I have a problem that has been plaguing me for two years, and across software and machines. 95% of what I do can be done without a Vee bit. I simply use small-diameter end mills and get good output. However, occasionally I want to use a Vee bit, and I have never been able to do it with accuracy.
I started out with an X-carve and Easel. I had the problem with that, and battled it with support and help from the forums with no luck. I eventually replaced my X-carve with a custom build using mostly OpenBuilds parts. I use Carbide Create and CNCJS with the new machine.
The problem is that across both machines and software, I get inaccurate inside corners during the finishing toolpath.
Tramming - I trammed both machines more than once and also leveled the wasteboard. Given that the problem shows in all 4 corners rather than just to one side (or top/bottom) I think it is unlikely that tramming is the issue anyhow.
Vee bit angles - I have used 5+ different Vee bits ranging in angle from 30 to 90 degrees, and also verified the angles. All show the same issue.
Z-axis play - Again, this is on two different machines, but I have confirmed the the Z axis is tight.
Software - I’ve tried Easel and Carbide Create, both with the same issue.
DPP - Tried reducing it drastically to the bare minimum. The end result is the same.
Calibration - I’ve done X, Y and Z calibration multiple times on both machines.
The example I provided is a simplified example, but this happens on all files I carve with a Vee bit. If I use a 1/16 end mill for the finish pass, all comes out great.
I should also add that this problem happens at the very end of the G Code execution. It goes around to the inside corners and tries to sharpen them, but is always too far outside the corner. The depth is correct, but the placement is not. Oddly, it is outside the corner by about the same amount all the way around, not just in one or two directions.
I’m down to the only reasonable issue being deflection, but I have used three different spindles at this point - two different Dewault routers, and now a dedicated spindle. Could all three have had the same deflection issues?
I’m just wondering if anyone has additional advice on this. I’m two years into CNC work and have been successful with everything except this one issue that continues to make me crazy!
There was once a thread where we collectively obsessed about this effect, which you may find interesting if this has been bugging you for 2 years
Bottom line: assuming everything else is 100% correct (e.g. programmed vbit angle is EXACTLY the angle of the actual vbit, don’t trust the manufacturer that it is any given value with high precision), this is probably down to tool deflection when cutting the straight sides (the corners are where they should be), and usually just running the vcarve a second time will address this (just because there is very little material left to (re)cut, so the tool/router basically sees a very small load)
The OTHER usual suspect, that you have not mentioned, is incorrect Z zero setting. And this is particularly true if using a vbit that has a small flat on the tip (they all do, to some extent). The Z zero should be raised by a value corresponding to that flat tip width and the vbit angle (we have a calculator somewhere for this…) such that the virtual point of the vbit is a stock surface, if that makes sense.
@Julien Thanks, and that was an interesting read. I tried a few things mentioned and didn’t have a lot of luck. Running it a second time had marginal improvement. I’m interested in the Z zero comment. I am using a probe, so I thought that would account for the issue. There is no visible flat tip, but I can do more experimenting there.
Here’s how I can work around it… Z Probe and run rough pass with an end mill. Z probe the vbit and then raise the Z zero slightly and run the detail pass completely, then lower the Z zero back to the original value and run the vbit again but manually STOP it before it gets to the part where it sharpens the corners. This effectively takes enough material off in the second vbit pass to compensate for the overcut from the first pass. It’s a pain to watch it and stop it at just the right time, but it does work. Maybe this is just what I have to do?
Interesting. This sounds like a combination of the “run it twice” approach and “raise the Z zero slightly to compensate for the flat tip”, except you don’t let it run through. Have you tried experimenting with the vbit precise angle ? i.e. run the exact same project a few times, each time declaring a slightly different vbit angle in the CAD. For a 60° vbit you could generate the toolpath based on the same square share but with vbit declared to be 59.8°, 59.9°, 60°, 60.1° and 60.2°, and check if one of the cuts is better.
Some of the inexpensive Vee bits may not be what they’re advertised at, the bit maybe listed as 30 degrees, but actually be 27. The other is as what Julian pointed out with the flat tip. When I’ve run into those odd corners, I’ll either mess around with the Z height, or angle of the bit.
