Any idea why I'm losing registration between operations. Sometimes, sometimes not.

Any idea why I’m losing registration between operations. Sometimes, sometimes not. super drag though, can’t get anything done wasting material and time. Cuts all the first operations cuts and then scoots over a few mils and ruins everything. Please help!!

What material? Could you be losing steps? Is it only in the x direction or y direction? A few mils is not much which could be steps. Have you tried less micro stepping cause you lose torque on micro steps. Is this a belt based machine or lead screws?

It is not obvious to me staring at your pics what you are calling “loosing registration” , likely because A. you know exactly what you expect and B. it is somewhat hard to tell with the surface paper still attached.
Anyway, a very typical loss of registration, by which I assume you mean loss of (0,0,0) reference, is caused by power management issues with steppers. You need to be at FW 440.14 (or higher, 440.16 available today 16July)) and have $_pm set to 2(suggested) or 1 for each axis stepper.
Do you already have these set?
If so, describe when the loss of registration seems to appear; While milling, when making a non-milling jump between pockets, when jumping between parts, etc.

Actually, now that I look at this more, I don’t think you’re losing steps. I think you’re on a belt-based machine and you are seeing the slop in timing belts. Those little circles not being real looking circles are telltale signs of timing belt mushiness.

OK, John is posting at 3:47AM PDT, guess he couldn’t sleep :frowning: . I’ll paraphrase his comments as "you may be pushing your tool faster than it can get the cutting done. "
You may be cutting too deep on each pass, cutting too fast, spindle too slow, etc. I find the easiest way to check this is to use the CP Gcode widget to run the job slower, such as at 0.5*Fspeed, real simple to do. If that improves the cut, then you can modify you Gcode (F speed, more passes) and/or modify tinyG parameters such as microsteps

If slowing down F speed helps, you might run into a separate issue: looks to me like you are milling acrylic or other resin material. Moving too slow at high spindle speed can result in burning the cut rather than chipping out the material. The best solution then might be more passes( incremental Z) at an F speed that keeps the mill chipping, rather than overheating. That is mostly experimentation as to what works best.

I’m cutting 1.5mm G10 fiberglass sheet double up with the paper comes to 3.5mm I’m cutting 4.2mm with 6 passes of .7 each at 150mm a minute. Seems pretty slow to me. My machine is a belt driven OX from Open Builds. you can see how the outer cut is shifted over a few mm. I think its my PC dropping steps or something like that.

Ah yes, now I see it. I agree, that is very conservative speed and depth. One Gcode file for all parts? Do complete parts get cut one at a time, or do all holes get cut, then go back and cut the outlines?
An interesting addition would be to bring your gantry back to 0,0 at the end of the job, and see if 0,0 matches what you set at the start.
You could do that all in air, so as not to waste material. This could possibly be accumulated math errors in tinyG as well, but these do not appear to be particularly aggressive dimensions and tinyG has done much more complex jobs.
Maybe post a link to the Gcode file?

Well I was trying to cut a field of parts at first. All the holes first then the outlines. It would make through the holes fine and even through one outine but when it went to these parts it would shift over and ruin the whole run but two parts (double stacked) Now I’m cutting one part at a time and getting the same thing, I wish I could choose the order of cutting and make it all one operation seeming how it’s between operations that it shifts over.

I am curious, what do you mean by “between operations”? You doing a tool change or something?
If it is just a long move between Outline cuts, perhaps $vm is too large?
Try setting $xvm=$yvm=800 (I consider that conservative for my SHO2).
Belt/tension issues, should they exist, are much more likely to occur on fast, long moves at $_vm, where the gantry might actually get to $_vm.
Accleration/decelleration can make belts slip a tooth.
Do you have a real heavy spindle?

FYI, in my CAD layers at Z1 are cut before Z moves to Z2.
So I can force some ordering by cutting certain portions at (incrementally) different Z heights.
Perhaps a thought - no feature standardisation in CAM world, for sure.

In Jscut I click all the holes and hit the create operation button detemine the depth for that operation and submit, then I move on to the outline and do the same thing. so it will do it in that order but it does the first operation but scoots over for the second.

Can someone with JS Cut experience jump in here please - I have none.
Something does not sound correct.
Does JSCut generate Gcode on the fly or a stand alone file?
Just how fast is “scoot”?

I don’t know, my belts are nice and tight I thought. I was thinking maybe it was skipping a tooth if it moved to rapidly to the next operation. just a thought but I doubt that’s the problem.

Back to JohnL’s earler comment on microstepping- at any given setting the NEMA17/23 has only so much slowing/holding torque. So if a really heavy spindle reaches near terminal (=$_vm)velocity and is ramping up down on a steep curve (high jerk), you may (magnetically) slip a microstep or two when ramping/deramping “too hard”. Best way to diagnose is to lower $_vm and/or $_jm until loss of accuracy is eliminated, then selectively increase. This is art with a little science.
And, this is just a guess, by the way.

I’m using the Quiet cut spindle from Open Builds and it’s not very heavy compared to some of the Bosch routers I see attached to these things. Turned out a Y belt was loose so I tightened it and I slowed the rapid to 500 and I’m on my third cut, first two were successful!!!

Good news!
I have a generic quiet cut., that is not heavy. A dewalt 661, thats heavy.

I fixed my statement above, I meant not heavy.

Steven, what belt tensioning setup do you have on your machine? The reason I’m asking is because I also run an OX and I currently just have the belts clamped down with screws through T nuts, which isn’t ideal. I was planning on doing an adjustable tension mechanism like on the X Carve but I was curious how you do it.

Brian: Here are a couple Pics of My SHO2 solution. Don’t know Ox, so may not be mechanically convenient for you

That looks like a nice simple but effective setup Carl, thanks for posting. The OX uses extrusions as well so your solution would work fine on my machine. Is that a small key ring you’re using to anchor the belt to the cap screw?