Worn leadscrew?

[Husky1]

I wonder what would cause this error on my parts. Till recently, all worked well, but on my last 5 parts i noticed error on x axis. It is the same error on all 5 parts. It is cut circle and there are also 40 x 7.1mm holes made circular around that circle. Error is 0.5mm only on x direction. See the illustration bollow what i mean:

https://i.imgur.com/wB1vJnK.png

The holes positions have same error, so i assume it is worn leadscrew / nut, but when i move the x axis by hand, i dont feel any movement, so i wonder, is this mechanical error or is there some electrical issue? i use chinese servo motors (bst motion, krs?? 400w) and cslab csmio ips controller.

[Husky1]

anyone?

[Doddy]

I've a couple of questions - really out of curiosity (and boredom - 2 weeks away from my machines) than progression to a solution, and I have zero experience of servos. But you should be able to answer the following with only a short time in the workshop.

Firstly, what, if anything, has changed since the machine worked well? I'm assuming that you'll reply will you have not deliberately changed anything.

Presumably you can swap the drive/positional feedback between the X/Y servos?, have you tried this?, what's the result? Then repeat with swapping the X/Y servo drivers. You should be able to isolate to either servo or driver in error... if not, then it's likely mechanical.

You don't mention: Is your part a progressive multi-depth cut?, If you was losing steps then you would expect any error to accumulate and might expect to witness this as you cut through the material. I'm inferring from your post that this (accumulated error) is not the case, and the error is repeatable.

You don't mention: If you move the workpiece centre elsewhere on the X axis and repeat the cut does the problem remain? If it did this would tend not to support the leadscrew argument.

Have you tried any measurement of x-axis positioning? get a sheet of material on your bed and engrave a rule (Y stroke every 10 mm across the X-axis, say) - from X=left though X = max. Cut one rule in the left-to-right direction, and a second in the right-to-left direction. In each case ensure that the X-axis velocity is expected to be zero when you're engraving the Y stoke (this is just to test the X-positioning). Are these the same? (If not I'd be checking the fasteners on the ball nuts, ball-screw supports, spindle and even machine bed).

Repeat but with a hand-crafted sinusoid (y = scale * sin (x * some-fettle-factor)) (cutter permanenting engaged with material) so that you're now testing the positional accuracy with the x-axis under load/ and not coming to rest) - measure the positional accuracy of the Y=0 crossing point. I'm not sure what this would prove other than ensuring that there's no significant time-constant in the servo positioning (is this the servo controller gain?).

Is this with your DIY spindle?, if so you should be very intimate with the build of the spindle - are you confident there's no play in the spindle under load? Again, check fasteners also.

[JAZZCNC]

The holes positions have same error, so i assume it is worn leadscrew / nut, but when i move the x axis by hand, i dont feel any movement, so i wonder, is this mechanical error or is there some electrical issue? i use chinese servo motors (bst motion, krs?? 400w) and cslab csmio ips controller.

If the drives were tuned correctly and you changed nothing and the error is repeatable then it's a most likely mechanical issue. However, I'd expect you would feel 0,5mm easy enough. That said it's the first place I'd go looking before started messing around with tuning. Set the drives to show current and see if they pull higher current than usual or in different parts of travel, they could be binding in one spot and causing the motors to fall behind slightly but then catch back up, this wouldn't fault the drives but the error would still show on the part.!

After this, I'd look for loose electrical connections. Then move onto re-tuning the drives.

[JAZZCNC]

Presumably you can swap the drive/positional feedback between the X/Y servos?, have you tried this?, what's the result? Then repeat with swapping the X/Y servo drivers. You should be able to isolate to either servo or driver in error... if not, then it's likely mechanical.

Doddy it's not that simple with Servo's Esp if running analog (And if Husky is who I think(Mat) then sure he is running analog) because each drive is tuned to the motor and it's load, so it's a lot of work not just a simple switch over like steppers.


You don't mention: Is your part a progressive multi-depth cut?, If you were losing steps then you would expect any error to accumulate and might expect to witness this as you cut through the material. I'm inferring from your post that this (accumulated error) is not the case, and the error is repeatable.

He's using AC brushless Servos so there are no steps involved. The encoders deal with position and feed it back to the drive so corrections can be made. You get what's called following error so if drives not tuned correctly to the motor and it's load then the error becomes larger, up to a point then drive will fault.
A correctly tuned setup will keep this following error to a minimum which doesn't affect part quality. But if anything changes like start putting much heavier than usual loads on table or binding or mechanical issues then the following error becomes greater and will show in the part. If the following error is on the edge of acceptable or binding etc is isolated just to one spot then because of how servos work the servo loop always allows the motor and position to catch back up so don't get a fault. The allowable following error can be adjusted in the drive so it could be set to a lower value which will help catch any trouble before the part is ruined, but it's a fine line between right and wrong.! I've also given a very simplified explanation of servo tuning , in practice, there's more to it.

[Husky1]

Thank you very much for the replies. I havent had time for the machine since i am away most of the time during the holidays. I will work on the problem next week and see if i could resolve the issue. I did have 5 minutes to check the voltage on the drives and this is what it shows (i could set only voltage info, very confusely translated servo manual) :

https://i.imgur.com/OhkgXSr.jpg

I see lots of voltage deviations, could this be an issue?

[Husky1]

I have engraved multiple rulers (10mm spacing) on a blank plate. First i engraved with x axis (from right to left and then vice versa) and the the same for y axis, overall 4 rulers. The result is the same as it was on my 5 parts with the error. X axis has 0.5mm error, y axis is ok. Engraving from left to right and vice versa doesnt show any error when comparing those two cuts. So i guess no mechanical error.

On the pics bellow it shows where the error appears. It is in the middle area, approx 0.5mm error. I circled where the camera shows correctly because of its lens.

2702527026270272702827029

On following pics i fixed the ruler in the center of the engraved ruler and it shows the same error as on the part:

2703027031270322703327034

This must be something with the servos. Is this fixable with servo tuning considering i have no clue how to fine tune the servos?

[JAZZCNC]

That just looks like your steps per setting is wrong. It's not moving the amount you think it is.? The first thing you need to do is check that it moves the amount you tell it with one move. Using a ruler to measure against isn't accurate enough you need to measure with digital calipers at least or ideally linear scale.

[Husky1]

I put the ruler there just for illustration of the error. Today i played with mach3 sets steps and the digital caliper and it shows exactly the same error. On pic1 i calibrated the x axis in mach so it moves exactly 460mm, then i set caliper to 200mm and placed it on 3 different locations, left, right and center. Pic2 shows what error machine makes. Have no idea how to fix that. I will inspect the connections next and swap motors. Maybe the encoder is faulty? Just guessing.
2706827069

[JAZZCNC]

How are the motors connected to the ballscrews.?

[Husky1]

With htd5 belt, 1:2 ratio.

[AndyUK]

Clearly something in the system is non-linear. Have you measured that the pitch of your ballscrew is constant along it's length?

Have you measured that DRO error in both directions? What's the Backlash?

Otherwise I don't know how it could be anything but a broken encoder or some screwy software based correction.

[Husky1]

I will check if the pitch is constant.

On previous post i posted picture where i engraved ruler in both directions, no visible deviations, backlash is minimal, around 0.03mm.

[Husky1]

I switched the motors, same error. I switched the drives, same error. Looks like mechanical error, worn leadscrew/nut i guess. I highly doubt mach3 would have a bug?

[Husky1]

So, i have replaced ballscrew with new one, reinstalled the mach3 and the error is still there. I have no more ideas what could be wrong. Could be CSMIO IP-S motion controller faulty?

[Husky1]

so i finally found the issue. One of the gantry linear rail was worn out causing z axis to swing left and right on the bottom at endmill.

[JAZZCNC]

so i finally found the issue. One of the gantry linear rail was worn out causing z axis to swing left and right on the bottom at endmill.

Well, that's a first to me, I've never seen linear rail wear out and cause such an issue. I'll put that one in the data bank to watch out for.!

[Doddy]

Well, that's a first to me

If true, could this be indicative of some other underlying problem than "wear and tear"?, misaligned rails?

[JAZZCNC]

If true, could this be indicative of some other underlying problem than "wear and tear"?, misaligned rails?

Absolutely and very probably is due exactly to that reason. Or they are really cheap nasty rails and poorly aligned.? I've never had a rail wear out to that degree and I've probably got 100+ machines out in the field using them and many of them used 12hr 6Dwk with minimal maintenance with several over 10yr old so if quality rails like Hi-win then I'd definitely be looking for some other issue causing this.

[Husky1]

i am using 15mm tbi motion linear rails on the gantry. I have changed my setup 6 months ago, moved leadscrew too high above the linear rails. I am always pushing my machine to the limits of the spindle (3000mm/min, 2mm wall, 15mm deep, 10mm endmill). I think this was the cause to wear them. I didnt have any issue before the change.

I am changing gantry rails now with 25mm, and putting leadscrew between the linear rails. Considering buying 6.6kw hsk63 spindle.

[Husky1]

Doddy: rails had 0.03mm error on 700mm

[JAZZCNC]

i am using 15mm tbi motion linear rails on the gantry. I have changed my setup 6 months ago, moved leadscrew too high above the linear rails. I am always pushing my machine to the limits of the spindle (3000mm/min, 2mm wall, 15mm deep, 10mm endmill). I think this was the cause to wear them. I didnt have any issue before the change.

I am changing gantry rails now with 25mm, and putting leadscrew between the linear rails. Considering buying 6.6kw hsk63 spindle.

TBI are decent enough linear rails and I still wouldn't expect them to wear out like this, also those cutting parameters while at the limit of the spindle shouldn't be anywhere near affecting the linear rails.
If the machine cannot handle 2.2Kw or 3Kw spindle then it certainly won't handle a 6.6Kw HSK63 spindle and just fitting larger rails may not be enough. I would look carefully at the machine to see if can find any other cause for the rails wearing out in that manner because I'm pretty sure it's not just because you have moved the ball-screw position. That large spindle will quickly find any weak areas and you could be back to square one chasing issues.

[m_c]

TBI are decent enough linear rails and I still wouldn't expect them to wear out like this, also those cutting parameters while at the limit of the spindle shouldn't be anywhere near affecting the linear rails.
If the machine cannot handle 2.2Kw or 3Kw spindle then it certainly won't handle a 6.6Kw HSK63 spindle and just fitting larger rails may not be enough. I would look carefully at the machine to see if can find any other cause for the rails wearing out in that manner because I'm pretty sure it's not just because you have moved the ball-screw position. That large spindle will quickly find any weak areas and you could be back to square one chasing issues.

If the machine is using this design - http://www.mycncuk.com/threads/10118-Finished-DIY-CNC?p=82769#post82769 I'd say it'll be putting quite a lot of leverage on the rails through cutting forces.
Looking at the diagrams, it must be verging on a 2:1 leverage between the lower trucks and the upper trucks, which I'd guess will be pushing 15mm rail to their limits.
However I'd probably look at doubling up the 15mm rails, rather upsizing given the design.

[JAZZCNC]

If the machine is using this design - http://www.mycncuk.com/threads/10118-Finished-DIY-CNC?p=82769#post82769 I'd say it'll be putting quite a lot of leverage on the rails through cutting forces.
Looking at the diagrams, it must be verging on a 2:1 leverage between the lower trucks and the upper trucks, which I'd guess will be pushing 15mm rail to their limits.
However I'd probably look at doubling up the 15mm rails, rather upsizing given the design.

If I remember right there are 3 rails on this Y-axis design with 2 bearings on each rail so I wouldn't expect them to wear out the rails in just over 3yrs. Even the bearings should last that long and they would wear long before the rails.? . . . But you could be right because the Z-axis is a long lever.?.

Matt which rail is worn out and how many does it use.? . . . Also how many hours does the machine have on it.? (mach3 will tell total operating time)

[m_c]

I had just assumed it was the Z-rails...

Y the load should be spread over a larger pivot, however it'll do far more movement than the Z, so any dirt ingress/lubrication issues could wear it out quicker.

[JAZZCNC]

I had just assumed it was the Z-rails...

Y the load should be spread over a larger pivot, however, it'll do far more movement than the Z, so any dirt ingress/lubrication issues could wear it out quicker.

No, I'm pretty sure Matt had 15mm rails on the Y-axis and 20mm on X & Z. This is why I'm surprised they have worn.
I've got a machine with 20mm Hi-win rails that I bought second-hand 15yrs ago so could be the best part of 20yrs old which have been completely utterly abused and neglected with hardly any maintenance on a machine that solely cuts Aluminium and still the rails are not excessively worn.

It takes a lot of abuse to wear these rails and even thou 15mm I'd still expect them to last 4yrs on a machine which I'm sure isn't working in a production environment.

[Husky1]

Sorry for late reply.

To explain: My first y axis setup did have 3x 15mm rails and the leadscrew in the middle of the rails. No issues. Then i redesigned the z axis so i could fit my bt30 spindle there. Two 15mm rails stayed there and i removed the third rail and in addition i relocated the leadscrew. Now the leadscrew is not in the middle of the rails but on top. This 2 modifications caused the rails to wear faster since only 2 rails left and the leadscrew on top made additional leverage on z axis. Machine works 8h/day all week, half of the time the 2.2kw spindle is on the limit. After changing top rail with the spare one, the error was gone.

I am changing the design again. I am getting 2x rexroth 25mm rails for y-axis and placing leadscrew in the middle and hope there is enough room left for the big hsk63 servospindle.