Wierd CNC Issue

[EmptyNester]

Hey All.

I haven't had any time for working with the CNC for quite some time. I'm finally getting back to it for a winter activity.

The PP50 is working great on my table.

So far, everything I've done has been for fun. Today I started working on a "real" project. At work we're having a chassis built for a product we're developing. It's going to be a month or two before we get it and we'd like to have one now to figure out where cables go. A perfect chance for subsidized fun.

I converted the Solidworks files to DFX'es and used Sheetcam and Mach.

I did a couple small parts and they went well. I just did a part that was 8" x 17". It is 3/4" short. All the dimensions to about 14" up on the part are correct.

When I traced the route with Mach, it actually did the appropriate height.

Any insights would be appreciated. (I thought I'd check here before I tried the CNC board.)

Thanks

[Rambozo]

If the small stuff is bang on, you could be losing steps due to something jamming the travel slightly or interference from the plasma.

[acourtjester]

To be sure I understand you did a part that was 8” X 14” and the 8” side was correct and the 14” was short by ¾”. If that is what happened did you have the part aligned so one side was in the X direction and the other in the Y. If that is the case one of the axis calculations is off. You could rotate the part in sheetcam and try again, it would be easier to just re-cal the axis.

[EmptyNester]

Jester - it didn't cut right, but if I job through the cut, it shows the right length. So, I figured if it could do it right when it wasn't cutting then the axis calculations are right.

Ram - I'm thinking that you may be right. I hadn't used the THC in a while and I think I had swapped in a newer one without much testing. I'm wondering if the torch lost steps because the tip was dragging?

I think I've found a new issue. When the arc crosses the slat I see a big voltage jump. I can't believe that I haven't seen this before. Makes me wonder if this new board is much more sensitive than the old ones.

[acourtjester]

One thing you may check is when you went through SheetCam to load the drawing did you have the selection in drawing options in inch not custom. This will change the scale for the part.
If the tip was dragging the floating head switch should have tripped.

keep us posted that is bugger.

[EmptyNester]

It turns out that it was the tip dragging.

It had been so long since I ran the table, I didn't recall where the THC software was at (I'm always tinkering). I finally figured out that I had screwed up the kerf detection so that controlling on voltage never kicked in. Once I fixed that, I was able to cut without a problem and dimensions came out as expected.

Since this is the first "real" thing I did, I had dimensions that I had to meet. There were a number of tab/slot features that were off just enough to not fit without filing/grinding. So, I guess I have to figure out how to "tune" the CNC. I don't know if this could be a backlash issue, and if so, if I can do anything about it.

The other thing is that I was cutting with water in the table. I hadn't been in the past because I never cut enough at one time to justify it. I saw a lot of voltage fluctuations. I'm not sure if it was due to the water, or something else.

[Rambozo]

As long as your kerf is consistent, you can always tune things to get better accuracy. Have you measured your actual kerf and told SheetCAM what it is?
Backlash compensation can do a lot, but it's best to remove as much as possible mechanically, before tweaking the last bit in software. With a lightly loaded device like a plasma cutter, there are a lot of tricks you can use that wouldn't work on a router. Things like spring loading your drive system to remove lash.