X3/X4 Z tower bending compensation for only 3KG

[neilw20]

I had a problem, taking a single pass very light cut on stainless steel.

The tool: a 10mm, 3 flute solid carbide end mill.
The cut: 11.5mm wide. Up one way, then back the other.

On pass 1, in one direction, the Z axis load is about 1.2KG causing the tower to deflect back a bit, meaning my cut was shallow by about 3-4 microns.

On the return pass, offset sideways by about 1.5mm the cutter load I estimate is about 180-200 grams, and now the tower leans towards the cut and it is 3-4 microns too deep, causing too much material to be removed, and uneven cutting.

Observing this, on the first pass I pressed on the head with my fingers (about 1.2KG) and released on the return pass. PERFECT CUT!

Not to be out done, and having no time to make the tiny tower stiff I reasoned that if I added about 1KG to the head during pass 1 and removed it on the return journey, the problem is fixed. My wife got tired fingers!

Solution: Use my 4th axis driver.
I mounted an old printer mechanism, by it's carriage to the top of the tower. With 400mm travel, the stepper motor being the movable weight I was able to measure 5 microns (0.005mm=0.0002") deflection when the carriage was fully extended over the head.
Now during pass one, the stepper is extended about 300mm giving 3 microns bend, equivalent to the extra cutter load on pass one.
On change of direction, the stepper retracts, to about 30mm and the CUT IS PERFECT.

I did not even consider having a 5 micron higher first pass as the 'HEAD BOB' inherent in this machine (about 15-20 microns + another 50-100 from friction etc) would just make matters worse.

Cutter life has gone from 1 to 2 hours, on this job which has a 2 minute cycle.

No big structures, concrete walls or expense (except for the new 4th axis plug - and some extra lines of code) has been incurred.
4 hours work and it works almost perfectly.

I can also add weight at the stepper motor end to shorten the transit time (about 1 second full travel). It sticks out less for the same weight.
The A axis is calibrated 0 to 50 giving 0 to 5 microns compensation.

After a few test cuts, it just gets better. If you get it wrong, it no worse than it not being there. Only just like putting a 1KG weight on the head.

I will post a video/pictures when I get time to do it.


Looks a bit strange, but I am sure it will cause some questions from, visitors.

[john_t_h]

Hi Neil,

It certainly sounds interesting and some pics would help make it a bit clearer as to what you have done.

I'm not sure if I'll ever need tolerances more precise than 0.02 mm, but you never know.

[neilw20]

Hi John.

I need the repeatability to get perfect surface finish with one pass.
I could do multi pass, but my 3.5 hours / day would become 10.

As it is now cutter life went from about 1.5 hours to around 3.
That's another bonus at $25 per cutter. 23 of them get recycled on the 4th axis tilt head CNC X3 grinder.

I will have some pictures soon. A video that is not too jerky.
Phone videos are just not up to it, as the tripod has a heartbeat, but here goes...

"http://www.youtube.com/watch?v=9Asq3GU-Qyk"]YouTube- X3 balance weight.mp4

Neil

[LEN1042]

Thanks Neil,
Very interesting concept.
Len

[ihavenofish]

i think im confused, by maybe not. 1kg surely doesnt deflect the column that much.

is this just a damping thing? the movable counterweight basically helps cancel out resonace? something of an adaptive damper.

seems like an interesting idea vs the typical "fill it with granite" approach.

[neilw20]

I don't need to attach it to the wall like others.
It does bend and is quite measurable.
1KG push down on the head is about 3 microns.
My tool load is 1.2KG on full pass and 200 grams on return pass.
Return pass was cutting to deep on back cut.
This is LOW COST PRECISION.

[ihavenofish]

I don't need to attach it to the wall like others.
It does bend and is quite measurable.
1KG push down on the head is about 3 microns.
My tool load is 1.2KG on full pass and 200 grams on return pass.
Return pass was cutting to deep on back cut.
This is LOW COST PRECISION.

ah, i see now.

i think. i cant even measure 3 microns here!

heck, theres probably alot more than that in backlash on the Y screw when you switch drections on the x.

i imagine theres probably multiple sources for your cutting error adding up, but its really cool that your little device could compensate.

have you writen into into programs commands to move to each side when it changes direction?

[neilw20]

ah, i see now.

i think. i cant even measure 3 microns here!

heck, theres probably alot more than that in backlash on the Y screw when you switch drections on the x.

i imagine theres probably multiple sources for your cutting error adding up, but its really cool that your little device could compensate.

have you writen into into programs commands to move to each side when it changes direction?

Not worried about other errors. Just need a perfect surface finish in 1 pass without spending $100K. Small x/y errors hard to see, but a crappy finish stands out. Good dial indicator measures 0.0001" (2.5 microns) readily.

As cutter direction changes so does the compensation, and for each part of job it also fudges for bumpy Y axis too. Hand scraped it a bit, but why bother now.

Mine is small DTI is 0.002mm per division. Push with fingers on head is 1.5 divisions,

[neilw20]

Now I have had a bit of time, I made it stiffer.
Not a concrete fix, but a brickwall fix, with a bonus flouro lamp attachment.
Machining mark variations in Z now almost perfect.
Although near the wall the electronics (hinged) can still be swung open. A bit harder to get at the Z screw.
Now it needs a tiny packer under the front of the column. Leaning forward very slightly.
No criss-cross cut pattern in y move, but x is perfect, because head can be rotated.

[g29cc]

Interesting.