Losing steps?? How do I know??

[JeLC]

I've had my system up and running about two weeks and have done about 6 or 7 trial cuts, with terrible results. I'm using a Gecko 540 with Keling 370 oz motors on a 34" (X axis) X 25" (Z Axis) table and using a Dewalt 618 router. My computer is a 2.4 Ghz with 512 Mb Ram running Mach3.
Yesterday I was cutting a small trial piece (14" X 12") doing an area clear around some lettering. I was doing it in two passes of .075" (Z) per pass. I had to rezero my Z axis multiple times because it was not maintaining a constant cutting depth. I've checked for a problem between my motor and my lead screw and am confident that I am getting no slippage because of a loose set screw. But what is confusing is that I'm only getting it on the Z axis. My X and Y seem to be consistent.

I'm using the Fine Line Automation antibacklash nut, so I should have very little backlash.

Does this sound like lost steps and if so, how can I remedy it???

Also, my motors are running extremely hot. So hot I can hardly touch them. I bought a package from Keling that included the G540, 370 oz motors and a 50V PSU. And I'm running at about 47 volts. Is this heat normal or do I have another problem?

[ger21]

Cut your Z accel and velocity in half and see if that helps. If it does, it's probably lost steps. Is the Z always getting lower? Or does it change?

As for the motor, do you have the current set correctly?

[Crevice Reamer]

Hi JeLC.

Have you calibrated the steps per inch on your axes?

Have you actually MEASURED your backlash?

You need to use a 3.5K resistor for the Keling 387. The G540 will WORK without a current set resistor, but will then be set to the maximum 3.5A. Also, without the resistor, the idle current does NOT reduce. Without that, the motors will DEFINITELY get hotter than they should.

Resistors of 3.5K are hard to find. Here's a source for a 3.48K:

You will need one of THESE 1/4 Watt 3.48K resistors for each 387:

3.48KXBK-ND

They only come in a pack of five. With shipping it's about $2.50

http://digikey.com/

CR.

[JeLC]

The z is going high, meaning when I had to reset zero, the DRO is saying the bit is at zero, but the bit is actually above the wood. (I'm guessing about a 1/16 or so, I didn't actually measured it.

Yes, I'm using current limiting resistors. (could not find a 3.5K here locally, so I bought a 3.3K and a 200 ohm (both 1%) and wired them in series. I have a cheap volt ohm meter, and they measured out at around 3.6K, which is a little high for the 1% tolerance, but that should result in lower current, if I remember my basic electricity correctly) And CR, you're right. I put in the original post I had 370 oz motors, but they are Keling 387 oz.

I have calibrated all of the axis several times.

One thing I forgot to mention was that I originally started out running Linux and EMC2 on this system. With Linux/EMC2, there is a latency test for checking out how fast the processor responds to the control program when it needs to generate a pulse. The system I'm using measured somewhat high, but, according to EMC2 documentation, within specs to run a machine. Mach3's printer port test did not show any problems with the speed of the PC. And I'm guessing that the Mach3 printer port test is the equivalent of the Linux latency test. (and yes, I do know the latency problem didn't just go away by installing XP and Mach3. Matter of fact, since I've gone back to a Microsoft product, it is probably gotten worse).

And no, I haven't measured backlash. How do I do that???? I would like to measure it on all of the axis, but don't have a clue how.

[BobF]

Here is what I would do.
setup a piece of scrap and cut a box outline that is exactly 1 inch. No need to go very deep. Measure it with calipers on all 4 sides and see what it is. This will give you an idea of several things. The bit should travel around the box from one end.
1. steps per inch are correct or not.
2. backlash - are you getting different measurements on opposite sides.
Check it for square with a known good square.
Are the corners square? This may show some misalignment in your machine.
All good?
Make a plunge cut to a specific depth, say 0.250. Measure with caliper. Is it correct? If so then Z steps per inch are correct.
Use Mach to move the machine all around, then go back to where you started (you did set all this to 0,0,0 didn't you?) Did it return to where you started? If not why?
Use a sharpie or something to mark the motor shaft, coupler, and screw. This is probably something to do before you start all this. Anyway when all is done, see if all the marks still line up. If not, something is slipping.
If all this works and you still have problems, I would bet on loosing steps.

[pminmo]

One thing I forgot to mention was that I originally started out running Linux and EMC2 on this system. With Linux/EMC2, there is a latency test for checking out how fast the processor responds to the control program when it needs to generate a pulse. The system I'm using measured somewhat high, but, according to EMC2 documentation, within specs to run a machine. Mach3's printer port test did not show any problems with the speed of the PC. And I'm guessing that the Mach3 printer port test is the equivalent of the Linux latency test. (and yes, I do know the latency problem didn't just go away by installing XP and Mach3. Matter of fact, since I've gone back to a Microsoft product, it is probably gotten worse).

Pulse generation problems would normally show up in multiple axis, not just one. But if the emc latency test showed marginal, odd's are that MACH3 won't do as well.

Any chance the coupler set screw is slipping?

[JeLC]

Pulse generation problems would normally show up in multiple axis, not just one. But if the emc latency test showed marginal, odd's are that MACH3 won't do as well.

Any chance the coupler set screw is slipping?

Not showing up on all axis is what is confusing. But I did do some searching around on the Yahoo Mach3 group site, and losing steps on the Z axis (only) is not unusual. But what is unusual is my Z is ending up above the beginning zero, and most on that site said their Z axis was ending up below the beginning zero.

And I don't think the coupler set screw is slipping. I'm going to scribe some reference marks on both shafts and the coupler, jog the axis around a bit and see. I don't believe it is, but I won't say for sure until I test it out more.

I did a cut like BobF suggested a couple of post before this one. I made a simple perimeter cut, .15" deep on a 2" square and a 6" square in MDF. Set my depth to cut at .1", so it would have to make two passes on each square and move the Z axis on both. After zeroing and doing the cut, the first square ended up at about .135" deep, but the second square ended up about .04" deep. That's over .1" error on a simple cut. I made the cut several times, lowering my Vel/Acc from 50, then to 40, then to 25, and finally to around 10 IPM. Pretty much same results each time. X and Y measured good on each cut, as far as being square and the right dimensions. Just a problem with the Z, and, if I had enough hair left to pull out, I'd be pulling about now.


Edit: I've decided to go back to Linux. I'm running the trial version of Mach3 anyway, and the problem is worse with the Microsoft setup. I'm only running 512Mb Ram, so I'm going to double that and reload the Linux/EMC2 setup. Plus in Linux, there is a way to modify the SMI (system maintenance interupts) portion of the BIOS. In reading, this is dangerous and could burn up the processor, but the computer I'm using is a spare, so I'm willing to take that chance. WinNT and XP are not designed for real time operations, and Microsoft is not my favorite software anyway. And it just looks like the motherboard/CPU/Microsoft combo I'm using is not suited for running a CNC machine. It's a high end (or it was when I bought it) DFI LanParty MB, designed for online gaming and video editing. If it dies, it dies. I'll just tell it rest in peace, throw it in my pile of discarded and junk PC parts, and forget about it.

[BobF]

What motors and drivers are you using?

[pminmo]

Can you give use more info about your Z axis, screw pitch, approximate weight, motor, driver..... What is the acceleration value for your Z axis? Screw pitch, etc.

[JeLC]

I posted my problem on the Mach3 yahoo group and got a solution from someone named Brian, who I think is associated with ArtSoft and Mach3. He told me to try adding 5us to the pulse width (bottom right corner of the motor tuning dialog box). Looks like that solved my problem.

I recut the perimeters of a 2" and 6" square to a depth of 1.5", which before the change, resulted in an error of approximately .1 above 0. After the change, the depth was practically perfect. It was cut in MDF, so getting an exact measurement is impossible, but it's really close.

Now I can start slowly getting my Vel/Acc back up to more than a crawl.

[BobF]

Glad you solved it.

[jasonjetski]

Hi, I have a similar problem today where my MF70 seams to lose some steps on the Y axis.
Iam cutting very small helicopter parts with a 1mm bit and each time I go back to X/0 Y/0 The cutter is 0.5mm out of position.
So I looked at my motor tuning box and I have 5us already in each of the box's.
Do I need to try adding 5us to each box,or just one box.
Also,Does it mean I need to make the setting 10us on every axis?
Regards Jason

I posted my problem on the Mach3 yahoo group and got a solution from someone named Brian, who I think is associated with ArtSoft and Mach3. He told me to try adding 5us to the pulse width (bottom right corner of the motor tuning dialog box). Looks like that solved my problem.

I recut the perimeters of a 2" and 6" square to a depth of 1.5", which before the change, resulted in an error of approximately .1 above 0. After the change, the depth was practically perfect. It was cut in MDF, so getting an exact measurement is impossible, but it's really close.

Now I can start slowly getting my Vel/Acc back up to more than a crawl.