Hello Guest it is November 08, 2024, 09:43:57 AM

Author Topic: "Safe Z off" + "go to zero" = lost steps  (Read 5814 times)

0 Members and 1 Guest are viewing this topic.

"Safe Z off" + "go to zero" = lost steps
« on: November 12, 2013, 04:28:48 AM »
Hi!

I've struggeled with this problem for few days. I use Mach3 to drive stepper motor in my stress testing device, just one axis. I've disabled splindle outputs, other axis and also "safe z" feature as I don't need them.

I'm running at 100KHz, motor is tuned to 750 RPM (1500mm/min with 2mm screw) with acceleration "300" (up to 500 works nice) driver is providing 1/8 step division (800 steps/mm, 1600 steps/rotation) It is able to keep that speed with axis loaded with 150N. The problem was that every time I've pressed "go to zero" I recieved a message like "safe Z turned off ignoring...." and I instantly lost steps even with acceleration 100 and speed reduced to 500rpm (1000mm/min). Driver and Mach kept going but motor just made some noise.

Then I've enabled Safe Z again (which is useless as I have no Z axis) and problems dissapeared.

What happened? Does Mach3 lose steps on error messages?

Offline Hood

*
  •  25,835 25,835
  • Carnoustie, Scotland
Re: "Safe Z off" + "go to zero" = lost steps
« Reply #1 on: November 12, 2013, 01:14:49 PM »
Never heard of that before, what version of Mach are you using?
If you attach your xml I will see if I can find a problem.
Hood

Offline BR549

*
  •  6,965 6,965
Re: "Safe Z off" + "go to zero" = lost steps
« Reply #2 on: November 12, 2013, 01:19:13 PM »
It can IF you are running a kernal speed FASTER than the PC can keep up. Lower it  and retry. I would say that a lot of PCs cannot run at 100HZ in mach3 and then push a LOT of steps per out the LPT.

Just a thought. (;-) TP

Offline Hood

*
  •  25,835 25,835
  • Carnoustie, Scotland
Re: "Safe Z off" + "go to zero" = lost steps
« Reply #3 on: November 12, 2013, 01:25:27 PM »
I never even noticed that, why are you running 100KHz? Your motor tuning suggests you only need 20KHz, so why go higher than 25KHz?
Hood
Re: "Safe Z off" + "go to zero" = lost steps
« Reply #4 on: November 12, 2013, 02:15:51 PM »
I have Intel C2D 2x 3,2GHz with 4GB RAM running on Foxconn P35A-S (the only MB i remember at time when I bought my PC to have both LPT and COM), so I assume speed is not an issue. I remember running it on P4 3GHz (which is much slower processor) at that speed with no issues.

What I understand is that Mach3 is running at constant speed of xx kHz and it can only give (or not give) a STEP signal on each cycle, so at higher speeds when it comes out that STEP signal has to be made eg. every 3,5 cycle it ends up running either too slow or too fast. At 25 kHz instead of 1500 mm/s it goes about 1680 mm/s, while at 100kHz I get 1531 units/min. As I need to set speed rather precisely (I also measure how much sample deforms at given force and I calculate position from time and speed) going with higher kernel speeds gives me more accurate results. Right?

Offline Hood

*
  •  25,835 25,835
  • Carnoustie, Scotland
Re: "Safe Z off" + "go to zero" = lost steps
« Reply #5 on: November 12, 2013, 02:21:02 PM »
Out of curiosity what is your time in int?
Hood
Re: "Safe Z off" + "go to zero" = lost steps
« Reply #6 on: November 12, 2013, 02:31:58 PM »
Oscillates about 2 (1,9 - 2,2) sometimes goes up to 2,5. Ah, my processor has Speedstep enabled and changes multiplier all the time so it can affect it I guess.  Is it good? Bad?

Yes SpeedStep clearly affects Time in int: At about 2,3GHz I get about 2,5, at 3,2GHz about 2,0)
« Last Edit: November 12, 2013, 02:35:45 PM by mibars »

Offline Hood

*
  •  25,835 25,835
  • Carnoustie, Scotland
Re: "Safe Z off" + "go to zero" = lost steps
« Reply #7 on: November 12, 2013, 02:47:06 PM »
Not a good idea to have any sort of CPU power saving, thats why Laptops can be problematic.
Time in int is decent and shouldnt cause you a problem as you have roughly 75 to 80% left for Windows to do its thing.
Have you run the driver test at 100KHz to see if it is a nice smooth line?
Hood