as you might have notices, Der Froschkönig and I have a fight. Funnily enough the error only came to light when cutting those pretty small finger joints with dog bones. If you have the time, I’d ask you to please cut the attached file (takes 34 seconds in 7mm deep plywood), measure the fingers and take a picture. I’d be really interested whether all the other CNCs around here are completely exact. And yes, I want to a) see whether it is maybe more common or b) know for sure, that it is my printer being off.
Looked at the gcode and ran it though Camotics (gcode simulator).
It appears you are trying to cut 3.7mm or so a pass, with that I can see problems. Reduce your step-down, and feed-rate. I’d go with a lot less a path. With the feed-rate and step-down I am seeing, the piece probably will either come out of the clamps, or bog down/burn and lose steps/slip/ruin the piece. I’d try lowering the step-down until you have at least 4 passes minimum, and halve the feed-rate.
Is those the issues you are having? I can, and use use estlcam, the thing needed was the actual gcode, which you sent, all good
Nah, those are not the problems. This works fine, I am pretty sure I could even go faster. Thanks for running it through a simulator and the tips anyway.
The problems are over/undercut depending on choosing climb or conventional because the router isn’t completely stiff. I am now reprinting all the braces because my printer was off by 1mm which made the braces do stupid things. With those small fingers one can see pretty well if the CNC is dialed in well and not wobbly.
Actually it could be related, maybe.
So you are seeing, from what I gather, is that it is not tracking on multiple passes, as in not following the previous pass?
Yes, thanks so much. At least I can see that it is definitely my fault. Not that I don’t trust Ryan, but I needed to see it. Thanks again. I love your accent, btw.
It tracks 100%, I cut 9 identical pieces yesterday, and all of them were exactly the same (but wrong). When I changed from conventional to climb it was absolutely obvious that it’s not something with the code but with the stability. It’s in the Froschkönig thread. Printing new braces now, maybe I’ll need a new core as well. Dreading that… -_-
No prob, you’re welcome. What’s the problem you’re trying to fix? Are your dimensions different, and/or getting crooked/rough cuts? And/or some other problem(s)?
Yeah, am wondering…
In general, these mostly 3D printed CNC’s, and bit life, remove larger volumes of material per sec, with slower but deeper cuts.
3D printers have initial layer height and initial layer speed settings. Starting prints with more conservative settings helps with overall job quality/success.
EstlCam has Finishing parameters. Does (or should…) it have Starting parameters too? Does similar concept make sense and exist for other CAM software? I didn’t find. Ideally EstlCam would let me configure bits (or bit-material combinations even) with :
Optional Initial passes count, default 1
Optional Initial pass DOC/Step-down
Optional Initial pass Feedrate
@christian-knuell, thank you for creating, and continuing to extend EstlCam!
How’s the router bearings?
To me and my brain you have a slop issue.
I am no expert but from what you have explained in the cutting direction and with it being the first layer.
Things I would try
Facing one edge with a gap larger than the cutting tool. Both ways (idk the cutting terms like upcut etc )
One is going to grab and try and pull or kind of claw it’s way into the product.
The other won’t do this as much. So it would show a slop or play more.
And also if it’s hapening more on the first layer that could also point to the bearings at the collet being worn. As there is nothing to help and re center the bit.
It has a lot more room to drift around etc.
Do you have a fish scale or a luggage scale ?
Maybe it would show if you had the steppers powered and added a bit of force to the collet to see for movement. Your hands and fingers are really good at feeling slight movements. Like for a gap to widen if you have your finger on said gap and it even slightly moves you hands will pick up on it.
It’s like checking tie rods for slop etc.
just some random ideas anyways man.
I’ve always found those little 1/8” or 3.17mm endmill flex just a little leading to problems with accuracy. Slowing the feed or making shallower cuts helps. I prefer using the 1/8” endmill with a 1/4” shank and that does improve things. This is only useful if the cut is not deeper than the 1/8” part of the endmill of course. For bigger projects I just go to the 1/4” endmills.
I rewatched the video at least 10 times now, trying to find my misake and I noticed that the dog bones also don’t look as pronounced as they should be, they should look like this: Measurements are off, but consistently - #17 by Tokoloshe
Or am I just imagining things now? At least your measurements are precise. I am milling it over and over in my head, the only thing that comes to mind is the YZ plate still moving somehow.
/edit: I figured out that it was the Diamantverzahnt endmill, a 1-flute cut it without problems, which is weird since I’ve always used those with the Primo. I am going to run more tests, but I think I had the RPM turned down too much. The picture I linked above was cut with the same settings that I used above, except the RPM! I will start at max RPM now again and work my way down, because the diamond bits are a) pretty cheap and b) leave a nice cut if you do it right. If nothing is working then they will be replaced by 1-flutes…
Thanks for running it through a simulator and the tips anyway. 
Not that I don’t trust Ryan, but I needed to see it. 
I am milling it over and over in my head, the only thing that comes to mind is the YZ plate still moving somehow.