[wastingtape]

I've had my Taig for a few weeks now and have been getting better at cutting test models in wood. I decided today to try something "real" which was the faceplate for the CNC controller case I have. I bought a generic project box and planned to just cut it to meet my needs.

I had been running my feedrates at around 26-30 IPM with wood, but decided to drop it down to 20 IPM on my first go with metal. The tool was an 1/8" square end mill and I had the speed set to pullet #2 at 6500 RPM. I'm using MeshCAM for my tool path generation which seems to generate paths such that the geometry for the entire model is cut at a single depth (z-axis move) progressively in stages based on the step down rate.

The first feature I attempted to cut was the holes for the fan mount and vents for air to be pulled in. The GCode started in the bottom left corner, then moved to the bottom right, then top left, and finally top right, before coming back around to bottom left to do the next depth cut. CNC Zone said the picture was too big so I put it up here: 2013-05-12_0812 - andstuff's library

Somewhere in the top right corner the y-axis seems to have gotten all messed up, such that when it got down to cut the next depth of screw holes it was completely off.

I'm a bit puzzled by this as the rest of the design seems more or less correct.

How do I troubleshoot this? Is this backlash or "loosing steps" (still not 100% what that means)?

[Bob Adams]

I can't help with your problem, but to get your slots cut right try this Gcode from my Write2Cad program.
The slots are for 80mm Fan and the cutting tool to use is .125 Dia.
Write2Cad> Panel
Cheers
Bob

[109jb]

Looks to me like it is losing steps in the y direction when it moves from slot to slot on the right side. There are several things that can cause lost steps. Tight gibs, acel too high, feed rate too fast.... from the look of your picture, it looks like the error likely us happening during a rapid y move. I would lower acel, and or max feed rate and try that.

[wastingtape]

I gave that a shot. I went from 20 IPM down to 12.5 IPM. And for the y-axis tuning in Mach3, I went from 8.75 acceleration down to 3.5, and from 29 velocity to 15. I wanted to change all the numbers enough to see if that was it.

The problem still continues, perhaps slightly worse than it was before.

I'm starting to think it's something too tight (which would explain why moving slower made it worse). I just recently adjusted the split nut on the y-axis following the instructions from the taig youtube video (this one: X Y GIB ADJUSTMENT - YouTube). I'm wondering if I've got it tuned too tight.

Also, I haven't adjusted the gib settings from the factory. When I had the machine taken apart a few weeks ago the y-axis did feel a little harder to turn, so I guess I'll check out that too.

[wastingtape]

Alright, I took the y-axis off the machine, adjusted the gibs, and loosened the split nut a bit. It feels a bit smoother. I loaded up the stock and re-ran the program and got the same issue.

I stuck a dial indicator on the machine pointed at the y-axis and started doing some tests in MDI mode in mach3.

When I move only the y-axis, there is no perceivable backlash.
G0Y1
G0Y0
G0Y1
G0Y0
(no backlash)

When I move both the x and y axis at the same rate, there is no perceivable backlash.
G0X1Y1
G0X0Y0
G0X1Y1
G0X0Y0
(no backlash)

HOWEVER when I move x and y at different rates, particularly when y is moving slower than x, not only does the y-axis sound like a dot matrix printer, but the actual location is 0.001" off from what mach3 thinks the location is. I did this test several times with 0.001" being the best and 0.007" being the worst when also doing motions with the z-axis.
G0X-2Y0.5Z1
G0X0Y0Z0
G0X-2Y0.5Z1
G0X0Y0Z0

This would probably explain why I'm just now seeing this issue, as most of the test cuts I've been doing have been square and not round.

So, where do I go from there?

From what I know, I'm guessing this is a problem with either...
- The power supply not being able to put out enough (Currently running Mean Well AS-320-48: 48v 6.5A)
- OR, the computer can't put out the signal consistently enough when "talking" to all motors at once (Currently running Del Optiplex 755, Core 2 duo, 2GB ram, Windows XP, G540 on parallel port).

Does anyone have any ideas on this? I'd like any second opinions before dropping $50 on a new power supply or $150+ on something like the smoothstepper.

[awerby]

It's hard to say anything about the adequacy (or lack therof) of your power supply without knowing about your motors. The voltage seems okay, but how many amps does each of them require? If the PS doesn't put out enough amperage, your motors might malfunction when asked to work simultaneously. (By the way, this loss of position is called "losing steps" not "loosing" them - an alarming number of people can't spell.)

The bad sounds made by your Y-axis motor are significant. That could be due to a number of things, like incorrect acceleration values, wrong pulse width, current starvation, etc. There could also be an issue with your computer - have you gone through it to eliminate anything that might be interfering with its concentration on producing pulses? Stuff like Windows update, Apple software, power management, and extra startup items can wreak havoc with CNC operations. There's something on the Machsupport site about optimizing XP; check it out. Don't neglect mechanical issues either, like lack of lubrication (you are oiling this machine each time you use it, right?) gibs adjustment, incorrect depth of cut, couplings slipping, and the workpiece shifting under cutting forces. If you've been cutting wood, you should also make sure the dust and oil on the screws and slides haven't combined to make a doughy gunk that can build up and interfere with their smooth functioning.

Andrew Werby
www.computersculpture.com

[wastingtape]

I'm running the soigeneris 166oz steppers, which according to their website are 2.8A. Stepper Motors - Soigeneris.com I picked up the Mean Well power supply as that's also what soigeneris recommended. MeanWell SP-320 Power Supply - Soigeneris.com

The "bad sounds" have been puzzling for a while and I had just kinda wondered if that was "normal". There are actually two distinct sounds which to me seem off -- first is that when moving very slowly, say 1-2 IPM, it sounds like a dot matrix printer. Second, when moving a little faster, say 5 - 10 IPM, it sounds like a bad 80's synthesizer. Seriously, with all 3 together I can easily get a techno orchestra which I've never heard that in all the Taig cnc mill youtube videos I've watched.

Regarding the other things you asked about -- I did a fresh install of Windows XP just for the machine, then went through and removed everything that wasn't needed (even including the games and such) before installing Mach3. The machine is not on the network and Mach3 was the only program I'm running. I've gotten rid of just about everything I could (including disabling services like DHCP), but I'm sure there's always more to disable.

When I took the machine apart earlier today I also cleaned all the ways and re-lubricated them. I'm using Vactra Way Oil #2 which I think was actually a suggestion I got from you. I re-adjusted the y-gibs and the y-axis split screw. And I know the gunk you're talking about. I've cleaned it off the machine few times now. It's pretty gross.

I'm not sure how to check for coupling slipping.

After I did the mechanical stuff I spent about 3 hours in the MDI interface with my dial indicator tuning the steps for each motor. To test this I pointed the dial indicator against the axis and reset the indicator to 0. I then zeroed that axis in Mach3 and ran the commands G0X0.1 and looked at how close to was to actually being 0.1". I then adjusted the steps configuration in Mach3 accordingly until all motors were actually moving 0.1" consistently. Each axis required a different value ranging from 40200 to 40800 which I was a bit puzzled about. I'm not sure if this is a discrepancy in the wiring for each motor, or something related to the G540, or what.

Next I repeated my tests above with the 3-axis G0 code...
G0X-2Y0.5Z1
G0X0Y0Z0
G0X-2Y0.5Z1
G0X0Y0Z0

And I was getting correct re-positioning!

I re-ran the program from before to cut the fan grill. It wasn't an exact match, however the Y-axis was only off by about 1/2 as much as last time. I'm guessing I resolved part of it and now there's something else I need to resolve.

I should also mention, my motors are *ridiculously* hot. And when I say hot, I mean nearing hot enough to burn if I were to touch them.

[awerby]

To check for slippage in your couplings, put a mark on them with a Sharpie that continues onto the shaft they're attached to. Any movement between the shaft and the coupling will result in an offset line. If you got this system from Jeff at Soigeneris, ask him what's going on; he's pretty good at tech support. Stepper motors can run a bit hot; that's not a big deal. But if they're burning up, that could indicate a problem. Did you install the right current-limiting resistors for your motors?

I don't think that adjusting your steps/unit settings to what you're actually measuring is the right way to go about solving your problems. This should be a fixed ratio, based on your screws (20 tpi), motor steps/rev (200) and microstepping factor (10). If you want to adjust for backlash, there's a separate setting in Mach for that. The wiring shouldn't affect it. You need to fix your machine so that 40,000 steps moves each axis one inch (not counting backlash). Otherwise, you're never going to be able to make anything accurately.

Andrew Werby
www.computersculpture.com

[handlewanker]

Hi, just my 2 pennoth worth, how much power are you using to hold the stepper in the holding position, IE when it's not moving.

If you are holding position with huge amounts of amperage the steppers will just get hotter and hotter.

I assume you have checked the tightness and adjustment of the gibs according to the previous video in post #4.
Ian.

[Jeff-Birt]

Are you using the parallel port? It sounds like you might be getting jittery step pulses. Shut down Mach3 and run the driver test program (in the C:\mach3 directory)and see if you get a nice smooth line or something that looks like an EKG. The other thing to watch out for is over-tightening the lock nuts on each axis. If you do that you will crack the ball bearings race and the axis will be 'lumpy'. For motor tuning values, you want to set the steps/unit to 40,000, the accel to 4 and the velocity to 10. This should run fine at these settings. If not then I would suggest trying to borrow an oscilloscope front a friend and taking a look at the step wave forms coming out of the parallel port (and also running the driver test program mentioned above.) If your getting good step pulses from the PC you should be able to set the velocity up to 30-40 range and the accel to around 6-8. I always suggest starting out with the values at a safe low setting until you get the hang of the machine. (I build the majority of my control boxes with the SmoothStepper as it helps me avoid all of the unknowns with the many thousands of different motherboards and types of parallel ports out there.)

On the stepper motors, did you install the current limiting resistors in the DB9 connectors? Keep in mind stepper motors are rated to run up to 180 degreed or so. They will always run a bit hot. On the 166oz-in motors you can reduce this somewhat by using the pot on the front of the SP-320 and setting the power supply to about 42V. There won't be a performance decrease but they motors should run a bit cooler when they are at rest.

I might add that I have drive boxes with the G540 and SP-320 running everything from 4'x8' router tables to Taigs/Sherlines. Anything the G540 can handle the power supply can handle

[Jeff-Birt]

@wastingtape - did you get this sorted out? If your still having problems please feel free to contact me.

[wastingtape]

@Jeff-Birt

Yes. The thing purrs like a kitten now. The motors don't overheat. And the power supply you recommended is completely capable. It turned out two be two problems:

Problem #1: Hardware

This problem I actually ended up discovering by accident. It turns out when I mounted the 166oz motors to the NEMA23 motor mounts, I tightened them to the mounts too much. It wasn't enough that I could see any problem or distortion, but I decided to test the motors off the machine but left the mounts on. I noticed that one motor in particular was "grindy" and gritty as it spun, and got extremely hot fast. I took the motor mount off to get a better look at the shaft to make sure what when I put the coupler on I didn't accidentally put it too close to the motor casing. It was then that I suddenly realized how different the motor sounded without the motor mount on it. My best guess is that in overtighenting the motor mount it caused the motor casing to distort, introducing friction where it shouldn't have been. This would explain the insane heat the motor was getting and probably contributing to positioning problems.

Problem #2: Software

I gave up on trying to tune Windows and converted from Mach3 + Parallel port to a KFLOP + KmotionCNC + USB. The difference is night and day. With the motors now properly set in the mounts everything works extremely well. The KFLOP system is difficult to learn but totally worth it. The motors are mildly warm compared to burn-your-finger hot as before. And my rapids are 2 or 3 times faster.