I may be completely talking out my arse but here's my thought.....
What if you considered it similar to the input pinion shaft like an automotive differential and used that as guide for the pre-load? Seems to me it might make sense because we're talking in ranges of the same high RPM. But again.....I'm way out of my league here.
Your arse and mine are on exactly the same page.
The Dwagon has a fair bit of mech experience so I'll wait for some input from him but as I'm beat after painting in the heat earlier and some hours in the workshop it's time for some now after a stiff nightcap.
See ya's all tomorrow when yea I can finally collect a shipment after takin 2 forking weeks to get through our totally retarded Auckland ports.
As these are plain helical cut gears, not hypoid like a car differential, you aren't going to set the preload the same way. The process for that is different now anyways; while we used to have to calculate shim stacks and offset the pinion based on gear clearance and correct contact region, most manufacturers have moved to a much simpler single shim based on contact area between ring/pinion, then the actual bearing preload is handled with crush sleeves. It's been decades since I last rebuilt a Detroit Locker and had to calculate shim sets...
Anyhoo... you're working on a rovatti T-series, right? T-80 size or somesuch?
http://rovatti.ricambio.net/site/pagece5.wplus?ID_COUNT=ce_5_home&LN=2&CEPV=Rovatti001&CELN=2&MP1=CE_3062&NDS=CE_3062&PRC=|R|CE_3076|CE_2|CE_3058|CE_3062&PRF=4#CE_3062I tried looking up the exploded view so I could see the preload mechanism; that will usually give you some indication of whether it is intended to have actual preload, or cold endplay. Unfortunately, all the pics are horrible resolution, and after looking at a dozen different ones, all I saw was models that use only ball bearings. That may be due to dumb luck, or may be that they've moved to double-row ball-bearings instead of tapered bearings and your model has been superceded.
Thermally, this assembly only has to deal with ambient plus friction losses, and it is all ferrous metal with oil-cooling, so expansion should be pretty even and not huge. As there is near-zero deflection load (only a relatively slight, constant load presented by the helical cut gears), actual preload is not as critical, so it is better to err on the loose side rather than tight.
That said... if it were mine... I'd probably do it the "rough hand" method; tighten the preload down until you feel drag when turning the pinion, rotate a couple full turns, and as long as you don't have any binding spots (if you do, do NOT hesitate to tear it down and clean the bores/reseat the races), back it off to where you can just detect endplay, then lock it down at a point halfway between the two.
That would be my
"err on the side of caution" or
"git 'er done in the field lying on muh back in a puddle of mud, etc..." game plan.
Oh schizz... you already have the drive side installed. That complicates things... you can't "feel" the bearings with that interference. Yeah...
unless you feel like dismantling the drive side, you may want to contact rovatti for access to their tech reference archive:
https://www.rovatti.com/technical-documentation They probably have a recommended procedure that's better engineered than my plan anyways; there are so many practical methods dependent on application. My guess would be either actual end-play with feeler gauges or dial indicator, or the method you suggested; wherein you place a scale at a given distance from the centerline of the set-screw and measure, or use a inch/oz torque wrench to apply preload.
Here's Timken's word on it; including the basis for most automotive tapered wheel-bearing adjustment:
I've attached the document that is excerpted from below; mostly it discusses differences in approach from service standpoint, where goal is to spend a few minutes performing a specific technical procedure (and how to choose that) to minimize likelihood of failure in a specific installation, vs manufacturing standpoint, where goal is to find a method which requires little or no technician skill, but yields fewest failed units in a group over time. A bit dry reading, but it does shed a lot of light on what many manufacturing choices that more mechanically adept folks would consider to be a
" were they thinking...?" design choice.
Cheers,
mnem