Ball screw travel error

billzweig · Sep 18, 2018

I've recently converted a mill to CNC . The X axis is a THK C3 ground ball screw. I got it used via Ebay from Korea – in great shape and looks like new - with ball-nut and bearings. It is a metric screw with 2.5 mm pitch and marked as such, together with the THK logo, on the nut assembly. The ball screw assembly was installed very accurately with the table parts bored together in the same setting and the nut housing secured with taper pins and screws. The total backlash of the table is under a couple of microns (little less than one tenth).
When advancing the table it moves further than it supposed. 10 turns should move it 25mm, but it actually goes 25.03 - 25.04. 20 turns moves it about 50.08 and so on. This is true in both directions. The most intriguing is that the error is not consistent, but always on the plus side. And any slip or backlash would result in a shorter travel.
Initially I though it is a problem with the stepper or the software, but it is the same when I've installed a hand-wheel with a dial directly on the screw end.
Any thoughts on this?
Thanks,
Bill.

DMF_TomB · Sep 18, 2018

billzweig said:
I've recently converted a mill to CNC . The X axis is a THK C3 ground ball screw. I got it used via Ebay from Korea – in great shape and looks like new - with ball-nut and bearings. It is a metric screw with 2.5 mm pitch and marked as such, together with the THK logo, on the nut assembly. The ball screw assembly was installed very accurately with the table parts bored together in the same setting and the nut housing secured with taper pins and screws. The total backlash of the table is under a couple of microns (little less than one tenth).
When advancing the table it moves further than it supposed. 10 turns should move it 25mm, but it actually goes 25.03 - 25.04. 20 turns moves it about 50.08 and so on. This is true in both directions. The most intriguing is that the error is not consistent, but always on the plus side. And any slip or backlash would result in a shorter travel.
Initially I though it is a problem with the stepper or the software, but it is the same when I've installed a hand-wheel with a dial directly on the screw end.
Any thoughts on this?
Thanks,
Bill.

most dro or digital readouts have a error compensation or calibration mode. you use a indicator zeroed to a 6.0000 gage block and zero dro then remove gage block travel to stop gage block was against til indicator is zero if dro says 6.0010 you press a special mode button and tell it it moved 6.0000. then you retest and confirm 6.0000 of movement reads 6.0000 on dro
.
i did this every time the dro scale reader was adjusted and its recommended to do once a year. quite normal to correct a .001 or .002" error

HuFlungDung · Sep 18, 2018

It'll be spot on when the machine is at -40°C

JST · Sep 18, 2018

Things have a tolerance, ya know...... What is the accuracy spec on the screw? Is it run open loop, as many are? Has there been any error correction done?

EPAIII · Sep 18, 2018

0.03mm off in 25mm is 0.12% error. Another way of saying that is it is about 1 part in 1000 or 0.001" per inch. Were you really expecting better than that from a screw from Korea? They don't seem to even publish specs on their web site.

THK Distributors of Motion Control Products & Automation Components

You say the error is not consistent. Have you checked for play in the nut assembly? How about the bearings used to mount the screw?

Use the calibrate function if you have it.

billzweig · Sep 18, 2018

I can apply correction in the control software and compensation in DRO. But what can make a C3 screw error so large? The accuracy spec on C3 screw, according to THK, should be much, much better. And gain, how the error can be on the plus side? A worn-out screw could only err on the minus side (and will show huge backlash).
As for setting it up at -40°C...not a bad idea, but I am not quite ready to move the shop to the Yukon or the Northwest Territories.

Milland · Sep 18, 2018

How are you reading distance traveled? As Tom mentions, a good reality check is to reference gauge blocks over distance, rather than just trusting a DRO or other electronic or dial reading.

Set up an abutted row of blocks, true to the direction of travel and offset one to another (like a square wave) so you can touch off with a good tenths indicator. You can use the blocks in an 81pc set, just account for the differing lengths of travel for each new block pair (one block for reading, one for spacing).

With care you should be able to map out the actual distance to whatever accuracy your block set has, should be good to around five microns over 250mm at least. Cleanliness of the blocks, good wringing, true to travel axis, temperature you're testing at - all will effect your final accuracy.

Is this a V-way machine, or did you set up linear rails too? Play in the way could affect readings, but then I'd expect some plus or minus values over travel distance.

billzweig · Sep 19, 2018

The mill is a high precision scraped V ways machine. I did not get any way near to the gauge block accuracy yet. I am using a Mitutoyo 50mm travel mechanical gauge that I've calibrated on a Moore jig borer. It has a small error, a few microns in various positions - but way less than the ball screw travel error. And THK is a good maker, I have used their ball screws before. As much as I am reluctant to dismantle the slide, I think I'll do it and test the ball screw directly. Maybe I am missing something here...

9100 · Sep 19, 2018

A knee mill will wear the table in the center and the ways it sits on at the ends so the table does not move exactly in a straight line but in an arc. A machinist I know installed a DRO on a considerably used BP and had the same problem. When he complained to the DRO dealer, they replied RTFM. Sure enough, it was all explained in the manual. Since the table is moving in an arc, the DRO will only agree with the leadscrew if they are on the same plane. Since I do not know your mill, I don't know if that applies here.

Bill

billzweig · Sep 19, 2018

Thank you, Bill, for this suggestion. Yes, this is in fact known as Abbe error. By coincidence (great minds think alike...

) just last night I though about this and have checked the travel with the indicator mounted at the height of the lead-screw centre. The error is indeed a few microns smaller but still almost 10 times the spec. I shall have the final verdict once I remove the screw and nut and check those on the bench but it does look that the pitch is for some reason more than the indicated 2.5mm.

Yan Wo · Sep 19, 2018

HuFlungDung said:
It'll be spot on when the machine is at -40°C

What about at -40°F?

motion guru · Sep 19, 2018

We just spent a crap load of time and money dealing with what turned out to be a low quality Chinese bearing in a surface grinder. I did some research and it is not uncommon at all for a Chinese “re-branding” shop to take low quality bearings, ball screws, and various other parts and mark them SKF, or THK, or NTN and sell them as such. They copy the boxes and packaging and sell as faux high quality Japanese or German components.

We just replaced a 51156 thrust bearing in a machine that had 27mm diameter bearing balls that varied +/- 0.0003 inches. The genuine replacement ********* bearings had no discernible variation in bearing ball diameter.

We would grind a chuck to within +/- 3 tenths and two days later it would be out +/- 0.002”!!

The kind of inaccuracy you have indicated would have me thinking you have a cheap Chinese knock-off.

Milland · Sep 19, 2018

motion guru said:
The kind of inaccuracy you have indicated would have me thinking you have a cheap Chinese knock-off.

One of the reasons I've been hording NOS spindle bearings from known-good suppliers. My thought is they're much less likely to be knock-offs, as the ability to make real looking packaging wasn't as advanced back 20+ years ago.

Or so I hope...

billzweig · Sep 20, 2018

Milland said:
One of the reasons I've been hording NOS spindle bearings from known-good suppliers. My thought is they're much less likely to be knock-offs, as the ability to make real looking packaging wasn't as advanced back 20+ years ago.

Or so I hope...

On more than one occasion I've decided to spend more money and to buy a reputable US product just to find out that it was made in China...and an identical item available for fraction of the price as import.
As for my C3 THK ball screw it is a good quality ground screw, though possibly originally some custom pitch (and yes, possibly relabeled and then sold as standard). And it is about 1.002 larger pitch.
As for the idea of low temperature correction....I would need to cool it in liquid nitrogen (-200C) to get it correct

.

johansen · Sep 20, 2018

billzweig said:
As for my C3 THK ball screw it is a good quality ground screw, though possibly originally some custom pitch (and yes, possibly relabeled and then sold as standard). And it is about 1.002 larger pitch.
As for the idea of low temperature correction....I would need to cool it in liquid nitrogen (-200C) to get it correct .

what are the chances they could have taken a rolled ball screw and grind it to make it look like a ground screw? The THK logo look legit?

some kind of clever follower rest and some spring preloaded ballscrew nuts could be setup to take a rolled screw and then grind it to improve the pitch diameter variation, but let the screw self feed though the machine so there would be no improvement to the pitch accuracy. i don't know how anyone could make money doing that though.

custom pitch for a liquid nitrogen application would be interesting, but why not just correct the difference in software.

Search

Largest Manufacturing Technology Community on the Web

Search

Ball screw travel error

billzweig

Stainless

DMF_TomB

Diamond

HuFlungDung

Diamond

JST

Diamond

EPAIII

Diamond

billzweig

Stainless

Milland

Diamond

billzweig

Stainless

9100

Diamond

billzweig

Stainless

Yan Wo

Stainless

motion guru

Diamond

Milland

Diamond

billzweig

Stainless

johansen

Stainless

Similar threads