Belt stretch affects CNC mill accuracy?

Ignoramus6705 fired this volley in news:nOWdnb7aaM_lW83RnZ2dnUVZ snipped-for-privacy@giganews.com:

I think that for any given feed vector, you could calculate (or measure empirically) the stretch, and have a lookup table that adjusted the feed to compensate. That would probably be a difficult programming job, but anything is do-able. Back in the early 80's, I wrote Z-80 code that would allow a bar code wand to scan a code at _widely_ varying speeds within the body of the symbol (over 20:1 and varying rapidly), and still decode it reliably. This is that sort of task. A wide range of feed rates and feed depths would need to be accommodated, and you'd even have to account for materials in the adjustments -- but it's just software.

I'm not sure what EMC enables you to do in that respect, but since it's open-source, it's extensible.

The stretch on a belt is not immeasurable, but with modern Kevlar reinforced belting, I think it would be minimal (supposing the belt is not used near its design limits). I don't think the BP guys would have liked not being able to mill round pockets. No... I don't think they'd have settled for it, at all. Probably, their belts weren't 40 years old, either .

LLoyd

Hey Iggy,

If it isn't "round", it isn't "belt stretch" !! If it's something to do with the belts and pulleys, it may be wear, but not stretch.

How does yours do?

Brian Lawson

Good point. Stretch is proportional to current.

Well, my mill is 23 years old and belts are fairly well tensioned.

i

You'll never CNC Gcode to bore accuracy. You'd have to check your machine but a circle to +/- .002 with G02/03 would be awful good for your machine. For accurate bores get a boring bar. I'm partial to devleig microbore - they do 2 tenths easily and I own a complete set from .250" to over 6.0". I lock the gibs down tight while boring to get this.

define "precisely round"

Gunner

One could not be a successful Leftwinger without realizing that, in contrast to the popular conception supported by newspapers and mothers of Leftwingers, a goodly number of Leftwingers are not only narrow-minded and dull, but also just stupid. Gunner Asch

I would think it is minimal since an entire turn of the handwheel is 0.2" or so. I would guess belt stretch to make less than a half thou difference. But if it does affect your parts, you have the capability to put DRO scales on your mill and use them with EMC2 instead of the motor encoders. That's what my mill has, an Anilam control with Anilam DRO scales , the motors have tachometers on them, they are stable even with backlash.

Have you found me a Syncrowave yet? :-)

RogerN

Roger, there is no handweel attached to the ballscrews.

I had one for sale for $600 last year. I asked for 600 here and 750 elsewhere. No one from this newsgroup wanted it, so I sold it for $750.

i

He could also move his current encoders to be directly connected to the ballscrews, eliminating any possible belt issues from the positioning accuracy. He would cut the counts per inch in half, but I think it's high enough not to matter.

My Anilam lathemate lathe does that, the encoders are coupled to the ballscrew.

RogerN

Early on I have decided not to alter the hardware of this mill. What I am doing is relatively simple. Changing hardware is complicated and will ultimately ruin this mill due to my incompetence.

i

OK, one turn of the ballscrew.

I've found some on eBay for over $2000 but I'm not that interested in one. If I put my old one on a trailer to I can move it to my current address then it will probably be sufficient. When I got it I ran some beads on aluminum but had some trouble welding 2 pieces together, a stainless steel brush and some more practice would have probably done the trick. I'm hoping to get a tank of Argon soon but for right now I should concentrate on getting the truck done and sold.

RogerN

No, but his point was the one full revolution of the ballscrew is a pretty small movement, and any belt stretch issues aren't likely to cause more than a couple degrees of ballscrew position error. I think you said 2.5mm ballscrew pitch, so perhaps something like 0.0006" positioning error under belt stretch? Backlash and proper backlash compensation will be a much bigger factor in accurate circular interpolation.

[ ... ]

The "virtual" handwheel. :-) It is still 0.200" per revolution of the ballscrew or the nut.

If yours is like mine -- the ballscrew for the X-axis does not rotate at all -- instead the ball *nut* rotates in matched bearings. And the ballscrew has a telescoping flat spring steel cover to keep swarf off of it.

Y-axis, however, does have a rotating ballscrew.

BTW -- looking at the photo you just (re)posted of the mill on the truck or trailer I now see that the Y-axis motor has the pulley end towards the vertical ways in the back instead of facing out towards the operator as mine does. I still should be able to mount the motor on the side of the knee -- if I can find a flat enough surface. :-)

Enjoy, DoN.

Baloney! Now, there is some tiny amount of stretch to the belts, but it is certainly small, in the fractions of a thousandth of an inch. I can't say about a Series-II BOSS machine, but there are other sources of apparent backlash. There's compressibility of the screw, rocking of the table when direction reverses, the constraint of the bearings that hold the rotating nut, and so on. My gut feeling is that these other sources of backlash, PLUS finite servo amplifier gain, will be much greater than the belt stretch. It almost certainly adds up to several tenths of a thousandth of an inch, and on an older machine, could be much worse. if the preload on the ballscrew has all worn off, then it will be in the several thousandth of an inch region.

If you need a really accurate and round bore, for a bearing or an engine cylinder, you'll never get it on anything but a VERY high-end machine tool in tip-top condition. When I do any sort of precision bore, I use a boring head for the final finish. This WILL be round.

I have heard of people in the industry with $100,000+++ machines that DO, indeed, mill precision bores in production. But, they have specially tested these machines to be capable of maintaining that level of precision and lack of backlash, and maintain them to keep them performing at that level.

Jon

Oh, and finally, if you don't have a Renishaw ball-bar tester, you just don't have the gear to find out what your machine really is capable of. You can try milling a bore and measuring the fit, or sweeping the ID with a tenth-reading DTI, but that will only give you a rough idea.

Jon

I saw you had them listed but by that time you had already changed your mind about selling it. I found a Syncrowave 350 on eBay around 400 miles away with a 499 starting bid and no reserve. Bigger than I want but a syncrowave

250 isn't exactly portable either. If freight isn't too high for 400 miles I may bid $750 minus freight charges, thinking it would be worth $750 delivered. It doesn't have the TIG torch, water cooler, and I didn't see a foot pedal. I already have a torch and water cooler on my old TIG so I'm ok there. If I knew the foot pedal specifics I could probably get my existing foot pedal to work on the Syncrowave too.

RogerN

Foot pedals are very simple. I actually have a pile of them.

i

Do you know if the Syncrowave 350 is the same as the 250 beside the extra weight and amps? Just wondering if I should bid on this item, assuming freight is reasonable, or wait for something closer or a syncrowave 250.

RogerN

It is an older and different machine. I resold one before (bought for $200 sold for $800) ask ernie.

i