Problem: Axis jump when exiting HandWheel mode

My machine has developed a problem that I would like some advice about. It is a 2005 Milltronics RH-20 with Centurion 7 controller. When I exit handwheel mode after finding and defining my datums, the table moves. The head also moves (up, thank God) when exiting handwheel from the Z axis.

Using a dial indicator, I determined the actual amount of movement to be 0.0026. Surprisingly, the movement is essentially the same in all three axes. I can compensate for this by adding 0.0026 back in to my datum definitions; Instead of using -0.1 as my edge finder position, I enter -0.0974.
It is difficult to know whether the amount of error is precisely the same from day to day. The discovery process is a discouraging expenditure of time and patience, since it calls into question the accuracy of the edge finder and indicators as well as the machine. I have even detected a slight movement when switching from 1 thou increments on the handwheel to 1/10.
One possible clue to the error's origin is that I can hear a faint "clunk" when these movements occur. (When I change modes.)

A search of the forums has not led to a direct discussion of this issue. I have found the thread "Y-axis vibration" and the referenced article (Axis Backlash - CNCexpo.com) to be informative. I have corrected for backlash. The correction was small; less than half a thou on all three axes.

There is a Handwheel Gain parameter; adjustments result in changes to the amount of movement for each incremental turn of the wheel. Apparently not the solution. I am hoping that tweaking a parameter will resolve it, but cannot find this described in the manual.
Thanks

Milltronics wont help? I have seen hurco do the same thing when you changed modes on the machine it would move .005 in x I believe. Its probably still that way today.

Axis balance (zero offset) on the servo amps.

I don't know what's in that Milltronics control. However, if they use a standard CNC rotary control for the handwheel, I have an idea of where the jump is coming from.

Handwheels generally have 6 electrical connections: ground, power (generally +5 V), A+, A-, B+ and B- (or something like that). The '+' and '-' suffix signify that for the same signal (A or B) the '+' and '-' will be opposite. eg., one will have a voltage on it while the other will not and vice versa. When the handwheel is sitting still at a detent, A+ and B+ will have a low voltage on them. In the time it takes handwheel to be rotated by one detent one way, the voltage on A+ will go up, then the voltage on B+ will go up, then the voltage on A+ will go down, then the voltage on B+ will go down. (The signals with the - suffix are doing the opposite at the same time.) When the handwheel is rotated by one detent the other way, first the voltage on B+ will go up, then the voltage on A+ will go up, then the voltage on B+ will go down, then the voltage on A+ will go down. The control interprets this sequence of signals as a command to step one way or the other.

What I think is happening is that when the control is switching to and from the handwheel, the electronics are seeing a partial step sequence and interpreting it as a jump even though one isn't intended. I think this may be fixable by changing the connections on the back of the handwheel. All the handwheels I've worked with have screws securing ring terminals on the ends of wires. First, after making notes on which wire went to which terminal, I would try exchanging A+ with A- and B+ with B-. This may solve the problem or it may make the axis step in the opposite direction. If the latter I would try restoring first A+ and A- and if that doesn't work then restore B+ and B- and swapping A+ and A-.

If the jump is gone, the stepping may now be in the wrong direction. If so, then exchange A+ with B+ and A- with B-. If none of this swapping has an effect, then it's probably something that Milltronics will have to address. If it does have an effect but none of the swaps completely solve the problem, please list the handwheel connections or post a picture of them.

Hope this makes sense,

Good luck,

Duane

Duane may be correct, but if it was me, I would balance the amps as Rooster suggested- its not that hard to do. Remember one thing- the machine most likely worked properly for a long time before this problem came up, and if so, its not a wiring issue. Dan

Thanks to all, and sorry for the delay. My reply notifications didn't get to me. I've had communications issues with Milltronics, too.
If the handwheel connections swap did have the anticipated effect, I wouldn't assume the wiring is at fault, just that it was a wiring solution. But now I do wonder if something could have developed a short.
Axis-balancing the servo amps sounds good, and not that hard to do. Would anyone care to elaborate? I'll search the forum, of course.

My Servo Systems are Yaskawa SGDH-10AEY220 amps on all 3 axes, SGMGH-09ACA61 motors on X & Y, and SGMGH-09ACA6C on Z.
I found the manual at
http://www.yaskawa.com/site/dmservo.nsf/(DocID)/TKUR-5PHSJH/$File/YEA-SIA-S800-32.2.pdf

There is no reference to a handwheel, of course. Several sections refer to setting offsets to precisely 0V, and "position reference", but everything appears to be under software control, even when labeled "Manual". Afraid I don't see what I thought I was looking for. Could this be why the dealer wants several hundred bucks to pay me a visit?

I don't know anything specific about Milltronics, but if this is a problem that didn't exist at one time and has gradually developed, I would suspect that it might be mechanical wear of the detent mechanism in the hand wheel. As Duane has mentioned, if it is in between detents when you switch out of the hand wheel mode, the logic will have to round off to one step or the other. This might be the source of the problem.

The few hand wheels I've taken apart were designed to provide a nice precise detents when new, but once worn a certain amount, they would allow some play in each position. If you can feel any play in your detents try switching in and out of hand wheel mode while lightly rotating the handle one way or the other against the detent and see if this has any effect.