Assembling & alignment precision concerns

I've been reading here and other corners of the web for quite a few weeks and greatly appreciate all the wonderful info and guidance. That being said I've not seen much mentioned with regards to assembly and alignment, procedurally speaking.

Specifically I'm pondering how best to align SBR to frame. For the purposes of this thread I'm looking at a vertical Z column as on a vertical mill, or a vertical plate for a Z axis on a high sided router.

I have means to make a precision plate to within +\- 0.015mm tolerances realistically. .. basically the limits of the machines and standards with which I can make and measure. However when it comes to installing the rails and screw ends onto such a plate I am not finding much in the way of publicly shared info.

So... my thought is basically this example:

100mm wide 200mm tall 25mm thick cold rolled steel plate as a base to mount the rails to. 2 trued and squared edges.
Mill a reference edge or lip by skimming 3mm off the thickness Leaving a 5mm by 200mm lip vertiaclly. Perhaps going so far as to run an additional cut to slot 2mm deeper next to the lip so as to avoid the tool nose cutting edge radius - leaving as perfect of a 90* corner as possible.
Now install (hand drill and tap while rail is clamped here?) a SBR against this edge and "measure" to the next SBR using a gauge block of appropriate distance as a spacer and install the second rail.
ASSuming I can install the rails perfectly with out them budging from the spaced alignment I should have them dead nuts parallel and only need to install the base plate dead nuts plumb and vertical (for which intend to use my imperial 0.0001 dial indicator) to the frame.
At this point I can use the same process on the other rail plates (vert mill example) and use the vertical plate as a reference to square the axii to.

In best case I realistically imagine I can get the initial axis within +\- 0.01 mm, and the same for the other two axii... so when finished I may have upto a cumulative mechanical error or 0.06mm... not very good.
Which leads me to ask, can I really expect to work out that much error with software compensation?

I am in the end only working with part print tolerances of +\- 0.025 mm on a rare occasion once I employ my diy machine, but I'd still like to have the machine as accurate as realistically possible.

What does the community think...?

Thanks as always,

Nick

Ok well your thinking is sound and milling a Reference edge is correct way. Regards aligning the other rail then it's not has difficult has you may be thinking.! The correct or official way will use all kinds of precision measuring equipment but the way I'm going to explain works just has good just takes longer.

Simply mount first Master rail to ref edge and bolt securely then align opposite rail parallel the best you can either by very close measuring or using gauge block. (I actually have a long bar with locking slides that locate into counter bores of rails mainly to speed process up a bit.)
Don't fasten this rail down fully just nip up enough so it can slide but not lose and sloppy.
Now fasten the Master rail bearings to attaching plate and same again only tighten opposite bearings so they can slip.
From here it's mostly feeling your way.!!
Put bearings on rails and run up and down rails to align bearings then tighten bearings down evenly. Chances are you won't feel any tightening of the bearings on rails but check they haven't locked up.? If so then chances are your plates are twisted or not flat.

Now it's a case of running the assembly up and down the rail and tightening the rail bolts feeling for tight spots or any binding at same time. If you feel any binding then quick nudge of the rail in either direction will bring into line. If the bearings lock while tightening rail then it could be mounting surface isn't flat and you may need to use shims.
When I say shims then I'm talking very thin shims, like paper thin or less, can often make huge difference and remove it completely.

That's all there is to it really just careful alignment and feeling your way.

One other thing is that I see you mention SBR.? If that's the supported linear round rail then most of this will be irrelevant because the precision of them is not that great. Here I'm mainly referring to Profiled linear rails.
With Round rail then the tolerances are far lower so they don't bind easily unless drastic out or fastened to bent/twisted plates and has such take much less to align. Down side is they are not that accurate.!!

Being totally honest you won't get the tolerance's your speaking of from Supported linear round rail. The procedure is prity much the same just far less critical.

Thank you very much for the feedback jazz, as well as all the other wealth of info and experience you've shared.

I can appreciate that approach for ease and probable reliability.
I would love to go profiled rail, but that decision is a matter of budget trade offs at this moment - I am at the point where I have a digital frame laid out and overall design idea but fleshing out the final product specs. I'll goto a seperate thread when the time comes.

As far as alignment, play, and tolerances... I'm shooting for repeatability more then accuracy. So long as the tool repeats (reasonably) I can make it hold size.

Which brings me to my last question on this topic then: What have been peoples experiences with regard to compensating for assembly errors with operating software? How much slack can the pc take up reliably?