Machsupport Forum

Mach Discussion => General Mach Discussion => Topic started by: Montezuma213 on November 19, 2012, 11:15:11 AM

Title: Losing Steps on x
Post by: Montezuma213 on November 19, 2012, 11:15:11 AM
Hello, I was wondering if someone could give me a few ideas to try.  I am out of ideas for fixing this problem. Here's some background on my machine. I run a gecko540 using a parallel port. I had issues with my first machine because it didnt have a parallel port, so I had to add a card with a parallel port. The problem was that that machine only uses PCIe. All the cards that are PCIe dont work well with my geck540. So I decided to upgrade to an older machine with a parallel port.

So here is my problem. WIth the new machine, my machine loses steps to the left somewhere in the middle of cutting. I take that same gcode, and run it on my old machine (after I plug it back in) and it doesnt lose steps. So its not the gcode or the physical CNC. There has to be something wrong with the other machine. I have no idea where to start. I am using the same mach file from one machine to the other, so the settings are all exactly the same.

Thanks
Title: Re: Losing Steps on x
Post by: Tweakie.CNC on November 19, 2012, 12:29:33 PM
If your new (old) PC has onboard graphics that could cause a problem - fitting a graphics card could be the way to go.

Tweakie.
Title: Re: Losing Steps on x
Post by: Fastest1 on November 19, 2012, 12:37:54 PM
Tweakie, I too have experienced this or a similar issue in 1 of my set ups. This is the only variable I didnt accept. I figured if it would run mach, good enough. But there is something subtly different. In my case reducing speeds did help but I dont find it an acceptable solution. I think I still have an extra card around here somewhere.
Title: Re: Losing Steps on x
Post by: Montezuma213 on November 19, 2012, 12:46:36 PM
I have a card at home I can try. Lets hope this works.  Any idea why this causes a problem?
Title: Re: Losing Steps on x
Post by: Fastest1 on November 19, 2012, 01:20:24 PM
   I replaced 3 of my PC's after a failure of an old power spec I had been using to run a mill via mach. That power spec had the max of ram and an extra graphics card. Ran flawlessly too. The replacements I used are Dell Optiplex GX280's. They do run Mach however I have had issues that never existed before, like the loss of steps at seemingly reasonable speeds. Oddly enough if IIRC, it was always in the X axis too. I liked the replacements for their slimmer form factor which I hope doesnt turn out to be a PIA. No real investment as I do use older computers.

At this moment I am powering up the old power spec to find out what had failed, then repair it and run a test. It was an intermittent issue I think. The more I thnk about it, I had just implemented 1 of these computers on a different machine and was loosing steps there but abandoned it for a while to play with the G0704. If that is in fact the culprit it is best anyway. The threads I read on machine set up and tuning only made my understanding and operation of the machines that much better. There has to be easier ways to learn it though.   
Title: Re: Losing Steps on x
Post by: Tweakie.CNC on November 20, 2012, 02:12:44 AM
I have a card at home I can try. Lets hope this works.  Any idea why this causes a problem?

When using an LPT Parallel Port, Mach3 takes control of the Win OS in order to output it's pulse train (step and direction signals) and if the PC has other refresh / update priorities there can be a conflict. Any interruption in this pulse train can lead to lost steps so fitting a graphics card can reduce the loading on the OS.

The definitive explanation is probably a lot more complicated than this but it is more or less the gist of it.  :)

Tweakie.
Title: Re: Losing Steps on x
Post by: Montezuma213 on November 20, 2012, 08:13:33 AM
Thanks tweakie, that helps. Do you guys think adding a wireless network card could cause issues also? I forgot that I added that also to the machine.
Title: Re: Losing Steps on x
Post by: Tweakie.CNC on November 20, 2012, 10:46:53 AM
Thanks tweakie, that helps. Do you guys think adding a wireless network card could cause issues also? I forgot that I added that also to the machine.

In my opinion yes, but like all things, I think you just have to try removing it and see if the problem goes away.

Tweakie.
Title: Losing Steps on x
Post by: Fastest1 on November 26, 2012, 09:32:50 AM
Well over the last few days I have installed a new CPU fan and hard drive. Fresh install of XP in the new drive along with Mach. Followed all optimization tips. Running a separate graphics card as required. I still get this weird loss of position in between the finish pass of holes 6 & 7 of 8. The rapids should be the same between all holes and no errors occur til line 2700 or so.
Title: Re: Losing Steps on x
Post by: Picengraver on November 26, 2012, 10:20:53 AM
What version of Mach are you using?  If not the lockdown version, try that.
John Champlain
www.picengrave.com
Title: Losing Steps on x
Post by: Fastest1 on November 26, 2012, 11:07:50 AM
John it is the latest lockdown as of 11/24/2012. It all works great except this issue. I am not sure if it is a Mach or Cad/Cam problem. I just know it is A problem. Tweakie just suggested it as a possibility. I have to rule out all variables. It is just odd as my machine seems to operate smoothly and controllably in all axis otherwise. I see nothing in the code that looks suspicious. The line where the loss of position occurs is no different. This particular hole pattern is 2 rows of 4 holes. Parallel and symmetrically spaced. Seems like the other transitions would be affected too but they are not.
Title: Re: Losing Steps on x
Post by: Picengraver on November 26, 2012, 12:37:30 PM
The reason I asked is that ever since I started using the later development versions to be able to use M10/M11 codes with my laser diode machine, I have experienced similar issues infrequently.  For no apparent reason, Mach will suddenly lose steps on a normal gcode line (ex. - G1 X0 Y0) that has worked flawlessly earlier in the same file.  Shutting down and restarting Mach sometimes helps, sometimes not. It is certainly not my controller or settings as I tried all the steps already recommended to you, and they do not help.  The problem seems random and unpredictable.  My opinion is that Mach is suddenly issuing step signals that ignore the feed rate as well as max. velocity settings.  I know it is not my code, as it only uses G1 commands for moves, but I don't know if it might be related to the use of the M10/M11 commands.

I am using a Dell Optiplex GX270 with pport- perhaps that's the commonality?

Hopefully someone else has seen similar behavior and will post about it.

Regards,
John Champlain
www.picengrave.com
Title: Losing Steps on x
Post by: Fastest1 on November 26, 2012, 01:45:13 PM
Is yours doing the same error in the same location? Mine is consistent. The optiplex was being used. I brought the power spec back from the dead to test as it has all of the minimum requirements Mach needs. The optiplex didn't have a separate graphics card. It didn't perform any different either.
Title: Re: Losing Steps on x
Post by: BR549 on November 26, 2012, 01:53:09 PM
Try using the latest developement Ver as a test.   OR back up in versions to see if it changes.  Some DELLS are known to be incompatible with MACH3.

There used to be the same type bug with the Z axis. While drilling it would ignore the Velocity/acell  parameters and TRIP the Z axis servo drive.

(;-) TP
Title: Re: Losing Steps on x
Post by: Rotax91 on November 26, 2012, 02:47:45 PM
I just came across this post because I too am having the same exact problem.  I'm losing steps on my X axis.  This wasn't always happening.  My speed is not jacked way up either. Problem just started last night.  I noticed it when I was vcarving a job and it was doing a multipass vcarve circle.  The start and endpoints don't line up.  The vcarve lettering looks horrible as well.  I thought it might be a loose motor coupling but I ruled that out.  Problem only seems to be with the X axis for me. It took a 3rd reboot and I was able to get a good run.

I'm also using the Gecko G540 controller.  My computer is an older GATEWAY pc.  I think I purchased it around 2005.  It has an onboard factory oem parallel port. I'm already using an aftermarket video card.  If I recall, it's a Nvidia Geforce FX-5500.  I'm running Mach 3 3.042.020     I'm going to try upgrading it to the latest Mach version tonight and also pull out the wireless NIC card in it.  The PC is dedicated to just my cnc.  It's a fresh install with Windows XP Pro SP3. No windows updates have been performed on the machine.  It's not connected to the internet and hasn't been in quite some time.  The wireless NIC just happens to be in it from the old days.

I will post back sometime tonight if the above helped. 
Title: Re: Losing Steps on x
Post by: Fastest1 on November 26, 2012, 06:01:11 PM
The power spec is doing the exact same thing as the Dell. The Power Spec is running an AMD Semperon 3100+ 1.8Ghz 1.25G Ram HT1600 motherboard, 500g Sata HD, 8400GS 256MB graphics card, XP Pro SP3 with the latest updates. Very lite SW wise. XP Pro and Mach 3. Also a dedicated PC for machine use.
Title: Re: Losing Steps on x
Post by: Hood on November 26, 2012, 06:23:09 PM
Not read through this post but a few things come to mind. Have you tried increasing the pulse width? Have you tried changing the Active state for the Step pin?
Assuming you are using the PP for the pulse width.
Hood
Title: Re: Losing Steps on x
Post by: Fastest1 on November 26, 2012, 09:16:37 PM
Hood, I have tried lengthening the pulse width with no change. I have not changed the state of the step pin however why would it only fault at 1 time after 2700 lines of similar code? I am using a PP.
Title: Re: Losing Steps on x
Post by: Overloaded on November 26, 2012, 09:35:20 PM
Hey John,
 Have you ruled out everything mechanical ?
A tight spot in the X at the problem area maybe ?
Can you do manual fast multiple axis jogs everywhere with no problems ?
Power supply ?

Didn't see these mentioned.
Russ
Title: Losing Steps on x
Post by: Fastest1 on November 27, 2012, 09:02:53 AM
Russ any input is good. It is a new PS 48v/12A, PMDX-126, 5056D's. During calibration I was running speeds up to 331ipm with no measurable loss of position (with my .0005 indicator). I ran many warm up wizards at that speed though during rapids the max speed is divided by how many axis are in motion. Still no loss. As the holes have 2 rows parallel I would have suspected an error in the transition from 2-3 also had the carriage or gibs been too tight. I did spend considerable time trying to adjust all gibs with an indicator. All of my fasteners are tight. I do have the typical backlash of a linearmotion2008 ballscrew, somewhere in the .002-.003 per axis. Backlash comp deals with it spectacularly.
Title: Re: Losing Steps on x
Post by: Picengraver on November 27, 2012, 02:33:07 PM
My mechanicals are linear bearings and Thomson shafts - very free moving.

I've recently changed to a different laser diode driver and don't use M10/M11 commands now, but too early to say my issues are gone.  If it happens again, I will revert to an older version (.022) to see if the issue continues.  Since others in this thread don't use the M10/M11 commands they don't seem likely to be the problem.  The only commonality we have seems to be the latest lockdown version (.066).  The last I remember, Ver. 066 was development and .022(?) was the lockdown.  I missed the change.
Regards All,
John Champlain
www.picengrave.com
Title: Re: Losing Steps on x
Post by: Rotax91 on November 27, 2012, 03:33:46 PM
I posted yesterday on this thread due to my X axis missing steps badly as well. This didn't always happen. Problem seemed to start on Sunday.  My multi pass circles weren't starting and finishing properly and my vcarve lettering looked terrible.  I pulled out my wifi card since I didn't need it.  I upgraded to the lockdown (.066) last night.  My X axis velocity was set to 100.2 in the motor tuning.  I backed it down to 80.  On the home screen I always had my feed rate slider down about 1/2 from the normal spot so I never thought much of the 100 value.  With only those three changes, I got (2) 1 hour long runs last night without a single problem.