[Kingjamez]

I have a X3 with a cncfusion conversion kit. The Z axis has double ball nuts. When I change directions in the z axis, the head actually continues going the original direction for about 0.003" ( im commanding it to go the other way), then actually reverses correctly and catches up to the original "change of direction" location at about 0.007" of instructed motion. This happens upon direction change no matter if going down or up. I can feel the screw change rotation direction correctly, so I know it's not a problem between the motor and screw.

Anybody seen this before?

-Jim

[hoss2006]

I heard about it once before, if I remember the problem was a warped gib strip.
I'll try to find the thread.
Hoss

[holbieone]

I have a X3 with a cncfusion conversion kit. The Z axis has double ball nuts. When I change directions in the z axis, the head actually continues going the original direction for about 0.003" ( im commanding it to go the other way), then actually reverses correctly and catches up to the original "change of direction" location at about 0.007" of instructed motion. This happens upon direction change no matter if going down or up. I can feel the screw chain rotation direction correctly, so I know it's not a problem between the motor and screw.

Anybody seen this before?

-Jim

if your backlash comp is set to a negative "-" ,try using an absolute value of the number "+"

i think it is looking for a delta not a direction

[hoss2006]

Found it, similar problem 13 months ago in this thread by thorpydo.
http://www.cnczone.com/forums/bencht...axis_woes.html
Hoss

[neilw20]

Known as head bobbing to the best of my knowledge.
As the head goes down, drag on the gibs/slides and the gas strut cause the lag.
I just program Z to overshoot by 0.055 mm then back up to the correct position and the spindle continues down the last 0.055mm. Before I got rid of significant slack in the gibs it needed 0.13mm

you really need a 0.001 resolution DTI to measure this properly.

Using this overshoot method, I am repeating to 0.002mm ALL DAY --- after the spindle temperature stabilizes.

Easy work around, but with CAM software, you need to do some fiddling to get the code generated.

I just program a ramped approach of 0.055mm.
My column is also attached to the wall so column flex is no longer an issue. Column bend from vertical cutter loads are much larger than the bob value.

[Kingjamez]

Hoss: thanks for finding that thread! I'll try some of the suggestions and see if I can reduce the error.

Nellw20: I do a lot of 3D work, setting overshoot seems like that would be near impossible in my CAM software. How do you do it?

-Jim

[neilw20]

Probably easiest to patch the Z moves.
You only need worry for final finishing, if the Z accuracy is all important.

I use parameters to achieve this.

#778=0.055  (TILT FUDGE TO SETTLE HEAD DOWN)
#777=-0.14 (Required Z DEPTH for example)

(the F#240 is a nice feed rate so no jerk involved. Usually about 800)

G1 Z0.7 F#240 (ROUGH APPROACH for Z)
Z[#777-#778] (Go below required level)
Z#777 (and this nods the head correctly)

or it could just patching the Z moves.

Z[-0.14 - #778] (Go below required level)
Z-0.14 (and this nods the head correctly)

I often include an X or y move with the Z-0.14 move to ramp gently onto a cutter path.
A Z up move does not jerk down. It is already there.
...
...
M30 (end program)

Fairly easy to write a program to patch the gcode file.

[Kingjamez]

Thanks for the help. It never occurred to me to write a simple script to modify the Z commands. Great idea!

I read the the above thread and started to play around with the z axis gib. I had tightened it down after tuning my servo so that it was just tight enough not to stop the servo at 120IPM. I loosened the gib, and tried the test again. What do you know, the effect GREATLY diminished, and now I only get 0.001" of "bob" and total backlash is around 0.0025" , I think I can live with that.


Thanks for the help, off to cut something and make sure I haven't messed something else up :-)


-Jim

[neilw20]

Write a simple program to move the Z up and down a few hundred times over maximum range and make sure it never loses steps.
Then tighten the gibs a bit more. A bit like running in an old car!.
Check final results with a DTI to ensure no loss of steps.
You need to 100% confident, so it wont destroy expensive cutters. They all cost too much.
I have 33 10mm 3 flute solid carbide, and they are not cheap.
They need resharpen after 2 hours each now, and before the head nod changes was less than 30 minutes. SX3 makes a good T&C grinder too.

[Kingjamez]

Hi Niel,
Thanks for the advise. Why do you think that the head nod problem caused premature wear?

I have a servo system, so I never loose steps. If I tighten the Z gib down any further, the servo has no problem moving the head (I've got dual gas springs like you do), but the head nod problem re-surfaces.

-Jim

[neilw20]

Just smoothing off the high spots, it seems, that the Chinese left behind.