You are confusing me with the word 'Bounce' Mach decelerates before it gets to the stop position so there is no bouncing of anything.  It in no way goes past the end point and then back.  If it did then the axis would be out of position from then on.

The acceleration and deceleration are set within the motor tuning so you can make the axis slow down over a greater distance if needed.  If the tool is hitting the work then you ether have a setting problem or major backlash issues and possibly tight axis gibbs causing drag and drop on the Z axis.

Graham

Thanks for the reply Graham - took 'till now for me to get the time to experiment. This is what I've observed: If I set the retract height to be high above the workpiece, then it seems I can escape the problem somewhat. The problem seems to be if the distance between the retract height and the last depth drilled is too short. If I leave roughly .8" between the work surface and the retract height, it works with essentially no 'bounce'. My theory is that it's still ramping up to rapid when it tries to ramp back down to feed if the retract / plunge height is too small (I was retracting to .1" originally). The result of increasing the R settings so far is that it pecks air for a few cycles while bouncing, then works properly when it gets to the top of the part: As the distance of the rapid increases, the 'bounce' at the end decreases until there is none.  I've thought about playing with the motor tuning ( would faster acceleration help, so it gets to its rapid speed sooner, although it seems counterintuitive to go faster?), but since it's a little out of my comfort zone, I thought I'd consult first. If I were making a godzillion widgets, it would really matter, but for now I guess I can live with it as is, or long code around it ( small job shop - short runs  usually). The spindle doesn't seem to lose position ( haven't yet tested it with an indicator, but it will drill multiple holes without a visible change - if I have time today I'll check it with an indicator at the start and finish of multiple cycles). I'm new to using Mach 3, and not the one who did the motor tuning - the people who did that are far more knowledgeable than me, so I'm reluctant to tinker with what they've done (it's a bridgeport upgrade, and I believe most of the values used are ones from bridgeport, but I could be wrong). So far this is the only cycle that's shown any issues - all parts have been within tolerance, although I've avoided G83 because of this issue.

Thanks again for the help, Alex

My theory is that it's still ramping up to rapid when it tries to ramp back down to feed if the retract / plunge height is too small (I was retracting to .1" originally).

Impossible without losing position.

I still suspect backlash as being the problem.
Assuming you have a ballscrew on the quill, the counter balance spring will hold the quill up against the drive at all times.   (or near neutral weight)
The momentum of a rapid Z down could overcome the spring and coast through the backlash.
If you have the rack and pinion drive method, it could be much worse as far as backlash goes.
I'd check the backlash closely.....then go from there.
Russ

I am not sure of which drives you are running but it sounds like the PID tuning is out of whack. When setting up mine it was possible for the position to oscillate if the tuning was off.

Hey alkassell

Any updates on what the problem was?

Mike

Thanks for the interest Mike - I haven't had a whole lot of time to spend on it - have to make those widgets, and I have a workaround. Here's what I have done:
I ran a program peck drilling 80 holes, with 10 pecks per hole, and touched off on an indicator before and after to see if it's losing position, and it was roughly .0006" off (from memory) - considering my quick set up, that the tool's not flat and the machine was made 20 years ago, I consider that to be no error. A friend of mine who is a lot more knowledgeable than me ( that's most people ) said "We should probably stop calling our term "bounce" and call it servo following error on rapid moves...as I understand it all machines have this, and that is why a programmer never rapids right to the start of cutting, or even .010" close!". He also told me on the phone that Mach 3 has a closed loop feedback with the galil motion control card I have so it knows in real time what the encoder shows, so indeed, it can overrun its mark, and then correct without losing position, unlike most other systems ( hope I got that right ).

Anyway, I think that motor tuning can cure it, but since it doesn't lose position it would mean that my rapid would be slower for everything else, and I don't do that much peck drilling, so I've been reluctant to slow it down. I probably shouldn't worry about speed - I do short runs, so the difference would be negligible, and if it keeps me from chipping a tool down the road, it would time ( and $$) saved.

Alex

FWIW,
A little off thread.
Haven't done much peck drilling, but have tried out the macros that Melee posted. The latest updates include macros for the mill. See link below. The ones on the lathe worked here and need to check the mill ones out.
I will say that if you are going to peck drill the axis doing the pecking needs to be solid ( that's why i don't do much
peck drilling on my Atlas mill ( quill just has too much backlash ). I did experiment with the lathe ones and will tell you they worked great. Peck drill little .020" holes and rapid peck drilling where the return to start the drilling is within .001" and if bad ....broken drill.

My only point is that if your axis is not accurate, has backlash, or just dosen't work very well ( for whatever reason), peck drilling may not be a something to do.

http://www.machsupport.com/forum/index.php/topic,13400.msg87932.html#msg87932

Like i said FWIW,
RICH

Thanks Rich, any input is welcome. If the macro allows me to change the height at which it changes from rapid to feed, it should cure all my ills ( at least for pecking ). Whatever the reason, .010 isn't enough room for my quill to decelerate to the feed speed. My workaround has been to drill a few cycles of air above the surface first, making it start feeding higher up - obviously a waste of time if it can be cured. I'll check out the macro when I get some time - I've got a few glitches to work on anyway ( VFD speeds with cone pulleys, soft limits, etc. ), so I'll have to make some time.
Thanks, Alex

Thanks for posting back Alex...I was very curious as well.

He also told me on the phone that Mach 3 has a closed loop feedback with the galil motion control card I have so it knows in real time what the encoder shows, so indeed, it can overrun its mark, and then correct without losing position, unlike most other systems ( hope I got that right ).

This is great info that I/we were not aware of......it explains a bunch.

Thanks again,
Russ