[Cowbell]

As I've started delving more into precision work, I've run into some real strange issues with the flatness/alignment of my table.

To start with, I trammed the thing using the standard technique of sticking a dial indicator on the spindle and rotating in between extremes of the table, tapping the vertical column to get it perfectly aligned.



My dial indicator has hash marks of 0.0005", and after endless tweaking and checking, I got it so that it ready to <= 1 hash mark on each end of the rotation. As far as my beginner's eye can tell, it's perfectly trammed.

Then I mounted the dial indicator directly under the spindle so that X motion of the table was perpendicular to the direction of the indicator head to eliminate push/pull drag. What I find here is extremely puzzling.

As I travel over the RIGHT side of the table (that is, from the middle out to the right end) there is extremely little deviation in the indicator. <= 1 hash mark. However, as I travel over the LEFT side of the table (from the middle to the left end) the deviation is large and seemingly linear. Over the 5-6" of travel, the deviation is well over 0.003". You can actually see the dial smoothly moving as you turn the feedscrew.



This really puzzles me.

a.) if I can get the reading during tramming to be pretty much spot on, why would I see a difference as the table travels?

b.) why would one half of the table appear to be nearly perfectly flat, while the other half deviates so sharply?

About the best I can come up with is that the table isn't sitting in it's grooves cleanly and as it crosses the midpoint, it tilts ever so slightly. But I can't feel this, and putting weight on the left end of the table doesn't seem to change things at all.


What goes on here? Is this kind of thing normal on such a low end mill? What options do I have? It would suck to have to restrict usage of my mill to half the table

[phil burman]

a) When you tram it you are only checking that the spindle axis is at 90 degrees to an imaginary plane that passes through the points on the table touched by the indicator. Try the same test at different positions along the x axis.

b) It may not have been machined flat or it may have warped after machining. Use a straight edge of known quality to check the flatness of the table.

Here's a test procedure with acceptance criteria that you may find interesting/useful.

http://www.tormach.com/document_libr...C_addendum.pdf

Regards (chair)
Phil

As
This really puzzles me.

a.) if I can get the reading during tramming to be pretty much spot on, why would I see a difference as the table travels?

b.) why would one half of the table appear to be nearly perfectly flat, while the other half deviates so sharply?

[cncuser1]

check the gibbs to make sure that they are not rubbing where they are not supposed to.

I found that my gibbs were installed crooked and rubbed in the wrong places, causing friction, in an extreme case they may cause movement too.

[acondit]

Cowbell,

I found that on my Mill-Drill when the table was all the way to the right, the weight pulling down lifted the table about 0.001 under the spindle. It actually raised it about 0.001" in the 4" width of the vise. If I adjusted the gibs enough to stop that it was too tight to use.

I repositioned my vise to the middle of the table and that helped with thing being machined in the vise.

Alan

[OwenL]

Table sag from the weight all the way to the oposite side and loose gibs.
Gib keys that are not true lifting the table as they travel.
Table actually not ground flat and true [thicker at one end v/s the other]

just throwing out things to check for.
Owen

[mjarus]

Well I just experienced the EXACT same thing today! I finally got around to tramming my table. Fiddled with it forever. Then I ran the X axis with the indicator in the spindle. Same problem except I get .004 on the left side instead of .003. I have a HF X2.

So, really miffed I came inside to search the forums and lo and behold same problem by someone else (so if I am crazy I have company)!

On another note I have adjusted my gibs so I don't think that is a cause... the X-axis would actually move if you shook it with your hand, I adjusted the gibs and now it is fine.

I think it would be odd that 2 machines would be out by almost the same amount in the same area if it were just an issue of a crap ground table. Maybe they made ours back to back in a training class?!

Anyways, I will give anything a shot. If I can't figure it out, back to HF I go...

mjarus.

[50062040]

what you just proved marjus/cowbell was the the t slot or edge which the dial indicator was touching while doing measurements is not parallel to travel direction.

i dont know if this is common but it shouldnt be a problem.

[Cowbell]

I'm not sure I follow. Assuming, under ideal circumstances that a table is perfectly flat, it shouldn't matter if the slots are parallel to travel direction. The table should be able to move in any arbitrary direction in XY and the plane of it's surface shouldn't change at all.

Or do you mean that the non-paralellism is the result of the table being twisted or otherwise distorted around X?

Either way, if you mount something on the "good" half, you get flat horizontal travel in both X and Y. If you mount it on the "bad" half, you get large, linear divergence in Z (height) over 5-6 inches.

The more I think about it, I just can't see what could cause this except for a non-flat table.

[mjarus]

i dont know if this is common but it shouldnt be a problem.

Could you explain more please? Doesn't that mean that, for example, if you place a piece on the table that is square to the table it will cut deeper on the left hand side of the piece than the right?

As an added note I checked the table with a long parallel and it seems flat. I just checked it by the looking for light underneath the edge method. Probably not the most accurate, but it is all I could think of. I figured I should be able to see some light under a .004" gap. Maybe not.

Any other ideas?

mjarus.

[mjarus]

Ahh posted the same time as cowbell...

Anywho, cowbell... I tried to check for table flatness. See above post. Looked ok.

mjarus.

[Cowbell]

I just checked with paralells too. They do indeed appear flat. But, running the table for the Nth time, the results are the same. <= 0.0005" over the 4-5 inches of travel on the right side of the table. > 0.003" over the 4-5 inches of the left side of the table.

Also - when the dial is at maximum error, I find that I can't even budge the dial more than 0.0005" by pulling up/pushing down on the table by hand. To me this implies the gibs/rails/whatever are highly snug. Yet the deflection is still here.

Man. Very confusing.

[50062040]

apologies, i assumed the readings were from the sides of the t slot

[nate]

when i worked as a 'grinder hand' in a tool and die shop we would routinely 'dress' the mags on our surface grinders. the mag on a surface grinder is very simmilar to the table on a milling machine. (a flat surface that moves in 2 axis) but ofcorse, its a magnet

to dress the mag we would actually grind it by lowering the surface grinders wheel down to the mag and then moving the table in the x rapidly and the y slowly. nice even smooth passes were a must.

why not "dress" your table with a fly cutter. then the surface of the table would be trued to the travel in the ways.

also you might try "stoning" your table with a bench stone (knife sharpening stone) use one that is at least 1.5x1x6 this will give you a large enough surface to prevent dig in by distributeing the force. move in a circular or figure 8 pattern. also periodic cleaning of the stone with alcohol and an old tooth brush will make it slide smooth and cut right.

the cast iron of the table will clog the stone quickly. cleaning with alcohol will really help!!

stone the whole table. be careful not to spend too much time in any one spot.

[50062040]

ive stumbled upon an article describibng exactly what youre talking about:

http://www.mermac.com/freemill2.html

"When we run the table back and forth and side to side with an indicator mounted in the spindle, we can see the tendency of the table to dip (move away from the spindle) at certain wear points. At the extremes of end to end travel on a big mill, we will also see the weight of the overhanging table cause the opposite end to rise a bit. The amount of dip can give us an indication of wear to the sliding surfaces. A few thousandths (.001-.0025") would be normal, more would be excessive. I said before that if the table passed under the indicator at a uniform angle, it would go undetected. How then is this dip possible? By uniform angle, I refer again to the example of the car driving up the ramp. If for argument's sake, someone cut several thousandths off one side of the saddle so the table angled to the floor slightly in one direction, that would cause a uniform angular displacement of the table. Wear tends not to be uniform and would only cause local displacements which would register on an indicator. "

[mjarus]

Well I solved the problem... I returned the mill and got another. HF was pretty cool about it. The new mill, with minimal fiddling, is perfect on the right side of the table and .0015 out on the extreme left. Good enough for me... I am done screwing with it!

Cowbell, see my other post about nut adjustment before you try this (as I have no clue what I am doing), but I think maybe adjusting the leadscrew nut may help. I had to do this on the new mill and before I did it I was getting really odd readings when I ran the table, but after it got a lot better. Maybe it had nothing to do with it and my machine was just off until I adjusted the nut, but it seems fine now... and that was just about the only thing I didn't try with the last one. Like I said, I have no clue if this could even remotely help, maybe someone else could shed some light on this?

mjarus.

[WhiteTiger]

I suspect you are looking at surface finish and adjustment issues. These budget mills generally have fairly accurate ways, but the surface finish is lacking, which makes them move stiff and irregularly when properly adjusted. Result: most users tend to adjust them for initial ease of movement which introduces functional (but not felt) slop in the fit.

The horizontal bed surfaces are usually fairly good, but the dovetail rail surfaces can be rough as a cob and since those are where the gibs reference to and ride against, the movement is stiff as a board during break in unless the adjustments are backed off for smooth motion.

The safest, easiest cure I know of for this is to strip the machine down, and oil up the ways with a good cutting oil (I like Marvel Mystery Oil, available at any auto parts) then reassemble just the saddle on the Y axis ways, adjust till you can just push it with effort, and spend a few minutes shoving it back and forth full stroke on it's ways. Then spend about 1/4 that amount of time short stroking just the ends of the ways, taking them a bit out of engagement if possible. When it gets easy to move, disassemble, wipe everything down to clean off the iron particles you probably will note turning the cutting oil greyish, then repeat at least one more time.

Once the Y ways are broken in and final cleaned, reassemble them dry and lock the gib down solid, then repeat the whole thing for the X axis. It's important to lock down the Y axis, as rocking in it can introduce force errors in your self lapping of the X ways and make them come out irregular in extreme cases.

When working the X ways, be sure to hold and support the free hanging end of the table when you are short stroking the ends of the table ways (which has to be done to offset the wear distribution, since the center of the table ways gets double the wear with each full travel stroke.

It's tedious, but it pays big dividends. Took me about an hour and a half all together to break in all the ways on my old bench mill, not counting the scraping of the saddle X axis dovetail surfaces.

Once you have all the surface roughness knocked off the mating surfaces, you can adjust the gibs correctly and not have them compress the high spots of their mating surfaces and give you deflection under load.


Tiger

[widgitmaster]

I just checked with paralells too. They do indeed appear flat. But, running the table for the Nth time, the results are the same. <= 0.0005" over the 4-5 inches of travel on the right side of the table. > 0.003" over the 4-5 inches of the left side of the table.

Also - when the dial is at maximum error, I find that I can't even budge the dial more than 0.0005" by pulling up/pushing down on the table by hand. To me this implies the gibs/rails/whatever are highly snug. Yet the deflection is still here.

Man. Very confusing.

This may be an easy solution!

The different amount of error for each direction of travel is due to loosness of the Gib itself, one direction makes it wedge in tight, the other makes it relax sloppy!

Check the adjuster screw & knotch in the Gib, to see if there is extra slop in the adjuster !

Just Another concept!

Eric

[Nava]

on my larger bf20l table I get about + .05mm (.0019") on either side due to the gravity of the table..even though i have my gibs pretty tight. I wish more people would do some how to videos about making these machines better by scraping ect.. so that ic an get rid of it. At some point even the best machines will need to be redone.