I would think if this were the case that it would show up in straight cuts. For example if the X axis screw were bent then cuts using only X would show this, but according to the OP this isn't the case.I'm thinking bent ball screw or the end work is not centered.
Good suggestion.
I'm no G code expert but I don't see a problem with your code. Arcs are programmed as arcs, not as a series of small straight lines.
In my opinion it is not a software issue.
You've done stiffness tests of the table, have you done similar for the head and spindle isolated(Does this machine have a quill)?
Here is a picture of the milling head. My plan is to check and adjust the Z axis gib then run more test cuts. I'll report back on how it goes.
Edit: I did check spindle runout and deflection horizontally and vertically. About .0002" spindle taper runout and maybe .001" horizontal deflection. I used a 12" screw driver prying up on the spindle, with the quill locked, and got about .010" deflection. That's what prompted me to check the Z axis gib adjustment. After Z adjustment I will run a series of test cuts and post the results.
OK, so you have a quill. The quill is locked down good and tight at all times , right?
There is lots of opportunity for problems between the cutter tip ...through the head...to the column.
Just for kicks put a solid bar into the collet...attach the base of the DTI to the column, pit the DTI tip on the bar ( in the toolholder) and see what you get for movement, should be verrrrry little. Try flexing and reading along the 3 axis. So this would be similar to the push-pull tests you did on the table, if possible apply the same force, so we can compare the movement of different assemblies given the same force.
I'm assuming the Z axis CNC moves the whole head, so test with quill locked( which is how it should have been while cutting )
I may have found something, only testing will prove me right or wrong. As I said above I checked vertical head movement by prying up on the head. I had about .010" deflection measured with a DTI. I've spent the last seven hours adjusting the Z axis gib (had to put a shim behind the gib) and have the vertical deflection down to .005". Checking clearance between the flat Z axis column and head slide ways (top left, top right, bottom left, and bottom right) I can get a .0015" feeler gauge in the top right only. It goes in from the top about 1-1/2" and stops.
I did what you suggested above about putting a dowel in a collet and checking deflection. I got very little deflection, .001", maybe .002".
I will take me the rest of the afternoon to put the machine back together. Hopefully can run some tests in the morning.
So If I understand correctly:
- Before you started adjusting the gib you had .010" play in the head along the Z axis. Was this strictly parallel to the axis of the column? Then this is backlash, it might account for the problem, where is it from, the ball screw or the bearings that fix the ball screw, worth tracking down.
-Also after your adjustments you can fit a 0.0015" feeler guage in the ways, between the column and head in about 1.5 " deep.? This is no good. You can do other tests and adjustments but this has to be fixed. You should determine if the gap is due to a crack/bend or bad badly made parts. If it is bad machining you can re-machine the ways in the head. To get a proper fit it would be scraped. When the gib is properly adjusted there should be no gaps between the ways on the column and the mating way surface on the head, and similarly for the gib to column contact surface. This is acchieved by proper machining, fitting and scraping, not just tighter gibs.
I understand what you are saying and do not disagree but this is an Asian mill. I don't expect it to be perfect. If I see an improvement with the arc and angular surface finish then I know I'm headed in the right direction. If there is no improvement then I still haven't identified the source of the problem and will keep searching.
I'm convinced the .010" Z vertical deflection was due to the excessive gib clearance because when I tightened the gib I was able to reduce the deflection to .005". It's not backlash because it's a double ball nut arrangement and I've checked it several times to make sure backlash was not a contributing factor. In one of my earlier posts I mentioned that backlash was less than .001".
I appreciate everyone's help. This has been a nagging problem for quite some time.
Your mill is easily capable of better than .005" which is likely a main source of the problem. I suggest continuing with the DTI across multiple locations to hunt down the exact source of the .005".
I ran the Z axis through it's travel range multiple times and was able to snug up the gib a bit more. I can no longer get a .0015" feeler gauge between the Z axis slide ways. Next step is to check the head deflection to see if that has improved. Got a few chores around the house before I can get back to the mill.
Most importantly, thanks for your help. I appreciate it.
I ended up putting a .008" shim behind the gib and tightened it as far as it would go and still no improvement in the head play. I'm chasing my tail! At this point I don't trust my readings. I don't have any .010" or .012" shim stock so I have to find some or order it. Hard to believe the mill came from the factory with an undersize gib but I have to deal with it. Probably going to diassemble the head and blue check the slide ways to see how bad they are. I've never done any scraping but I may have to learn.
You have a tapered gib that doesn't make full contact along it's entire length?
Can you use the DTI near the gap to correlate the gap to the deflection?
In any event the gib needs to make full contact in properly working mill. But in the mean time can you put a shim in the gap, with no shim where there is no gap and then tighten the gib to completely jam/lock down the head?
If you can then do it and retest with the DTI for deflection from another source. If the deflection disappears when the head is jammed the you found the culprit.
As a test you can pull out the gib lock screws and replace with a bolt and a nut and adjust the gib as if it were a screw adjusted gib. This only gives you 2 adjusment points, but will tell you what you need to know. My G0704 has been running like that for years due to a tapered gib that doesn't fit right.
I'm trying to isolate the source of the deflection as suggested. For my own satisfaction, using a .0015" feeler gauge, I checked for gaps between the column ways and the head slide ways. I also checked for gaps between the dovetail ways on one side and between the dovetail ways and the gib on the other side. No gaps larger than .0015" anywhere.
As I said in one of my previous posts I have .010" deflection with the DTI mounted on the table and the needle contacting the underside of the head next to the quill (see first picture). This setup is vertical so no deflection compounding. With the DTI mounted on the column ways and the needle touching the underside of the head next to the quill I get about .003" deflection (see second picture). The distance from the column ways to the needle contact point is 13". The last test I mounted the DTI on the Y axis bed ways with the needle touching the column ways just beneath the head slide (see third picture). I got .008" of deflection. This setup is virtually vertical so no compounding. These readings add up: .003"+.008"=.011". Close enough to .010" for me.
The .003" deflection from the column to the head seems reasonable over a distance of 13". This tells me the clearance between the column and the head slide is not excessive. Now I need to look at the column mounting and see what's going on there.
Thanks for the tip. I'm doing something similar. I got rid of the original adjusting screws and made a spacer that fits the gib screw counterbore with just a few thou of clearance then use a socket head cap screw for adjusting. This gives me slightly better contact on the end of the gib.
I isolated the column flex by putting a temporary spreader bar between the column and wall (see attached picture). This reduced the deflection to .002". I ran some profile passes and there was no change in the angular and radius portion of the cut surface. I would have thought by reducing column flex there would have been some change. I also changed the steps per as Jim suggested and no change in surface finish there either. I'm stumped.