Ongoing Cutting Dimension Problem D:

Last week I posted about this problem I was having with aluminum. Basically my OD’s were small and my ID’s were big and the slot radii were slightly out of round. I was going to do some more testing on aluminum but I ran out of actual ideas for things I need/want to make for now so I decided I would retry carbon fiber with my mister to keep the dust down (which works btw!) A few months back I designed my own frame for a drone, and was going to have somebody else cut it; didn’t end up working out. Fast forward to me building an MPCNC, It’s been close to a month since I’ve flown my drones and I wanted to cut/build/fly my design finally. When I designed it I 3D printed it out before finalizing it (so my drawing is very slightly different now then what it used to be, hence the small dip on the inside end of the cf part). The 3D prints were purposely cut rough but are still very close.

I cut an aluminum part that looked very close to my numbers with trochoidal cutting, but with these CF parts that’s not an option due to the size of the slots, holes, ID’s, etc. I’m currently using one of the diamond cutter endmills, 3/32" in diameter. But it was the same deal with different 1/8" endmills on aluminum also. Also obviously tried different speeds and DoCs, with no avail. The cf pictured is the most recent one I cut, and that was a full DoC (4mm) at 200mm/m and it wasn’t struggling at all, not sound wise at least. Belts are nice and snug, random but thought I’d include it. Verified tool size in Estlcam and double checked it with calipers also.

One thing I thought was maybe the acceleration speed I set up with Estlcam was too fast. It was set to reach max speed after 2mm of distance. WAY upped it to hit full speed at 32mm. Didn’t help. I thought that maybe the inertia was set incorrectly at the default 85% but then reconsidered that thought when the OD is cutting smoothly (while still undersized lol).

My last thought is that something is going wrong in Estlcam’s gcode generation. If anyone knows how to read it and find any errors, I can provide a .dxf of this part and the gcode file for checking out. I would’ve tried camming it in Fusion, but I need to learn more about the cam side of it, I’ve been putting it off…

You are using estlcam firmware? You aren’t really giving any info other than your part is bad in some way. You really should not waste expensive material until you run a test cut and are confident in your skills. HD foam is $5 and will last a really long time. Aluminum and Carbon Fiber are not good beginner materials.

Switch back to my firmware so more people can help. I have zero experience with that firmware.

As for the rest of it, I’m guessing your OD are small and your ID are big, acting like the bit is too large.
1-What cam strategy , speeds, rpms, etc., you can zip and upload your e10 file so we can check it out.
2-Firmware, board, drivers, driver voltage?
3-What spindle?
4-What bit, specifically?
5- Most importantly how big is your machine in all axis working volume?
6- Do you have a vacuum attached?
7- A build pic would help as well.

That D: face really got my attention. Now it is time for the standard troubleshooting sequence where we find out of it is electrical, software, or mechanical failure of some degree.

Have you done any sort of calibration? As in, does it go the correct distance when not under any load at all? Jog and measure. I read your other post, so it sounds like at one point you were in correct tolerance. What did you change to get it so out of whack?

Like Ryan and Kevin are getting at more info is needed. You didnt note this time how far off your OD and ID dimensions are. Since your OD is small and ID is large makes me wonder if you are telling it to cut directly on your line and you maybe didnt take into account the curf, material lost during the cut.

Also like was stated by Ryan, go back to something cheap and simple. Get some foam, or even strap on a ball point pen and put some paper down. Make some squares and circles of specific known sizes and run the program. Then measure to see what you get. Confirm what you are putting in is what you are getting out.

Oh man, google chrome just literally shut down on me for no reason right as I was almost done with my reply…Let’s do this again, hopefully more quickly…

Ok so how would I use your Marlin firmware as far as jogging the machine goes (besides LCD & knob), repetier host? I know that I can use basically any gcode sender but jogging the machine, zeroing it, etc can be a huge pain with some of them. That’s why I’ve been using Estlcam’s firmware configured for grbl. Super simple jogging (I mean who can resist using an xbox controller to jog with or without crawl speeds, zero it out, etc.), going from cam to machine in one program, etc. I’ve been meaning to switch to fusion for cam. I was supposed to be learning it at work but I believe that’s been put on the back burner for now.

I’ll just go through your numbers listed to make it easy and concise hopefully…

1.Normal machining strategies in estlcam, hole, hole, part, etc. The bit is just to big to do any trochoidal strategies (which I had luck with on aluminum) on anything but the OD and maybe the larger hole on the inside. I’ll make a sub list of the speeds I’ve tried. All at (allegedly) 30k rpm. Keep in mind it BARELY sounded like it was putting in any work except for the first cut listed below

  1. 4.5mm DoC, 300mm/m XY, 100mm/m Z
  2. 4.5mm DoC 200mm/m XY, 100mm/m Z
  3. 2.25mm DoC 300mm/m XY 100mm/m Z
  4. 2.25mm DoC 200mm (or maybe it was 150mm/m, can't remember exactly) XY 75mm/m Z
2. Estlcam firmware, cnc shield board (basically just a grbl shield clone, it may have a very slightly different pinout, can't remember 100%). DVR8825 drivers @ 700mv. I ran the first two cuts at 500mv but thought I'd try upping it just in case. No difference obviously. Wired in series.
  1. DW660

  2. This isn’t the EXACT bit, but it’s the same thing. I didn’t buy it, I know somebody who has cut carbon fiber on their Tormach with it so they gave it to me to try out so I don’t wear out normal carbide bits so fast. Keep in mind this happened with 1/8" bits on aluminum too.

  3. My machine has about an 18"x18"x3" workspace. The last time that I cut aluminum I had this same dimension problem too. However, I was cutting flat bar in a vice on parallels. I was basically as far up to the middle of the machine as I could get and was given the same outcome.

  4. Not attached but basically the whole time it runs I’ve sat there with one in my hands. Getting out aluminum chips, making sure the water mist doesn’t get everywhere, etc.

  5. The picture I put isn’t new, it’s while it was still clean :P. I’ve done some wire cleanup since then but it’s pretty much the same as in the picture.

I’ll post an e10 file and a dxf of the part in the zip file. I’ll even throw in the gcode too. I’m not sure how an e10 file will transfer to somebody else computer with another tool list. I’ll set up the e10 file to match what it said in the 1st set of feeds and DoCs listed above for reference.



Kevin, the only thing that’s changed since the nice trochoidal cut was me taking off the bottom cover of the DW660 like you recommended. The machine was still giving me trouble before I took it off so I know that’s not it haha. (208 KB)

I don’t actually have pins small enough to check the slots or holes. The major problem with them is that they are either out of round or aren’t straight where they should be. The larger hole is actually pretty close on the sides that aren’t out of round. I believe the flat spot (at least on this one hole) is where the bit plunges.

The OD is small by like .050" on both axes. This would be less of a concern to me if the motor mount end had the design looking even somewhat close to the 3D print. I know it won’t be perfect because the bit radius is ever so slightly larger than the radius on the drawing. That’s alright though, I can get it really close with a finish pass as long as the rest of it comes out correct.

You guys are right. I should be using less expensive material. These aren’t all trash parts though. In fact, the aluminum clamps that didn’t cut correctly I can still use, and used to hold down the CF for this project. These arms I cut don’t look perfect, but the hole pattern seems to be correct so they should all still be usable on my quad when I build it.

I guess the plus side of machining foam is that it’s the closest thing to snow I’ll be getting here in California :P.

I think that is the problem, It is not as simple as you are making it out.

1-Rapids? Kinda slow but I can’t say for sure. You are not using a finishing pass on any of this, that is a must.

2-Estlcam firmware and a uno based board, can’t help much here.

3-Nice, is it perpendicular? Far off, non perpendicular would cause the exact same thing.

4-Could it be bent giving you a wider cut than normal? That would explain everything.

The reason we say foam is there are no stresses put on the machine you can verify dimensions.

I believe estlcam’s rapids are just the max feedrate, which is set to 3000mm/m. I don’t remember off the top of my head where that number came from. Perhaps somebody else’s settings I blindly copied?

I need to print the perpendicularity thing you made to check for sure. But using a square it seems pretty dang close. Close enough that I don’t think it could throw it off by almost 1/16".

Also you’re right, I purposely left out a finishing pass because I knew that my OD was already small. In hindsight, I could’ve just added one for the inside until I can figure out what’s going on to throw off the rest of it. I assume that would help clean up some of the flat spots. Although, I’m not sure if it’s visible in one of the first pictures I posted, you can see two of the slots on the opposite side aren’t even in line. It’s kind of like they are more parallel to each other but farther apart when the sides should be parallel but the slots themselves in line with each other. Hopefully that makes sense.

If you have 3000 for the Z axis that can crash your board, 50mm/s, the z can only do 8.4mm/s.

1/16" doesn’t take much. The finishing pass won’t make your cuts wider it makes the dimension tighter. You leave .5mm or so and it comes back under extremely light load and cleans it up. Under high loads the bit wanders more and actually cuts too big.

Use the stuff we just talked about including the finishing pass and cut some foam, see if the dimensions come out wrong. If you don’t have foam use a pen and draw it, if you can’t get it accurate with foam or a pen there are larger issues.

I’m done for the night, good luck.

Thanks for all the help, don’t have any foam on hand so I decided to just draw it out and see how it looked. The 3rd one down is the best I got. Most of the problems in the other ones above is just from the pen being too far down so it flexes too hard. I did my best to get as much flex out as possible on the sides so everything that’s supposed to be round comes out round. Buuut nothings perfect. Anyway what this told me is that it should be cutting correctly. The dimensions are much closer to the nominal numbers, and that’s based off of a guess that a ballpoint pen diameter is roughly .5mm. Which it’s probably not.

The carbon arms that I cut all fit inside the outline of the drawn ones. I’m assuming my next step is to hopefully find some foam at lowes or home depot to cut and hope that it comes out correctly. If it doesn’t my guess is that the DW660 mount is worn or something and isn’t holding it square enough. I guess I may as well print the perpendicularity tester for that while I’m at it…

I’m not sure what you did there, did you run your cut gcode or did you do an engrave. How close to nominal numbers is it, dead on correct?

At this point you need to get some foam or some some soft wood (pine) and try your gcodes out on it first.
Here is my version of What I would do, I didn’t mess with your speeds but I do think they are too slow, try it as is though and only adjust the speeds on some scrap pieces and a small square part with a hole in it.

Alright I took a mental break from messing with this thing yesterday, and I reprinted my tool mount for my DW660 since one piece was cracked. I thought that the foam I ordered would be getting here sooner. From what I could find at Lowe’s and Home Depot it seemed like they mostly carried 4’x8’ foam sheets and I didn’t want something that big for now. I guess I could cut it down but eh. Anyway I decided to try cutting a design with similar features to my quadcopter arm. It seemed to come out correct! At most a square may have been .5mm different on one side to another. Machine might not be perfectly square, but I am easily happy with that. That’s a tighter tolerance than 90% of the parts I have to deal with at work lol. The common dimension seems to be small on OD’s by .25mm. I need to cut an ID big enough to measure to be sure how far off that is, but if it’s anything close to .25mm big/small I’ll be happy.

I’d like to note that before I attempted cutting the wood, I played around with the amount of flex possible from the gantry, tool, etc. I noticed that the tool mount wasn’t tightened all the way and if I flexed the router, it seemed to move also. So I tightened that up…

I’m nervous to try carbon fiber out again lol.

Are those dimensions with a finishing pass? They are good but not great.

Yeah finishing pass just set for .1mm. Haven’t messed with it too much apart from these few little tests.

You might want to go a little more than that. I do .4mm usually in wood.

Cut another with a .4 finish pass and added a couple other things in I had trouble with on the carbon (angled slots). The ID profile cut was near perfect. .05mm off at max. The tabs on the sides I think were something like 7.98mm and 8.00mm with a nominal value of 8! The OD of the square still seemed to be slightly small by ~.2-.25mm. The slots look pretty good, they’re a little too small to measure well with calipers, but I’d imagine they should be pretty good since the other ID is. The hole might be VERY slightly out of round but I couldn’t add a finish pass .4mm because the hole size is too small. I was also cutting pretty fast so may be better slower. It’s still not terrible though. Also couldn’t .4mm finish pass the slots because again, size. I think I may be able to get away with a finish pass on the smaller bits on carbon because I have a smaller bit than I used for this. (1/8" vs 3/32")

You’re doing it man!!! Hooray, turn that D: upside down.

Now I just need to transfer these tight of dimensions over to carbon fiber lol. And I guess aluminum while I’m at it :stuck_out_tongue:

Welp. Back to square one. After successfully cutting wood I tried a small test part in some 2mm carbon fiber to see if my tightening of the tool mount helped in actuality. Nope.

As it was cutting I could still physically see the router flexing. I think it must be some sort of play/flex in the gantry itself. It’s hard to tell, however, because it obviously can roll around while I try to flex it so it may provide a false sense of there being more play than there is.

Been using the same 3/32 diamond endmill for carbon fiber. 2mm DoC, 500mm/m. It never really makes it up to that speed though because of the acceleration. So it really only got that fast on the OD cut.

Sounds like the carbon fiber already killed your bit, you probably need a new one. Carbon fiber is a crazy abrasive and eat bits for lunch.