What's new
What's new

A scraping mystery

JST

Diamond
Joined
Jun 16, 2001
Location
St Louis
Now, this obviously has a reasonable explanation, but I don't see it yet. Sorry, I did not take pictures, it seemed easy enough to identify until it didn't.

I have a small benchtop mill, which a couple years ago I went through, scraping every surface for alignment, and doing other repairs and improvements.

I of course checked tram, and it was quite acceptable. I do not recall the actual measurement, but it was fine.

So, the other day, I was doing some work on it, and found that a piece being squared up and flattened did not turn out with the faces parallel after top and bottom were milled.

I checked tram, and yep, the tram was way out between the operator side of the table vs the column, a couple THOU, not tenths, lower at the operator side. It happens that the head does not nod that direction, so it has to be corrected elsewhere. This was a bit odd, as the tram had checked out before. So I kept investigating.

Method was a piece of shaft held in a collet, with a clamp on that carrying a 5 tenths DTI on an adjustable clamp. Spindle was swung to move the DTI over the table, or other surfaces in question. Standard stuff, and nothing but the DTI has to be accurate.

Knee was not wiggling vs the column, checked that. So I measured the 4 corners of the table, from work surface to the way surfaces. Distances there checked out as equal, to a couple tenths. For a tie breaker, I used a granite flat that I can lift to check the table, and could find nothing going on there. Blued up all over OK. Table provisionally eliminated as the cause.

So I pulled off the table, and measured the top ways on the "saddle". Same result, spots at all 4 corners were fine, and the surfaces checked with the flat. Double checked after removing the "saddle" by measuring the 4 places the two sets of ways are above each other. All the same within a small margin. OK, now the "saddle" is provisionally eliminated.

Pulled off the saddle and checked the knee ways. Well, THOSE checked out as in-tram using the same setup. They blue up OK. The knee ways check out as ALSO provisionally not the issue.

So, I reassembled the thing. First put the saddle back on, no issues, same result as before with the DTI.

Put the table back on, and there is the couple THOU again. Took off the table again, did not move the knee, adjusted the DTI down to the saddle ways, all fine, just as before.

So, I checked the distance from table top to ways, at 5 or 6 places along the ways, on both sides, choosing points toward the inside, as well as the outside edges, randomly. Again, I get a close match with a mic at every point. Ditto when I do it again.

Thinking like a fox, I checked the table ways for any evidence of a bright line where it might ride up on a corner that was not flattened enough. Nothing there, and I did not get a feeler gauge under the table anywhere.

So, I have proved that the scraping is fine, but yet the tram is out, when it does not seem as if it can be, or at least there is no reason for it to be "out". BUT, I get a very repeatable error.

The scraped surfaces all check out. I cannot find evidence of a "ride up" of a corner on any possible unremoved wear ridge. Bit it is "out" and not a tiny amount. I suppose it's possible that I bungled the original tram measurement, but even then, the surfaces and other measurements do not turn up a cause for an actual out of tram, but there is one.

Anyone have an idea of what I may be missing here?
 
Well, I cannot go back and take pics of what has been done. Not sure what they would show anyhow, just the setup and the DTI readings, really. I cannot reliably get a decent pic of a blued surface.

As for the gib deal, these gibs are screw adjusted from the side, the screws have pockets where they go. They are not taper fitted gibs with a screw on each end. I don't think that problem can occur with this type.

The issue is very repeatable, and does not move around. That's more evidence that it is not something shifting.

I though of another test, I can take out the gib, and shift the table over. If it really IS a "corner lift" problem, that should eliminate it. The table then should, at the least, show up a different error.
 
It sounds like the blued area is high in the middle. Did you rotate the straightedge? Ais your SE a camelback or parallel shape?. If it is a prism type they bend on a convex surface. I hate long prism straightedges
Just think if the surface is convex .002" and you don't hinge or pivot to find the airy points, the SE would blue up the surface acting like a rocking chair.
As fare as the gib goes, leave the inside loose. Or leave all the screws loose and with the DTI on the table, tighten the screws and see if it changes. Also blue up the table top. Those cheap import tables Cann warp or pean when you tighten the T nuts to tight. I see this on Bridgeport tables. On old used machines I've seen them as bad as
008".
 
Well, I cannot go back and take pics of what has been done. Not sure what they would show anyhow, just the setup and the DTI readings, really. I cannot reliably get a decent pic of a blued surface.
I just mean that it is hard to be sure my mental model of what you are saying is happening is correct without a photo or diagram of the machine itself. Seeing the machine can help to identify other possible sources of the error. I try to make quick illustrations of an assembly to help understand what is being observed, but if that's difficult to document I understand.

As for the gib deal, these gibs are screw adjusted from the side, the screws have pockets where they go. They are not taper fitted gibs with a screw on each end. I don't think that problem can occur with this type.
I've had serious issues with poorly made gibs of that type as they could kind of rotate around the tips of the set screws. Anything you can do to isolate the gib would be worth while I think. There are not a lot of other places where this could be going wrong.

The issue is very repeatable, and does not move around. That's more evidence that it is not something shifting.
Unless the shift itself is very repeatable :-) I've had carriages coming into unintended contact with a protruding object that cause an extremely repeatable deflection, it took a while to realize it was hitting a slightly protruding screw.

I though of another test, I can take out the gib, and shift the table over. If it really IS a "corner lift" problem, that should eliminate it. The table then should, at the least, show up a different error.
Good idea. Isolation tests that eliminate possibilities will help you narrow it down.
 
With gib out, the error is little if any different.

Shifting the table does not appear to change it

lengthwise movement of the table appears not to make a difference.

The ways have been cleaned, so probably not a matter of debris in the ways.

I have rechecked the dimensions several times, and there seems to be no error there.

I will do a diagram of the thing. I do not know that photos will be as clear.
 
So, I stoned the sharpish corners off the "saddle" ways where the corners fit into the table ways just in case. Essentially the same result.

There is no change on either side of the table as the gib is snugged.

Views of the end of the machine ways:

Table/saddle. The side registering high is the left in the picture. You can see the corner relief on the table way.
WHZGvfE.jpg


Saddle/knee. Again, corner relief is visible.
lDVjxLs.jpg


Knee/column, typical except the other side has no gib, and does have corner relief.
L56rrnv.jpg


Tram check. I see I did not get all the blue off the table from the check I made.
sWjVJW1.jpg
 
I know where you are going with that. Bridge over the scraping marks.

I'll do something on that order, to exhaust all the possibilities.

But really there is no need. The amount is quite consistent, regardless of the table movement, and is not much if any affected by the knee movement either. I have done the measurement a couple dozen times at random locations on the table, with the same result.

The indicator is a half thou indicator, and the scrape marks just cause the needle to wiggle a little. With an 0.0001 " indicator that could be necessary to get sense out of it.

Also note that the total error is nearly 3 THOU. I'd not expect that to be the result of falling into a scraping mark.

I have also run the table back and forth using the same indicator on the table surface. The needle barely moved over the entire table travel, as I would expect. Yes, I did check to see if it was stuck, or jammed at one end of the travel... it was fine, and a light pressure did move it, but it came right back. That seems to suggest that a "averaging plate" is not going to change anything (and that the table is not "wedge shaped" from end to end).

Just to recap:
* The knee ways measure in-tram at all 4 corners, and spot as flat.
* The table ways of the "saddle" measure in-tram at all 4 corners, and spot as straight and flat.
* the saddle measures the same at all 4 corners from table ways to knee ways (measured where they are above each other).
* The table's own ways measure the exact same distance from way to table top at all 4 corners and at random points along the table.
* The table top measures flat and parallel at front and back when spotted against a flat. (The table top does have some minor wear in the center area where work is usually placed, but it is under a thou.)
* The table was moved around on the ways, with the gib removed, and no change was noted, so a 'corner ridge problem' seems unlikely.
* The corners of the "saddle" ways were stoned down to round them back a bit, just in case, no change.
* The table shows a gradual rise from operator side to column side, equal on both sides of the spindle (CCW on right, or CW on left).
* That rise is not dependent on table position
* all the ways have been cleaned multiple times before measurements.

The problem is that the column side of the table consistently measures higher than the operator side when tram checked. The amount is stable and consistent, despite the assembly having been taken apart, measured, and put back together, more than once.

But, since the saddle shows no such rise, nor does the knee, the spindle either is not out, or there are exactly compensating errors on either the knee, or the saddle. And, if there were, the table should repeat that compensating error, and not show an error.

Given measurements that have been made of the table and the saddle, etc, that seems as if it cannot occur. Obviously I have overlooked something, but I don't know what it is.

Spindle bearings:

Obviously I cannot check the roundness etc of the bearings as at the factory. These are war surplus precision angle roller bearings, and appear to have burnish marks for correct orientation, which I did follow when re-installing them after cleaning and re-greasing. They looked very clean and in good shape to me, and I have seen all sorts of bad bearings before.

I have preloaded them per the instructions, and have checked for any looseness, found nothing. The very consistent results at various conditions of rotation, and points 180 deg out, don't sound much like bearing issues.

While I cannot prove the bearings are good, I also cannot show any evidence they are bad. So, for the moment, they are in the clear. I have no desire to take them out for any further checking.
 
Last edited:
Table weight causing it to sag? Set the DTI at the front of the table and pull up on the table from underneath to see if there is any big change.
 
Did that first off. Nope, the knee is still tight since the total re-scrape. Could not get a thou out of the thing lifting as hard as I could.

Also, no significant difference between results lowering the table vs raising the table.
 
Here is an idea out of left field from something I was told more than 30 years ago. On some knee mills the machine base knee ways are not 90 degrees to the table top. Apparently this was done to stop juddering when the knee was raised or lowered. What I was told may be total bullshit but I’m sure someone like Tyrone or Richard would have a definitive answer. I’m sure you can work out the implications of this if it’s the case.
 
This gib looks like it may be touching/rubbing on the clearance surface. Could that be causing issues? But I thought you mentioned removing the gibs and it's still the same, so maybe not.

lDVjxLs.jpg
 
That one is at the left side. The other is on the operator side, so neither of them would raise the table at the back. I cannot get a 1.5 thou feeler into any way, and the error is almost 3 thou. If they were raised up, I should be able to.
 
With no picture of the overall machine I could be way out in left field here but... you seem very focused on the table/saddle/knee. But tram is a measurement of the spindle axis vs the table. So the _head_ is just as important if not more important than the lower half of the machine. You say there is no nod, but are there ANY adjustments of any kind between the column and the spindle? Anything that could have loosened or shifted? How is the head attached to the column? I doubt it's all one casting, so where there is a joint there could be an error.
 
I've stripped it to the knee twice, saw nothing. Both the knee and saddle also check as in tram and flat on their bearing surfaces.

However, if you have checked everything, and nothing can be causing the problem, then something you checked is not as you think it is. The question is which one....

There is a round ram, movable. Only adjustment is turning the ram. I rechecked that.

The ram is and has been in the same trammed position throughout almost the whole process. Everything was scraped to it as a reference, since it was the fixed point, either using it as the reference, or using it's carrying surfaces (per Connoley).

Plus, supporting parts are in tram, so why is the table "out", when it measures to be flat and parallel?

Something has to be not as I think it is, and I need to find it.
 
Here is an idea out of left field from something I was told more than 30 years ago. On some knee mills the machine base knee ways are not 90 degrees to the table top. Apparently this was done to stop juddering when the knee was raised or lowered. What I was told may be total bullshit but I’m sure someone like Tyrone or Richard would have a definitive answer. I’m sure you can work out the implications of this if it’s the case.
 








 
Back
Top