So, I was cutting a large circular pocket this weekend. Depth was supposed to be 4 passes of .0937 to finish at .375, after the second pass things went screwy. This was supposed to be a 3 and some change hour cut, but I watched the first pass and most of the second. Then I decided all was good I’ll go for a walk and get some things done. When I returned it was a quarter of the way through pass 3 and was at a depth of .2485, completely ruining the piece.
I have checked the gcode and the z depths are correct, and machine belts are tight, so that leads me to believe it was carbide motion. Does this sound like something that has happened to anyone else? Using CM 4 and latest GRBL.
I never had toolpath errors due to Carbide Motion myself, and never saw anyone report any such issue on the forum. It seems unlikely that Carbide Motion would corrupt the G-code values it sends, while still sending sufficiently valid commands for the job to continue.
Post the G-code maybe for a double-check ? 3hours for a 0.375 pocket seems very very long, maybe share the endmill and feeds and speeds too in case this helps finding hints at what could have happened?
Also, I cannot agree more with @WillAdams about never leaving the machine unattended. I stumbled upon this video soon after I bought my Shapeoko, and it was eye-opening for me.
I’ll post the G-code when I get home tonight. Just off memory though its a .25 2-flute upcut endmill, with I believe 60IPM cutting a 30" diameter cirlcular pocket and “3” on the dewalt (~17000 RPM?). This was my first time cutting ash, so I was trying to be conservative with my cuts. Judging by how well it handled an erroneous ~.25" DOC, I plan to adjust my working DOC to something greater than .0937.
Feed, speed, and depth of cut is a balancing act — you want conservative enough to be reliable and not cause missed steps, you want fast and aggressive enough to be done in a reasonable period of time. The testing technique at: https://www.precisebits.com/tutorials/calibrating_feeds_n_speeds.htm seems to work well.
Any chance your router or bit moved causing the deeper cut? I have had a bit slip in the collett and cut too deep, but never had an issue with Carbide Motion sending the wrong values.
Blockquote
Any chance your router or bit moved causing the deeper cut? I have had a bit slip in the collett and cut too deep, but never had an issue with Carbide Motion sending the wrong values.
Im using the precise bits collet and didn’t notice any slippage in the bit or router, but will double check.
Blockquote
Good point. The other notable possibility is the Z-axis belt slipping due to insufficient belt tension or a loose pulley set screw.
I know for sure the Z belt is tight, but havent checked the setscrews.
Blockquote
Feed, speed, and depth of cut is a balancing act — you want conservative enough to be reliable and not cause missed steps, you want fast and aggressive enough to be done in a reasonable period of time. The testing technique at: Calibrating Feeds and Speeds When Using Carbide Microtools seems to work well.
Quickly looking at their recommendations they suggest an initial plunge of 1 x D for a wood like Ash, which in my case is .25". This seems aggressive considering Carbide3Ds feed and speed chart doesn’t it? If I’m mathing good the recommended feedrate is 85 IPM, which seems feasible at a lower DOC. I understand these are very broad general guidelines, I’m just trying to make sure I’m not babying the machine too much.
this G-code file is fine, with four passes of the same depth, as expected: so the problem was most probably something mechanical, along the lines of what William and Dan mentionned.
Side note: I have seen the “1xD” DOC recommandation, but experimented and always end up using much less than this value, at least in hard wood. What works for me is closer to 1/3rd to 1/2 of D…at least when slotting is involved, and when using a “naive” toolpath generator like Carbide Create’s. I suspect people using Fusion360 can pull off deeper DOCs using smarter/adaptative toolpaths.
Do you feel like a 1xD DOC is achievable with a lower stepover? Currently I think I use around 75%, but if I drop that to something like 35% perhaps. Furthermore, would the lower stepover negate the decreased cut time benefit of a larger DOC?
Probably so, but indeed for carving such a wide pocket it may not be beneficial in the end. You could try and generate multiple versions of the G-code using different combinations of DOC/stepover, and simulate the running time (in case you are using Carbide Create that does not include runtime estimation, you can use any online g-code viewer, my usual choice is this one)
What I would do to reduce the total job time is go for a (even) higher feed rate, say 75ipm, bump up the RPM accordingly to keep the same chip load, and use a small enough DOC that works ok with this high feed rate.
Also, there are probably CNC veterans here that would have a better opinion than mine on this. There is surprisingly little info/guidelines about optimizing DOC, compared to the abundance of tables for feeds and speeds, this always seemed a bit weird to me since DOC is just as important. I always choose the easy way out = experimenting. For a given material+RPM+feedrate, I start with a conservative DOC, then do a few test runs while increasing DOC, then stop and dial it back a bit when the sound of the cut tells me it is too much.
Well when I had a standard Z on my main shapeoko I kept loosing steps or jamming up - I initially thought it was due to me milling too much aluminium or going too hard etc. However since building my new machine I’m getting into this mad world of electronics which is teaching me allot about motor power and requirements. If your motor speed is too fast or acceleration is set too high, or you have too many micro steps on your machine you might find it skipping steps due to the controller not being able to power the motor sufficiently. This will result in the cuts getting lighter and lighter.
The simple fix is to fine tune your machine. Drop your micro steps and acceleration and it will make a world of difference.