Hi Mike
I bought a "new" MP8 off ebay a while back. I haven't had chance to get it set up yet but it came with a test certificate from 20 Jan 1998 (the probe is however in unused condition) and the test results are amazing! From over 60 measurements only one had an error of over 0.5 microns. It would have been even more amazing if the test certificate had the same serial number as my probe!!! I've no idea yet how well my "new" probe will actually perform but I suspect if you're not using the probe electronics the errors could possibly be significantly higher.
My interpretation of backlash correction:
Backlash, as opposed to lost motion, is how much "play" is present in each axis. To make the calculations simple let's say that a particular axis has 2 thou backlash (I like to consider this to be +/- 1 thou) and the rest of the machine is almost "ideal" (stiff, low friction etc). When the axis is moved in the positive direction to a set position, say zero, the mechanics will usually lag behind the commanded position by half the backlash, or in this example -1 thou. If additional force is then applied in the direction of the previous movement the mechanics may continue to move to the other end of the available play. The actual position will now be +1 thou. The actual position can therefore be within +/- half the backlash, or +/- 1 thou maximum. Symmetrical around the requested position.
Now if we add compensation such that when the axis moves positively to zero, the position commanded by the software is actually +1 thou, the mechanics will lag 1 thou behind and stop at exactly zero - great! However, if additional force is applied, the mechanics will be able to move an additional distance equal to the backlash i.e. 2 thou. The worst case error would then become +2 thou - twice the value compared to the maximum value
without the compensation! Bearing in mind that this is the tolerance for one surface. Machining 2 parallel faces could produce an error of 2 thou per side - a total of 4 thou error.
These forces can be cutting forces or, in the case of the Z axis, the weight of the head alone. The Z axis is however slightly different. Most mill probing routines set the reference while moving in the head in the downward direction. On my machine (and I presume all similar machines with properly adjusted gibs) the head will lag behind the command position under these near static conditions. However, when the spindle starts, and at the latest when the tool enters the cut, the head will experience a downward force (c.f. tool pullout). This will pull the head down to the other end of the backlash, so for the above example the head would now be 2 thou lower than expected. If all Z movements into the workpiece are downwards then correcting for backlash on the Z axis won't change this, it won't be any worse, but it won't be any better either. I can imagine that the surface finish of 3D work could suffer with backlash correction enabled, but I haven't tried it, so I can't comment.
I'm constantly amazed when people report they've been machining with backlash values of 3 to 5 thou and didn't even notice! Adding 5 thou backlash compensation to a machine like this could result in a total error of 10 thou between 2 faces! This is not simply a cure for a poorly adjusted machine.
The first priority must be to set-up the machine to reduce the backlash to within, or better than the specified limits - yes I'm aware that this can be difficult. The specified tolerance limits are more than good enough for the vast majority of Tormach users. Backlash correction then becomes irrelevant to all but a few nuts like me
Maybe you see this differently?
Regards
Step