I've been trying to solve this problem for a week to no avail, so I'm hoping somebody on this board can help. My system (XP on computer) is communicating through a gecko G540 to drive 4 stepper motors (a slaved to x). The problem is that the y position seems to change as I step down through a part. I will attempt to insert a picture that will hopefully show the problem. When the issue first occurred, as I roughed successive layers (3D roughing with a full retract) I noticed that the bit was off around 0.030 inches with each cut. The full roughing was about 1.5 inches deep and each cut was 0.25 inches in z. At the end of the roughing there was a series off steps vertically for a total error around 0.150 inches. Unfortunately I did not take a picture of that part.

During trouble shooting of this problem my coupler on the Z axis started to slip so I had to tear apart the Z drive to tighten up the coupler and then went through and realigned everything (to less than 0.1 deg error). The problem is still present but the steps (always to the -y direction) are much smaller. Somewhere in 0.007 inch give or take a mil.

I've tested slowing down the feed rates (typically I've run x, y around 90-120 ipm, and z around 50 ipm). The accelerations prior to the drift were around 0.0052 G's and have dropped down to around 0.0038 G's. Pulse width was 4 micro seconds before and was increased to 5 microseconds. None had any measurable affect. Presuming the picture insertion works, you can see the problem on this test run. The speeds are 120, 100, 75, and 50 ipm starting from the edge nearest my hand and moving upward. X is in the short direction while y is the long direction. This program sent the bit to an xy position dropped down in 0.25 inch steps and made a straight cut in x for 3.75 inches. The bit fully retracted and moved back to the x start position. It dropped another 0.25 and cut in x again. This process repeated multiple times. So in each slot there was never a command to move in y. The steps seem very consistent.

After the first slot was cut, the bit moved to a new y position and again cut a new slot at the next lower speed. The steps seem consistent for each slot and was not reduced by the slower speeds.

It seems to me that errant pulses are being sent to y with each z movement. I don't have an oscilloscope so I can't confirm this. My finger isn't apparently sensitive enough to detect it on the stepper motor.

I'd appreciate any suggestions as to how to track down the source of the problem and to resolve it with resorting to swapping everything out.

Regards,

Greg

It seems to me that errant pulses are being sent to y with each z movement.

Don't know how you came to that conclusion.

Looks like the Z axis is not perpendicular to the table. I can't see the other side of the groove.
I am assuming that your steps per unit are correct and that backlash is not a factor.

RICH
 

That is what I thought at one point, but I was at the machine when it originally happened and didn't see or hear anything out of the ordinary that indicated the z axis mechanism shifted. The Z axis was perpendicular to within 0.1 deg (the limit of accuracy on measurement tool). I always check when I put in a new tool but I did not check after this happened. That was silly.

For the part in the picture the bit is also perpendicular to within 0.1 deg. If I did my calculation correctly a 0.1 deg error over 0.25 inches would be on the order of 0.0004 inches. These steps are more than 10 x larger and would suggest an angle error more like 0.4-0.5 degrees.

The original problem would have suggested a misalignment closer to 7-8 degrees or have I made a mental error? I think I would have seen that happen. I will try to check my angle measurement tool by buying another one tomorrow just to be certain.

I didn't show the other side of the groove but it is straight as expected. The groove is formed by a single pass with a 0.500 end mill. As it steps in y minus direction it just keeps cutting the opposite side.

I might have better said that something seems to happen to y when there is an x and/or z command given.

Any other ideas come to mind?

Thanks,

Greg

Greg,
I will assume that you are checking the Z bit being perpendicular to the table by measuring against the bit.  I think you may find that while the bit may be perpendicular to the table, the Z axis rails/rods are not.

Measure the Z rails and assure perpendicularity first, and then the tool bit/spindle.

John Champlain

Hi this looks like something I used to get whilst doing manual machining in my apprentice days which was caused by cutter deflection. I assume you have no movement on the  X,Y and Z moves on the digital read outs on Mach. I think this conventional cut verse climb milling application
Use a dial indicator to check the following.
There is no cutter deflection by pushing the cutter by hand when its stationary.
Put a dial indicator on the Z axis and run the machine to see if you get any physical deflection.

If these are Ok check again by taking a small cut off the edge of you block so you are only cutting on one edge of your cutter and see if you still get the steps.
then try either coming from the other direction or the other side of the block, this will tell you if the axis is moving and eliminate intermittent pulsing

The best example was when I was an apprentice and I used a 1/2 collet to hold a 12mm cutter so it deflected in the collet holder. In my defence this was in 1971 when the UK was changing from imperial to metric. So the balling out by the training instructor was unjustified, in those days a tap over the knuckles with a metal rule was allowed.

Best of luck Greg sorting out you problem
Jim

John,

you make a good point. While we didn't remove the rails on the z, we also did not check them after reassembly. I will do so later today. Not sure how that would have happened to create the original event but maybe I'm getting stronger in my old age and torqued the chuck too much when putting in the bit.  So long as I can banish the demons for a while and make some dust, that is forward progress.

Jim,

in theory there is always deflection when a force is applied, but I think (hope) that it is minimal as I typically only cut low density modelling board. I will also run the test you suggest as  a thousand theories are not as good as one set of measurements in my experience.

I appreciate both of your suggestions.

Regards,

Greg