Since getting the wood CNC set up I’ve attempted to use it a couple times, and have been disappointed with the repeatability of the Y axis. I investigated with the help of @Gear105 and we found that it seems to lose steps, but not due to mechanical slippage as far as we could tell. I think it’s an issue with the electronics. The two Y motors are wired in parallel to a single driver, so perhaps it is not delivering enough current at times, either due to configuration or capability.
Today I made a test case G Code and made a video…
The test case goes back and forth between Y=100mm and Y=200mm, and dwells for a second so we can read the gauge. Here is the file (YTEST.TAP)
Can you change the acceleration/deceleration parameters to lower values and see if that helps? Also add an equal dwell on the opposite end of the travel from the gauge so that you are doing a truely equal movement in each direction, it might be trying to change direction while still physically decelerating.
What is the driver set to deliver for amps and what are the motors pulling?
To test if its the drivers we can rewire the X driver as a Y split the motors and test its repeatability. Otherwise it could be a current setting on the driver, or maybe problems with the driver itself.
I’ll be there later today and might try to do this.
I already tried adjusting the parameters and it made no change.
@Lukeo
Wiring two motors to one driver is not recommended. You are loosing a lot of power because the set draw will be split between the motors. I think there also some other strange downhill effects of doing this as well, in that when you are energizing two different coils on different motors along with the effect of micro stepping between coils, things get… muddled. I recommend splitting the step dir before the drivers.
Swapping the drivers is a good place to start. If swapping the drivers ‘fixes’ the problem then it’s likely an adjustment on the Y driver board.
Of course also clean the X and Y axis slides to make sure there are no sticky spots, check for warps etc. Try running the text in sections of say 1/4 the distance and see if there is a sticky spot that you can’t see visually.
Reduce the speed and see if that helps, if it does then likely you have a current problem, weak motors or as was pointed out it may be the motors just don’t like to play in parallel together.
@emerson told me that @lukeo added a second driver recently, so each of the Y motors has its own driver now. I re-performed the same test just now, and repeatability is very much improved.
It moves a few thou here and there, but not always in the same direction so they cancel out over time. I think that’s within the expected tolerance of this machine (heck, even leaning on the machine slightly was changing the gauge by a couple thou).
I just rewired the machine and finished installing the drivers yesterday. The motor step distance for Z and Y need to be recalibrated in mach 3. And I’ll be trying to upgrade the limit switches soon as well.
Please tell me that you’re not cutting anything while the dial indicator is in a position to get dust on / in it. Please also return it to the metal area so that it doesn’t collect any dust from other wood-dust-producing applications. (It was dusty when I returned it to its shelf)
Thanks @lukeo. I did some more tests today. And things are looking pretty good!
X and Y were already fairly close. They seem to be traveling about 98% of the commanded distance. I’m not sure if that’s within the tolerance of the system so I didn’t change them.
The Z axis was traveling 200% the commanded distance, so I reconfigured the “steps per” in Mach3 (It was 1600, and I changed it to 800). After that my test cut measured 100% commanded distance in this axis.
The Y limit switches are still not working, but I would say this machine is now pretty usable for lots of projects.
Today I calibrated the X and Y axes by doing a series of 4 holes spaced 12" apart, measuring, adjusting the “steps per”, rerunning, etc. They’re now pretty well calibrated.
But now I’m limited by the rigidity of the machine. I tried to cut through a 1.25" piece of pine lumber, but after about 0.25" the machine vibrated too much, the endmill caught a sidewall and jerked a few steps. I stopped it quickly.
To solve that, I raised the workpiece up so that the Z axis of the machine is at the top of its travel. I recommend we lower the machine back down to how it was at Cook St. There’s not enough rigidity for this machine to work well when the Z axis is extended down past 2 or 3 inches, so we might as well lower the whole machine back down to how it originally was, reducing the work envelope so that we don’t have to build up our work holding so much.
With the workpiece up high, and the Z axis almost fully retracted, I was able to cut a bit deeper than 0.5" into the pine before the machine started vibrating so much that the endmill caught an edge again.
Next time I will try go even slower (but this was already really slow).
I checked out the machine today. Seems like a lot of the rigidity issues are in the Z-axis assembly. Quite a bit of play up and down (looks like that could be tightened) the router clamp twists quite a bit along the axis of the gantry and the gantry isn’t very tortionally ridgid anyway.
Otherwise there’s just a few rollers that don’t contact the rails (maybe that can be solved by swapping around parts). We could also throw a few more and bigger triangles in the corners if we don’t mind losing some travel.
I’d be interested in partnerning up with someone who knew the machine better to get this sorted out together.
Reading materials = Foundations of Mechanical Accuracy - Moore
@thovthe thanks for checking that out. I’m just pinging @Lukeo who built the machine originally and also @Emerson who has gotten it to work in the past. I know both are really busy right now but maybe they could get you going. Sounds like you have some experience and some good ideas to improve the machine.
