be pretty hardcore to understand it without pictures!   TBH I've been confusing myself regularly with the parts in front of me.  I'm really posting here for my own benefit, but hopefully anyone doing a DIY rebuild will now be aware of the pitfalls, even if they have no idea what I'm on about!

If you picture (in your mind....)   The bellhousing casting and the main gearbox casting, each has a semicircle milled out to accept a 80mm diameter 30208 bearing to support the diff (Theoretically 40mm depth for each semicircle when measured from the machined faces where the bellhousing and gearbox joint is.

However measuring the depth of east bearing semicircle, measures greater then 40mm on the pinion side and less than 40mm on the bellhousing side (.e. the pinion side is 7thou too deep and the bellhousing side is 7thou too shallow.  (so if I added the full 14thou difference I just switch from one error condition to the opposite error.

NB - As belt and braces I'm also going to make a 'slip gauge' from a section of silver steel, at the length etched on the pinion, less 40mm (as I will measure from the bottom of the bearing cutout, not the centreline.  This will allow me to rest a ground round bar across the pinion side bearing cutouts.  If done correctly the slip gauge should just contact the ground bar when positioned vertically on the pinion.   This should give a true reading without needing to reference the machined face at all       

 

 

Edited June 17, 2022 by 910Esprit

[rgh0]

When the gearboxes were made the bell housing and gearbox casing would have been bolted together and then the 80 mm hole bored.  Nominally this bore would have been centred on the split line but it appears there is some ( systematic?) small error in the machine setup used to do this boring so the centreline of the bored hole is moved 7 thou from the split line to the gearbox side.

This type of tolerance / error is not uncommon in machinery fabrication which is why it is always wise to check the location of the centre line of bored holes precisely as done, rather than assume the split line is also the centre of the hole

 

cheers

Rohan

 

[snowrx]

Yes, I've been in there before doing a Quaiffe and bearings. I could visualize the .007 off center bore and the need to compensate pinion height, just lost me on the doubling to .014. Someday I need to swap in my CWP set and I want to be prepared.

[910Esprit]

A couple of tips....

We  all know the box was shared with the Citroen SM.  But the 5 speed version of the far more common Citroen DS is virtually identical too (maybe it has different ratios, I've not checked).  

The DS parts catalogue has far better quality exploded diagrams than the lotus manual and also contains really useful dimensions of the bearings, clips, springs, washers etc.

http://www.citroen-ds-id.com/ds/files/D_spare_parts_catalogue_648_vol_1+2_june_1974.pdf

You may also find much cheaper genuine parts by using the Citroen part number, but searching on 'Fiat 242', which used a 4 speed version of the same gearbox and shares lots of the sundry parts, but retains the Citroen parts number.  If that fails, the Citroen DS is well served by a network of suppliers in the UK and Europe.  It also pays to shop around as there is a massive variation in prices   

 

 

[910Esprit]

I've always thought the Lotus manual was pretty vague on a few aspects of setting up the diff correctly - It has now become apparent why:  Lotus have copied content directly from the Citroen DS workshop manual, but have not used the many original photographs of processes and various tools being used that accompany the text!  My advice would be to ignore the Lotus manual and just refer to the original Citroen source.  Its easier to follow and makes more sense.

[910Esprit]

New input shaft from Lotusbits as the old one was pretty tired from the fretting corrosion that seems to occur at the primary shaft splines  (seen this on a number of boxes).  Circlip was also on its last legs..  Also a small package of sundry items arrived from Italy, which were many orders of magnitude cheaper than the UK.   

Diff now set up.  The overall pre-load is not too difficult and is within spec using both methods detailed in the manual.   However getting the backlash within tolerance is a nightmare, as it is so sensitive to the pre-load, so things like the torque of the housing bolts significantly affects it, as must any gasket sealant used, so a small amount of judgement is used.   Its also tricky to measure the true backlash with the DTI probe positioned against a tapered curved surface (a tooth), not at 90 degrees to the probe. As a final check, I superglue a nut on the edge of the crownwheel, this allows the DTI probe to rest against a flat surface perpendicular to the differential centreline.

Made my own 'secret recipe' marking compound, which was just a tube of acrylic paint from the hardware shop, thickened a little with corn flour...   Happy with the pattern, which looks good to my eye.

Think I now have everything for final assembly

[910Esprit]

Put the box back together, which is straightforward.   The only tricky thing is holding the box while applying a lot of torque to the Primary & Pinion shaft securing nuts.   That's achieved by holding the casing in the press and the selecting a forward gear & reverse simultaneously to stop the shafts rotating.

So far so good, especially as I'd previously selected the shims for pinion height, end float & backlash.  Now, for some reason unknown (although I suspect worn Citroen tooling), the pre-load is going out of tolerance when I bolt up the bellhousing to the box.   I can double check this by removing one of the drive shaft housings and measuring the bearing depth, which I'll do in the next day or two.

 

[910Esprit]

I must have the diff apart/rebuilt 10 times.  Actually have done the final assembly 3 times, with gasket cement etc.  Still not happy, as the pre-load always measurably increases on final assembly.    

Anyway, I'm now certain (again...)  that I know what's going on.   I am equally certain that the 2 main transaxle castings gearbox/bellhousing were not machined as a single unit.   Presumably the Lotus specific bellhousing casting was made as an independent thing and united with the gearbox after initial manufacture.  (This is evident in the line bore for the diff bearings, where there are clearly unmatched machine patterns on both castings.   

I believe the pre-load issue occurs as the bellhousing flanges are 10 thou narrower than the corresponding gearbox flanges when measured from driveshaft  'flange to flange'   As the calculated pre-load is measured from the gearbox flanges before you assemble the bellhousing, everything measures fine, but when you bolt up the driveshaft housings the narrower bellhousing will allow an increase in the preload.   

Haven't fully decided how to address this without having the 2 castings matched.   Could possibly subtract 10 thou from pre-load calcs, or more likely to introduce a 10 thou shim/spacer to make the bellhousing/gearbox the same width.      

 

[drdoom]

Not long ago Harry Martens @dsvitesse1 posted somewhere in TLF how remarkably tight were the tolerances in manufacture of these 'boxes, adding that he ships his units with a spare paper shim gasket or two, if I recall correctly. FWIW

[910Esprit]

I've overstated the discrepancy, which is nearer to 7.5 thou, but as the specified pre-load is only 12 thou, its bound to have an impact.

All measurements I have taken are using bona fide engineering tools (engineers square and calibrated depth gauge for the above) so I'm confident that these are material discrepancies.  Its entirely possible these castings are the exception to the rule.  I have been mulling over the ide of switching to another set of cases, but will probably persevere with this set for now, as they have previously been fine for circa 75k miles.

[910Esprit]

Well there we have it! - Note split gasket.   Happy that I was on the right lines

 

[rgh0]

I find it hard to believe that the output shaft bearing supports between the GB and bellhousing were not machined as an assembly.

May be not on the centre line of the split or on the centreline of the diff carrier but i believe impossible in a practical sense to do separately rather than as an assembly.  A measure of vertical versus longitudinal diameters would confirm if not a a true circle  - but hard to do.

Certainly if on assembly the output shafts are being moved as the centre line of the output bearings is a few thou different form the diff carrier centreline the preload would be affected

 

cheers

Rohan

[910Esprit]

I would wager they were machined separately.   That's not to say there was not some final process to process the bearing cutouts to the finished size.    Apart from the witness marks and inconsistent measurements, a 'smoking gun' would be the fact that the gearbox and bellhousing are not dowelled.  In fact the bearing cutouts are subsequently used to pull everything into alignment on assembly.