losing steps, maybe

Hello all,

Its been a while since i was here but I've got another problem or maybe the same one i've always had and just now noticing it. I had a problem before of losing position and it wound up just being a cheap timing pulley issue. Then i had the same problem but a cam program update took care of it.

So here i am again, i lose distance on my x axis. I have verified that there is no backlash or very little. I've marked all shafts and pulleys to check for slip. I'm not using a generated g code. I'm just making small movements with incremental moves or g0 x.1 for example. My Y axis is set up exactly the same with the exception of motor size and reduction ratio. I've set up a dial indicator on y and a .1 move equals a .1 movement. On x though, ill call for .1 movement and get .08 and repeat at same direction and get .08 and repeat and get .1. It's not consistent. Doesn't matter is if it is a same direction move or reverse direction. I made a chart and about 2/3 of the time its about .02 off. Doesn't matter if i move in increments or with g code, same result.

Things i've done
marked all shafts or points of possible slip with no problem found
everything is very tight.
ive check all my numbers over and over. My x is setup just like my y and my y axis is as close to perfect as i think this machine can ever be.
I've used different driver types in setup to see if it changes it.
I've used the step time and space and hold that is recommended as well as slower and faster.
I've increased and decreased the max jitter.
I've made the movements slow to fast and everything within with the same results for all
I adjusted the scale for x to compinsate for the .2 or so that it loses, but then about 1/3 of the time it over shoots.

What really confuses me is that the problem really show up on small circles. Like .5 inch cut with a .125 bit. So it would seem that the x has backlash and it shows at the point where the motor has to reverse direction. But if you cut a zig zag line that x constantly has to reverse direction, I don't see the flaw.

I've been working with it for a while now and i'm sure i'm forgetting some of the stuff i've tried but any suggestions that i can get would be greatly appreciated.

table details
5x5
plasma/ router
X axis=kl-8070d keling digital drive driving 600 oz with 14 to 78 timing reduction pulling a 3/8 timing belt
y axis=kl-8056d keling digital drive driving 600 oz with a 11 to 78 timing reduction pulling 3/8 timing belt
z axis=kl-4042d keling digital drive driving 276 oz with a 8 to 1 planetary box driving acme thread with pulley counter balance
pmdx-126 breakout card
hp computer with linuxcnc of course

Hello,

In resume:
You say that in Y all is fine, but in X not.
X is moving something less than supposed to, and sometimes the correct amount.
You have a similar set up in X as in Y.

X is your largest axis, more heavy, fixed in two points on your table, maybe driven only in one side. For me, it looks like your mechanics get stuck in a few places sometimes, as the force needed to move the axis is larger than your set up can handle.

My recommendation:
Take X the motor out, and mark the shaft. Try to do it precisely.
Now send an increment movement that will coincide with 10 exact turns of the motor (test a value that will do that), in a speed that you can visually check it.
Do it again, again and again.
If it always turn the same amount (for example 10 turns) then up to the motor all is fine and it is mechanics, if not the motor/drive ....

Hi

I would concur, it is very unlikely to be lost steps.

The most likely causes would be slippage or flex and we don't know enough about your machine to guess which.
If the error occurs upon change of direction with a belt drive, it is probably 'taking up the slack' or there is flexing occuring as the machine moves in the opposite direction, which accounts for the lack of linear motion

regards

Thanks for you replies guys.

I will remove the motor and try that.

As for the backlash issue, it will happen with same direction moves. Its not just between direction changes.

As for it being strong enough, that has always been one of my concerns as well. But it has proven to be a lot stronger than i thought. The only time the issue really shows up is in small holes. Just testing i have run the router through .5 inch plywood with a 3/8 bit until the 2hp router started to struggle and then rehomed and hadn't lost any steps. I would never try cutting that fast, just trying to push the machine to find its limits.

My X and Y have alway seemed strong. Like accidently hitting things and bending them, oops. My z is slow and not very strong though. My plans are to upgrade my x and y to bigger nema 34 motors and then move one of my stronger x or y motors down to the z position. So everything gets upgraded with only two motor purchases. I just want to figure this issue out before i move on. And it would be nice to use it for cuttiong out some parts i need for the upgrade.

you say your using timing belt what make ? also check the pullys are made for the same belt ,
also were the pullys pre drilled to shaft size or did you have to drill them , i'm wondering if one of the pully is rotating elliptical

either way sounds mechanical
check if moving say 1" at a time forward , all ways gives the same repeatable results or is their a pattern to the error

on a different point what power supply are you using ?

My belts and pulleys came from sdp-si and seem to mesh well. I didn't have to do any resizing.

I agree that it sounds mechanical except it isn't repeatable. I've tried testing with a few different travel distances. Ill make a chart though to help see repeats in patterns. Maybe i'm just missing the pattern. Maybe it is an ellipitical pulley or something. I didn't think about that.

My powersupply is homemade. The power supply transformer is a multi tap step down transformer turned around backwards. It's huge and way over kill but better too big than too little. The caps are new from a local electronics supply. The diodes are from a wrockwell vfd. The chokes are some i had saved from i don't remember what lol.

Take the motor out and test like I suggested.
It is not a too hard test to perform and you will not damage nothing. Why guess if you can test?