Mach3, 203V and steppers losing steps - looking for items to check

[LacL]

@Bell 430, thanks for the input. I have been using a surge protector for the CNC and my case is grounded to the 3rd pin on the power cord which is earth ground, but I did not have the steppers grounded to this.

@HelicopterJohn, how do I run Windows 7 in XP compatibility mode or do you mean run MACH 3 in XP compatibility mode?

[HelicopterJohn]

Hi Leo,

I meant to say run Mach3 in XP SP3 Compatability Mode in Windows 7.

John

[LacL]

Unfortunately XP mode didn't solve the problem.

I also turned off everything I could think of in windows including virus checker and firewall.

[Alexey Vorobyev]

what chipset is on your motherboard?

[LacL]

I'm using the Asus M4N68T-M with GeForce 7025 / nForce 630a chipset.

[Alexey Vorobyev]

We've experienced a couple of times the similar problem, and also on nForce chipsets. STEP freq. generated sometimes is very unstable. It's seems to be hardware problem.
Try to use VIA, or any other chipset, non-nForce.

[kreutz]

Hello;

Did you read this? : What to do about "hitching", or missing / skipping steps in Mach3 | Building a CNC in Las Colinas it might help.

Best regards,

kreutz

[kreutz]

I have the exact same setup and same issue. When I use MDI to make moves, I don't seem to loose any steps. Always returns to 0 on the dial indicator. When I run a long program, the parts come out smaller by about 0.045".

I am now suspecting Windows 7 too and found this post after a quick search.

Who else here is using Win 7? I have tried everything to resolve this issue and nothing has worked. Thinking of installing XP if it will solve this.

Two things to check: Backlash compensation adjustment and steps/inch settings. If the difference increases with the size of the part it might be due to leadscrew non-linearity, see: http://www.me.mtu.edu/~microweb/chap2/ch2-5.htm

[kreutz]

I took your problem very seriously and tried to reproduce it using my mill.

First I made a g-code file to exercise the x axis at different feed-rates from 5 ipm to 90 ipm (@ 5 inches/s/s acceleration) traveling back and forth decrementing distances (from 6" to 0.075" ) and used a dial indicator on the zero position. There is a 0.75 sec dwell time when arriving to zero so I can monitor if there are lost steps. Backlash was previously adjusted (0.00008") and backlash compensation is on. There were no lost steps at any feed-rate.

Second: I wanted to separate the hardware portion from Mach3, so the prior test proves that the mill/controller hardware is OK. Now, I started a g-code file for a 24 tooth gear cut 3" O.D. did the same test starting from 15 ipm to 80 ipm. The zero position is to one side of the gear so there is no interference with the dial indicator. The table returns to the zero position at the end. Kernel speed is 75 Khz (didn't want to test it at maximum speed).

From 15 to 30 ipm there were no lost steps. From 30 ipm up there was a 0.090" +/- 5 mils difference in end position. There is no stalling or audible difference telling that steps were lost.

Probable Culprit: Mach3 (R3.043.022) and/or my PC (HP Xw6200, 3.2 Ghz, dual xeon, 4 Gb RAM) Mach3 optimized, no extra services active. It would be nice to have a SmoothStepper board available to try to go deeper into the root of the problem. A loaner will be sincerely appreciated in order to rule out the PC as the root cause. I don't think it is Mach3 algorithms' related per se, but step rate jitter induced resonances, no noticeable otherwise, because there are a lot of acceleration/decelerations involved into the gear cutting g-code ( a 3.5" circular pocket test cutting MDF, a few days ago, at 80 ipm didn't show any problems). Mach3's PC test says that the PC is excellent ... I have stepper dampers installed and there are no evident resonances at any step-rate.

Best regards,

kreutz

[LacL]

Thank you for taking the time to look into this problem.

I just tried on 2 different mobo's with different chipsets & CPUs. Same exact problem on both.

What's strange is when I issue MDI moves I do not get any lost steps, only when cutting a test part, in this case a 3" square. It's always 0.040"-0.045" smaller. I suspect the accel and deccel at the corners is causing the problem, but why?

[kreutz]

Thank you for taking the time to look into this problem.

I just tried on 2 different mobo's with different chipsets & CPUs. Same exact problem on both.

What's strange is when I issue MDI moves I do not get any lost steps, only when cutting a test part, in this case a 3" square. It's always 0.040"-0.045" smaller. I suspect the accel and deccel at the corners is causing the problem, but why?

It is exactly my point, when I test longitudinal back and forth movements using g-code (which behaves as your MDI tests) there is no loss of position even at 90 ipm. It is only when doing composite (circular) synchonized movement with more than one axis that the problem arises and only above a certain feed-rate.

I went further today and loaded LinuxCNC and configured it from the liveCD. The linux latency test shows that my PC's step-rate is not reliable above 37 ipm, so that might be the culprit. Tested with the same g-code (cutting a gear) at maximum speed (37 ipm) and there were no lost steps. I don't need to do any machining above 30 ipm but it would be a good test to try a SmoothStepper board. Meanwhile I will continue to use Mach3 (because I am used to it) but limit my rapids to 30 ipm which is not a problem because my mill is a Sieg X1 micro-mill (with extended Y and X axes).

Best regards,

kreutz

[kreutz]

How did you set the steps/inch settings? Did you watch this video? http://www.youtube.com/watch?v=ZAQHNYSYR2I&feature=player_embedded]G0704 Axis Calibration.wmv - YouTube