is my gear rack is inaccurate? how to isolate

[IsaacH]

Hi folks,

I appreciate this forum. I lurked on it for the last 9 months and built a decent machine. Never posted because I could always search and find the answers I needed. Thank you to everyone who contributes to problems and helps the group learn.

I'm having a problem with my rack and pinion setup on my Y axis. I use Y for the shorter of the two major axis (gantry).
I think the gear rack is inaccurate. I have parts come out undersized on the Y0-Y20.000 area and oversized on the Y20-Y48.000 area. I thought my steps were calc'd wrong, but updating those values results in even worse parts.
I made a test program which cuts a slot every 5". Used calipers and measured the slots. They were inconsistent. some were .004 or -.012 off, some were worse. They did not scale linearly. I identified areas that were right on the money and some that were not.

Gear rack was purchased from McMaster. I appreciate that company but perhaps their gear rack was not accurate enough.
Gears and belts are from CNCRP. I don't suspect them as the problem as many have used that system without errors (I believe).

general setup:

I'm using UCCNC with UB1 board.
Linear rails and blocks from Hiwin.
Frame is welded. Nema 34 with Gecko V203 drivers.

X (longest) axis works fine. I sample measured 1" sections of X with dial indicator and it hits 1.000" pretty well. I used a long tape measure and told it to go 94" and it hit it pretty well. Same mcmaster rack used.
During calibration Y axis steps calculate differently then X. X calcs at 2037.xxxx and Y comes in at 2040.xxxx.

I have removed the track and ran the gear alone on it. Some areas did not run as smoothly. The same areas as seemed less accurate from the cuts.
I'm thinking of buying some track from Moore Gear to see if it fixes it because I'm out of ideas.

Looking to hear if others have had problems like this with rack and pinion and what tests I could conduct to isolate and confirm the gear rack as the problem.
Thank you

I used this rack:
https://www.mcmaster.com/5174t21

with this gear setup:
PRO Rack and Pinion Drive, NEMA 34 | CNCRouterParts

Similar Threads:

• Gear rack
• Gear rack for the Z ?
• Spur Gear / Gear Rack
• rack gear

[Eldon_Joh]

take the y axis gear rack off and mesh it with the X gear rack. you should easily be able to see if its of inconsistent pitch.

as for the friction in the linear rails.. that's a lot more expensive to fix. clean them out..


"I have removed the track and ran the gear alone on it. Some areas did not run as smoothly. The same areas as seemed less accurate from the cuts. "

what does that mean? i interpreted that to mean the linear rails are stiff but you could have worn out that unhardened 1018 steel rack.

[IsaacH]

take the y axis gear rack off and mesh it with the X gear rack. you should easily be able to see if its of inconsistent pitch.

as for the friction in the linear rails.. that's a lot more expensive to fix. clean them out..


"I have removed the track and ran the gear alone on it. Some areas did not run as smoothly. The same areas as seemed less accurate from the cuts. "

what does that mean? i interpreted that to mean the linear rails are stiff but you could have worn out that unhardened 1018 steel rack.

I don't know if fiction in linear rails is the issue, but I didn't know that it's not. I thought the big nema 34s on 3:1 reduction would be enough to overcome most small fiction. I have lubricated the blocks. I have taken off the rack and meshed it with the x axis and couldn't see any real obvious issues.
Tonight I am going to flip the y axis rack and see if it results in opposite innacurate areas to try to isolate the problem.

[wizard]

Friction, especially variable friction is a bad thing in a machine tool. On larger machines all sorts of techniques are used to over come friction to eliminate its negative impacts. A very close relative of friction is mechanical binding, by being is best characterized as mechanical parts in mis alignment resulting in those parts deflecting as an axis translates.

The first thing I considered when reading your post is the possibility of the pinion binding on the rack. A bad rack isn’t impossible but northern is the possibility of a gear with an offset bore. It is kinda hard to tell from your post exactly what is going on. In a nut shell though your axis carriages must move one one extreme to another without a noticeable variation in movement effort. The precise amount of effort is not as important as not having huge variations in effort across the axis length. This isn’t to say having a high effort to move is not a problem; it is but smoothness of movement is very significant.

To put this into perspective with respect to old machinery,, Bridgeport mills would often wear the table bearings in the middle of the ways. Eventually Gibs would be adjusted to remove slop resulting ina tight fit outside the common work zone. This would often translate into poor dimension control over long parts. Binding means that not only are you changing the force required to move and position but you are displacing something. Displacement generally means an axis isn’t in the position desired. So in your case if you are feeling significant binding then most likely something is being deflected a bit which could easily cause a 0.001”(or far more) error at the tool.

I don't know if fiction in linear rails is the issue, but I didn't know that it's not. I thought the big nema 34s on 3:1 reduction would be enough to overcome most small fiction. I have lubricated the blocks. I have taken off the rack and meshed it with the x axis and couldn't see any real obvious issues.
Tonight I am going to flip the y axis rack and see if it results in opposite innacurate areas to try to isolate the problem.

[adam_m]

It seems to me that if you're using the same rack for your X axis problems would show up there as well can you swap some rack with the X axis to test if its rack?

I'd be more inclined to think its a stepper/steps issue, can you swap the stepper maybe with one on your X axis or your driver with another driver?

Adam,

[wizard]

Hi folks,

I appreciate this forum. I lurked on it for the last 9 months and built a decent machine. Never posted because I could always search and find the answers I needed. Thank you to everyone who contributes to problems and helps the group learn.

I'm having a problem with my rack and pinion setup on my Y axis. I use Y for the shorter of the two major axis (gantry).
I think the gear rack is inaccurate. I have parts come out undersized on the Y0-Y20.000 area and oversized on the Y20-Y48.000 area. I thought my steps were calc'd wrong, but updating those values results in even worse parts.
I made a test program which cuts a slot every 5". Used calipers and measured the slots. They were inconsistent. some were .004 or -.012 off, some were worse. They did not scale linearly. I identified areas that were right on the money and some that were not.

The first thing to realize here is that routers are not milling machines and light weight routers even less so. It is your expectations need to be realist with respect to the machines mechanical abilities. The second issue is that there are many possibilities that could be generating these Y axis errors, including issues with the Z axis. More info on how the slots where cut would also help.

Given that 0.004” might not be that extreme especially if calipers are used to measure the slots. The materials being cut can be a factor too. 0.012” is starting to be excessive though.

Gear rack was purchased from McMaster. I appreciate that company but perhaps their gear rack was not accurate enough.

You ave not indicated that you have gotten to the root cause yet. So I would not get to excited about pinning the issue on one part. Is it possible the rack is defective - certainly but also just as likely not a problem.

Gears and belts are from CNCRP. I don't suspect them as the problem as many have used that system without errors (I believe).

The fact that some have used those parts means nothing here. There is always the potential for error or defect. Trying to debug a machine based on manufacture reputation is foolhardy. It is better to keep one mind open because often the root cause is a surprise. That doesn’t mean that the rack isn’t a problem just that it hasn’t been proven.

general setup:

I'm using UCCNC with UB1 board.
Linear rails and blocks from Hiwin.
Frame is welded. Nema 34 with Gecko V203 drivers.

X (longest) axis works fine. I sample measured 1" sections of X with dial indicator and it hits 1.000" pretty well. I used a long tape measure and told it to go 94" and it hit it pretty well. Same mcmaster rack used.
During calibration Y axis steps calculate differently then X. X calcs at 2037.xxxx and Y comes in at 2040.xxxx.

Did you change the gear ratios to come up with those different numbers?

I have removed the track and ran the gear alone on it. Some areas did not run as smoothly. The same areas as seemed less accurate from the cuts.

I don’t follow this statement at all!!! What track did you remove and why?

Generally when dealing with axis problems, at least on smaller machines, it is best to remove the drive mechanism and move the saddle by hand. You do this to make sure the linear rails themselves are not binding. In the case of a rack based system this is also a good time that verify that the rack is parallel to the plane the linear bearings for the saddle create. Ideally that rack should be parallel to the linear bearings to under 0.001”.

It just occurred to me that we don’t have any indication of how you are engaging the rack. For example some use spring loading to attempt to control backlash. However if there is binding resulting in a significant load increase the spring may be causing some of that locating issue.

I'm thinking of buying some track from Moore Gear to see if it fixes it because I'm out of ideas.

From what you posted here I think it is too early to do that. There are all sorts of things that could potentially be the cause here. For example the pinion could have an issue. I would imagine that there is some ratio between the pinion and the stepper, if that is beltdrive a pulley bore could be off. Beyond that the Z axis could be the problem. Let’s not forget that spring loading if used.

Looking to hear if others have had problems like this with rack and pinion and what tests I could conduct to isolate and confirm the gear rack as the problem.
Thank you

I used this rack:


with this gear setup:

Are these parts the same as used on the X axis. A quick swap might allow you to isolate the problem area.

Also pictures / videos might help. Especially of the Y axis test runs.

You didn’t mention the materials you are using for testing. I’m kinda hoping it isn’t wood, at least not wood that doesn’t cut cleanly. Even grain direction can impact how well the cuts measure up.

Speaking of pinions and gear reductions, a black marker comes inreal handy to place an index point on your pinion gear. The idea here is to see if the binding repeats or follows a pattern across the rack, pinion or both.

I hope some of this helps! Ideally we can prevent spending more money before we have a higher probability that we are on the right track.

Speaking of that tape measure, it isn’t a bad idea as long as we realize that they are generally not that accurate. It n your Y it might not hurt to mount a long scale permanently to make the same checks and to have a long term reference. Scales can also be terribly in accurate but there are quality options to consider.

[ger21]

I made a test program which cuts a slot every 5". Used calipers and measured the slots. They were inconsistent. some were .004 or -.012 off, some were worse. They did not scale linearly. I identified areas that were right on the money and some that were not.

Decrease that axis acceleration and velocity by 50%, and try it again.
If it gets better, than your probably losing steps.

[adam_m]

I also have rack and pinion from CNCRP on my machine... it works very well.

Wizard, the engagement on those units are by spring tension by either a turn buckle or bolt and sprint compression.

Adam,

[IsaacH]

Thank you everyone for suggestions. I flipped the rack over and rotated it 180 deg to see if the issue would move. It did not.
After moving the Y axis back and forth in 5" increments I'm beginning to see a pattern of difference happening between the outside of the gantry and the middle. That is, the middle hits closer to 5.000 and the outside is closer to 4.960 and varies. One side worse then the other. This happens regardless to whether the rack was flipped, which sort of eliminates the rack alone as the culprit.

I did move the machine at 100 ipm and at 50 ipm and 20 ipm and didnt' see a noticeable difference in measurements. I reduced acceleration to 20 from 60 and no noticeable change.

In UCCNC when I calibrate the Y axis, I send it 40" and run it along a tape measure. When I have it hitting 40" as close as I can, the steps display at about 2040 per in. vs 2037.xxxx for the X axis.

The exercise of sending the machine in step mode in increments of 10" shows something:
With a small .125 bit just above a tape measure I set 0 and send it from 0-10" it will be about 9.95.
Then if I send it 10 more inches from 9.95" it will sum to 19.85.
Then 10 more inches will end at 29.9.
Finally I send it 10 more inches and it catches up to 40.

I think wizard may have suggested a possible thing to look into, excess friction or binding. I had not been concerned about it because I thought my stepper was large enough that it would overcome it, but it may not be the case.
A real possibility is that due to how I installed my gear reducer the gear is just slightly off perfectly seated in the track due to sag. See the pictures I have attached.

Test runs were cut into MDF with a .125 endmill at .05 deep. It's possible my work with the calipers isn't perfect, but i did try to measure several times.
Lately tests have been using calipers and dial indicator clamped to the linear rail and touching the body of the z axis. Just to remove any spindle flex or endmill deflection.
The machine is very rigid, but I did weld it, so it's possible that the gantry has some slight warpage toward the ends where it was welded which are contributing to the binding/friction.

Thanks for much for all of your suggestions. I'll go out to the shop now and try moving it around without it connected to the stepper and see how smoothly it works.
As I said, I really appreciate all the helpful comments!

[ger21]

The exercise of sending the machine in step mode in increments of 10" shows something:
With a small .125 bit just above a tape measure I set 0 and send it from 0-10" it will be about 9.95.
Then if I send it 10 more inches from 9.95" it will sum to 19.85.
Then 10 more inches will end at 29.9.
Finally I send it 10 more inches and it catches up to 40.

Try sending it to 15, then back to 10 and measuring, and see if you get different results. Then 25, and back to 20. And so on. This may tell you if there's any backlash present.

[IsaacH]

Try sending it to 15, then back to 10 and measuring, and see if you get different results. Then 25, and back to 20. And so on. This may tell you if there's any backlash present.

Thanks ger21
I have several times done backlash tests both in that manner with a tape and with a dial indicator and did not measure notable if any backlash.

I pushed the z axis assy back and forth along the y rails last night and the action is not free, it takes some effort, but without a strain guage I'm not sure how much. I will try this afternoon loosening everything a lot to get the assy sliding super easily and see if that results in different values.

[adam_m]

Just to get some clarity on your machine setup. You're using R&P on all 3 axis, 3 same stepper motors, the same rack and the same drivers, correct?

If that's the case I would start swapping parts... if your X axis is working fine I would make a note of the current position of the steppers and rack and pinion drives and then swap a rack and pinion with the Y axis, re-test.

If there's no change there, then I would swap it back and I would move to the drivers to try and isolate if its a driver problem.

[IsaacH]

I am using R&P for two axis, ballscrew for Z.

Today I swapped one moter/gear assembly, no change.
I then put a tape measure up next to the rack and clamped a small indicator (screwdriver) to the gear area right above the tape measure so I could see exactly where it was.
I made the machine do the usual increments of 5. It was perfect. Right on the money.

Did the same test down at the spoilboard using the endmill as a probe on the tape measure. It was as off as always.
I believe I have confirmed the moter/gears/rack as fine.

It appears that as the assembly slides along the Y axis it must shift/bind/ ever so slightly causing it to rotate and the endmill, being far away from the center of rotation shows the effect by being off by about .100 at the worst point.
thus the spindle becomes slightly out of tram and then comes back into straight again at the ends of the y axis.

Guessing I'll just have to do a lot of loosening/fiddling to try to figure out what is causing the spindle/z axis assembly to shift when it's in the center of the gantry. One thought I had was that the two linear rails aren't parallel enough and the carriages have so little play that they are causing the rotation. I'll try to isolate this by disconnecting the lower set and see.

Anyways, thanks for listening/helping.