Timing belt backlash??

I've just been testing a timing belt drive that connects a Lichuan servo motor to the cross slide ballscrew of my lathe. The parts I used are:

150-3M-15 Timing Belt (150 long belt)
32-3M-15 Metric Pilot Bore Timing Pulley (32t driven)
16-3M-15 Metric Pilot Bore Timing Pulley (16t driver)

I have the belt nicely adjusted but when I turn the motor against the locked ballscrew using my fingers, I can clearly see the best part of a mm or so of movement between the pulley and the belt at each pulley before the teeth of the belt and pulley actually make contact. This is crap, as it equates to something like 100um of backlash and sort of negates the cost and ballache of bothering with a half decent ballscrew and bearings. I'm fairly certain this isn't to be expected.

The belt is a Transflex brand and the pulleys are fairly generic looking items, all from Bearingboys in the UK.

Has anyone seen similar issues and if so, what would you suggest? I could buy (hopefully pukka) pulleys from somewhere like RS in the hope the pulleys are the issue.

I may have to get out the calipers and make some measurements but I don't have a standard or drawing to compare against, only the belt and the pulleys themselves. It feels to me that the pulley teeth aren't deep enough, causing the belt not to seat fully. Not something I was expecting to have to deal with....

Duplicate

Nobody seems to be able to reply to this post - I must have annoyed the timing belt gods!

I hope you don't mind me pasting the guts of your PMs as a sort of manual reply method:

From Voicecoil:
I used AT5 belts (Syncroflex) and pulleys from Beltingonline and got good results - approx +/-10...15um final accuracy. There are different tooth profiles available for belts, the main ones being trapezoidal (lowest backlash) and trapezoidal (best power transmission), I'm wondering from what you said about the teeth not seeming to reach the bottom that maybe your belt is a different profile from the pulleys? Hope this is of some help, Trevor

From m_c:
Have you tried knocking a flange of a pulley, so you can see how well/poorly the teeth mesh?
They should be a pretty tight fit without any movement.

I'll have another look for alternative versions but after a first look I couldn't find an equivalent small pulley of a similar diameter to the 16t one. However, I have a flangeless pulley on the motor, so may be able to look more closely at how the belt and pulley mesh. I'll look at beltingonline as an alternative supplier, as I can't help thinking this is a quality issue.

Thanks for the replies!

Hmmm... I am not sure if this is entirely relevant but it may help so I offer it in that spirit. I have a Shapeoko S3 CNC machine that uses belt drive for each of the stepper motors (NEMA 23) and I first got the machine 3 years ago from new. I was unable to keep the belts tensioned correctly because they went through a right angled connector that was difficult to use well. Some symptoms were uneven tensions, size inaccuracies and the inability to find the best tension for the belts.They kept slackening at varying frequency but every few days. I wanted to mill metal and there was no way that I could even predict the result with my poor set-up.

The belts were 9mm wide GT2 timing belts supplied by Gates. I thought the belts were good quality so I set about looking for a better adjustment solution. I found a 3D printed item that kept the whole belt absolutely flat rather than doubled back on itself after following the right angled bracket and it looked to be a much better solution. I modified the end supporting plates of my CNC machine and attached the 3D printed revised adjustment mechanism. Gates provided an application to tune their belts by sound but it was very fussy to use and difficult to get an accurate reading. I tried a simple guitar tuner which worked on vibration to detect the frequency. This worked really well and I discovered that the tension of the 9mm wide GT2 belts, using the Gates data, could be anywhere from 100 to 150 Hz. I tuned my belts to C3 on the guitar tuner (130.81Hz) and they have stayed at that tension for 14 months!

You may also find this article by Liam Newcombe to be informative. It has a lot to say about belt deflection and backlash.

https://community.carbide3d.com/t/ba...ibration/28669

I had better luck with cast iron toothed pulleys when CNC'ing my mill :beer:

Quote:

---

Originally Posted by Robin Hewitt

I had better luck with..:

---

Ooooooh, beer! :beer:

Hi Muzzer this is an opportune post of yours as I have a very similar problem with HTD drive to a lathe cross screw.

Small drive pulley 12 T and large 26T driving a 1605 ballscrew via a rotating ballnut.

The measured lateral backlash with the shaft locked is 40 half steps or 9.3 thou—total rubbish. I can actually see the belt moving and taking up the slack while the driven pulley is stationary.

The only reason I used HTD was the rotating ballnut came fitted with it. To be fair to the manufacturer it was designed for high speed, low force and low precision and I want the exact opposite of all 3.

To compound matters, the open loop stepper I was using on the supposed half step actually steps more like 1/3 step and then 2/3 steps. Ho Hum.

So plan is to change to AT pulleys with an ATL belt after changing the stepper motor to a closed loop one with encoder etc.

As far as I can ascertain, this combination of AT pulley and ATL belt gives the least backlash, although the availability of ATL belts is sketchy.

Cheers
David

Mine was rather extreme too:
https://www.youtube.com/watch?v=aXv_H8yqnt0

I ended up making my own pulley, as it seems that many of the pulley suppliers make their pulleys from the same extruded stock made in some shack in China. Consequently, trying pulleys from other suppliers may simply give you more of the same duff pulleys.

https://www.youtube.com/watch?v=4g2qvoNKiqc

This fixed the issue but was rather an extreme solution!