Repeating

[pavlic28]

             Hi!
Does any of you guys who have been already installed DSPMC controller problem whit repeating of G Code.
I Can make a perefect circle, but I can not repeat it. It is also interesting that I can't put the axis in to the 0.000 position. they always go to the same  position as you can see on the picture bellow.
Thanks

[TOTALLYRC]

I have made several dozen of the same part in a row and they all repeat within my ability to measure.

If you go to the MDI page and do
G00 x10 y10 and then
G00 x0 y0 you should return to 0,0 plus or minus the number of encoder counts you have set up as the deadband.
I am around 4 counts IIRC which is a few tenths.

I am in imperial units.

Mike

[are1987]

pavlic28,

I just sent you a email!

Andrew

[pavlic28]

I have made several dozen of the same part in a row and they all repeat within my ability to measure.

If you go to the MDI page and do
G00 x10 y10 and then
G00 x0 y0 you should return to 0,0 plus or minus the number of encoder counts you have set up as the deadband.
I am around 4 counts IIRC which is a few tenths.

I am in imperial units.

Mike

I did as you suggest with
 indicator. My conclusiongs are folloving:
-I can repeat this porheaps 20 times and ewery time is going to the same position, but then at once the aix does not return to the same position. On the MACH is position OK, but on the indicator I can clearlly see a difference. It's a small difference  like 5 microns but it's adding together.

-I allso found out that the bigger speed I have, more often this mistakes are.

I was assuming that could be problem in the frequincy of the encoders, Today I reduce pulses per rotation (i have emulated encoder so, I can do this) but the probllem still stay.

Does any one have any suggestion?

[TOTALLYRC]

I could use a little more information.
How many counts per rev are your encoders?
What is your dead band setting?
What is your steps per inch?

After thinking about this your encoder emulation might be the problem.

The DSPMC figures out where it is by counting encoder pulses, nothing more, nothing less.

Then it feeds that information back to Mach3 to display in the DROs.

Mike

[pavlic28]

-On default I had 1000 pulses/rotation. First reduce them to 500 pulses/rotation and then 250 pulses/rotation. It  didn't help.
-Dead band settings are on 0
-I have 800 steps per inch

Today I will try to check if encoder emulation if is it OK as you suggested, but I doubt...

[TOTALLYRC]

HI Pavlic,
I think first of all you are going the wrong way with you encoder counts
You really want more counts, not less. With the DSPMC set up for 20k steps per inch, if I some how loose a step, it is .00005 of an inch. It takes a lot of lost steps to make a difference. At 800 steps per inch as you have, a lost step is ~.0015. a huge difference and a big difference in the smallest move that can be made.

You also say that you have 800 steps per inch. At what encoder setting? When you change the encoder counts/rev you have to reset the steps per inch.

Mike

[pavlic28]

I thought that DSPMC controller is not capable count so many pulses in higher speed, that's why I reduce pulses.
I did one more test. I test repeating (with indicator) just in DSPMC AxisWorks softwear. Here I made the same conclusions like in Mach, at higher speeds it sometime loose position (just a few microns). 
I have max. emulation 1024 pulses/rotation.
In the previous post I made a mistake I didn't wrote correctly. I have 800 steps per mm.
So, I'm allmost in the same ratio as you are.
 Thanks

[TOTALLYRC]

I thought that DSPMC controller is not capable count so many pulses in higher speed, that's why I reduce pulses.
I did one more test. I test repeating (with indicator) just in DSPMC AxisWorks softwear. Here I made the same conclusions like in Mach, at higher speeds it sometime loose position (just a few microns). 
I have max. emulation 1024 pulses/rotation.
In the previous post I made a mistake I didn't wrote correctly. I have 800 steps per mm.
So, I'm allmost in the same ratio as you are.
 Thanks

The DSPMC is guaranteed up to 2 MHz encoder frequency (manufacturer spec) and will most likely go faster based on the chips that do the actual work.

I have not seen the DSPMC "loose" position like you are seeing except for when my encoders were going bad.

I would check what the accuracy of the encoder emulation is. It should be an order of magnitude better than the resolution you are looking for. So for .001 mm it should be .0001 mm to not induce cumulative errors. At least in my way of thinking.

If you have a servo with an actual encoder I would try that. If not I am not sure what to do except to contact Rufi from Vital System.

Mike

[Tks]

Check your wireing of your encoders, and make sure that you connect the parrallel resistors over the channels. We had some noise problems, so adding 330ohm over the A and /A signal (and B /B and Z /Z) helpens allot. This makes the signal more resistant to electric noise. Connections have to be rock solid always!