I am not sure if this is a Carvey problem or an Easel problem, but I think it’s Easel…
I began noticing this a couple of months back with under sized diameter holes, and I have been fixing them by manual filing after carving. I have been assuming the problem was worn bits that were slightly under-size. This happened I believe when a new algorithm was introduced in the order of carving.
The project I am doing now involves making precision grooves, and the errors are not so easy to fix. I have specified a groove of 8.2mm wide, 172mm long with a depth of 0.6mm. I am using a 1/16" “blue” tool. The material is two color acrylic. I am using custom cut settings of Feed rate: 300mm/min Plunge rate: 300mm/min Depth per pass: 0.3mm.
The resulting groove turned out to be 7.8mm wide or 0.4mm too narrow. I have measured my 1/16" bit, and it is under-size, measuring .0610" rather than .0625". I believe this can account for only a part of the error, 0.03mm of it by my calculations. Most of the error is from something else. I was thinking that perhaps the error is caused by the shape of the tool’s end, perhaps it is not quite the same diameter at the tip. But looking at in under a microscope suggests this is not that case, besides the “walls” of the groove are nice and vertical.
Interesting the depth of the grooves were too deep, measuring 1.0mm rather than 0.6mm.
Am I asking too much in the way of accuracy from this machine?
Kenneth, pardon my ignorance. Us Easel/Carvey types just do not get into that area, nor should we based on the way Carvey is promoted, IMHO. The language used between Easel and Carvey is kind of a “black Box” I am pretty much afraid to open…
I have been operating this machine for a couple of years now, without this issue. Hole sizes in the past were accurate. This is something new, that I believe started happening a few month back when Inventables changed their algorithm for driving the machine.
I dont think (but can’t say for sure) that the algorithms affect this at all. GRBL-speak = the machine just do what its told.
But any machine will show wear and wear will undercut pockets / overcut outside shapes. Backlash is the key word here. Backlash is the distance lost when it change direction. X/-X etc
At this level where you detect 0.2-0.4mm discrepancy we need to make sure the following is true before going deeper:
1 - Use a V bit as suggested above, test actual travel vs commanded travel.
2 - make sure the machine is tight, collett is tight and clean the machine if neccesary
3 - Cut a single line path and measure actual width. Tools do also wear.
I tried making a single line cut as you suggest, I used an engraving bit which has the sharpest point of any of the bit I have. I was not able to accurately measure the resulting line, so the results were not definitive. I also cut to rectangles, one with a 1/4" straight bit, the other with a 1/6" up-cutting bit. This rectangles were specified to be 8.2mm by 20mm.
The 1/4" cut one came out as 7.98mm by 20mm
The 1/16" cut came out as 8.01mm by 20.1mm
The Y axis accuracy looks good, but the X axis is undercutting by about 0.2mm
I am not able to measure 250mm to the required accuracy. The best I can think of to do is make a couple of parallel lines in the Y direction 150mm apart. And measure the distance between them. Would that be useful? Note the lines would have a bit of thickness (width) to them, reducing the measurement accuracy.
Anything to see if the error compounds over a greater distance would help. If it does, that would point to a calibration issue / solution.
Also, for the precision you need, the bits need to be measured. The method @ChrisRice suggested is a good one.
In my projects there are a number of parallel groves, they are spaced 20mm apart. Measurements taken between different numbers of the grooves, shows that the error does not grow with greater distance in the X axis.
The Carve have a 300x200 (X/Y) work plane, if you could jog 280mm both ways for X,180mm for Y and report back your measured travel that would be a great help.
Also go into Easel machine inspector, type and send $$ in the command window and report back your $110-111values that would also be a big help.
You report a discrepancy of 0.2-0.4mm which isn’t that much, so we need to establish that the base parameters are good.
No point in chasing 0.5mm if the calibration is 0.3mm off
Another test, to see the amount of backlash try this:
Jog all the way to the right, jog once to the left 50mm and measure the travel. Then perform a couple more 50mm jogs to the left before doing the same in the opposite direction.
You might see 49.2mm for the first 50mm jog, then 50 for the next until you change direction.
49.2 - 50.0 - 50.0 - 50.0 on direction, then
49.0 - 50.0 -50.0 - 50.0 in the opposite direction. The travel last on the first move after a direction change is your backlash.