Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: M250cnc on November 24, 2007, 11:47:13 AM
-
Hi I had a problem the other day in which I thought I was losing steps in a canned drilling cycle and this was the first time I had run a canned drilling cycle. I came on here and asked for advice in that in the canned cycle I could not use CV mode only Exact Stop and I thought this was the reason I was losing position. Graham Waterworth helped me out with some example code for a subroutine, I wrote the code “Very Similar In The Amount of Moves In The Canned Cycle” and lo and behold I got the same sort of error.
I then just tried to test the axis and not scrap any more metal.
I then changed my code to move Z up then down by 60mm in a subroutine 60 times I ran the code and got an error of +0.1359mm
I then thought that my motor settings were incorrect “My Motors Are Running At Only Half Speed” I slowed it down some more and reran the code same result.
I then changed the acceleration value from 300mm per min to 150mm per min and ran the code again I got an error of +0.1283mm
I then changed the acceleration value to 75mm per min and again ran the code this time I got an error of +0.1308mm
I the came to the conclusion that there was something wrong with the mechanics of the Z axis so I was about to start redesigning the Z axis when I thought I would try the Y axis
It should be noted that I use 795 steps per rev on this axis and 400 on the Y axis
I changed the code to suit the Y axis and ran the code starting from Zero to -60mm
This time I got an error of -0.2585mm
Then I changed the code to run from Zero to +60mm I got an error of -0.2575mm
I then changed the code to run from +30mm to –30mm I got an error of -0.2550mm
My conclusion now was that there is nothing mechanically wrong with my machine and that I was not losing steps as the results were pretty much the same except one axis was losing and one was gaining I then checked the motor direction, one is reversed compared to the other and that the fact that one axis error is double the other axis is down to the steps per rev being double in these two axis and it just can’t be a coincidence.
My motors are configured as
Dir Low Active X
Step Low Active X
The Drivers are from Motion Control Products, which are very popular and highly thought of in the UK model MDS980
So is this a bug, I am using version Mach3 2.60 by the way
Here is the last version I ran on the Y axis
Starting with my dial indicator nearest to me at Zero on the dial the Y DRO reading +30 I then run the code “Thanks Graham for the help”
F100
G90
G1 Y28
Z15 F1000
M98 P123 L50
Y28
Z0
M30
O123
Y-30
Y30
M99
If someone also using 2.60 could run that code using a driver other than the one i am using say a Gecko and check the results.
Or maybe someone can tell me another reason i am getting this error, i am all ears. ;D
Phil_H
-
Nobody got any ideas what the problem could be ??? I'm desperate to get this working correctly.
Please I'm begging.
Phil_H
-
Have you rechecked the steps per unit>>>
If you command a move in one direction does the axis move that "exact" amount?
(;-) TP
-
Thanks vmax549
I did check that the distance moved was correct before i came to this problem.
You see that there is an adjustment to the Z axis to get it correct so i just changed it to 780 from 795 i got the same error result.
I don't know how all this stuff works but to my mind it seems as if every time a move is it sent it will move the correct distance, but the next move it either adds or subtracts some steps depending on the motor direction settings, if that makes sense.
Phil_H
-
Could this be the reason.
In theory my system is zero backlash but no system could be totally backlash free.
So i have just had a thought, with some backlash that hadn't been compensated for i start off thus.
I move my table onto a dial indicator, set indicator to zero, set DRO to ZERO and if i then continue in the same direction and go 100mm it would in my case go 100mm. But if i instead go back and forth 100mm say 100 moves would a tiny amount build up to give the sort of error i am getting.
Could this turn into a backlash calculator ;D
Phil_H
-
Hi, Phil_H
Have you tried Dir & Step Low Active to "Checked", Some Controls are sensitive to leading and trailing pulse''s.
If you change it then you'll need to change in Config, Home & limits, Reversed coll um's around for correct axis Direction.
Backlash should not be cumulative.
Hope this Helps, Chip
-
Hi afn09556.
I changed first DirLowActive to ticked same result
I then changed StepLowAcxive to ticked "Both Now Ticked" same result
I then unticked DirLowActive same result
So that means i tried all combinations always ends up -0.255
Phil_H
-
Phil,
How far of a move did you make? I have found that the closer to full machine travel you can move & measure the more accurate the machine will be.
One other thought, If the machine worked correctly on an older version of Mach, just reinstall that version.
Hopethat helps.
Cutmore
-
Hi CUTMORE thanks for the help
When i set my steps per rev i did it over 100mm
Even if that was incorrect i should still end at the same place i started. IE go 100mm goes 99 come back 100mm comes back 99.
I didn't do this particular check on an earlier version so it may have been wrong then as well.
Getting more desperate by the hour.
Phil_H
-
What break out board are you using? I had one that was opto isolated and whenever you changed direction, it would cause a crosstalk on the step channel and add a step. So just about every change in direction caused an increase in steps.
Vince
-
Hi N4NV
I am using a C11 from cnc4pc
It is an opto isolated board also my drivers are opto isolated :'(
That is the problem, i am getting some sort of accumulated error.
So how do i fix it please.
Phil_H
-
Hi:
Its possible you have a direction prechange thats too low, did you try making the Direction wdth to 10 on the motor tuning screen? Some drivers need a delay that is a bit wide when changing directions.. Id try a few values, ( same test proceedure) and see if it varys the amount lost , or even corrects it..)
Art
-
Hi Art thanks for your help
I have them both set at 5 is there an upper limit ?
Tried them up to the limits 25 and 15.
The error has now gone from 0.255mm to 0.21mm but they are not losing but gaining ?
My test is from Zero to -60 after the 50 cycles Zero on my DTI is -.210mm now
Phil_H
-
Phil,
You said :-
"It is an opto isolated board also my drivers are opto isolated"
Apparently this can cause problems but not always. It would be worth connecting your step and direction signals from the input of your BOB to your drivers,just to eliminate this as possible problem.
Ian
-
Could you please make a simple prgram to test with that has not Subs ? Please make a file that has a simple Z+ Z- and tell me if you have the same problem.
As a test have you tried the sherline mode under ports and pins?
Thanks
Brian
-
1) What motors are you using?
2) What is the native step resolution of the motors (steps per rev)?
3) What microstepping are you using on you controller?
-
Phil,
Have you made sure you have taken precautions to put noise under control? Some earlier versions of the C11 board you are using could be affected by noise and cause that problem. What is the revision number on the board?
Thanks,
Arturo Duncan
http://cnc4pc.com
-
Hi Brian
I run code as you said the result was an error of -0.2275 "From a consistent-.0.2555mm"
Then ran same code in Sherline mode now error is -0.2625mm
Hi Andrey
The motors are Astrosyn MY259RE
Y axis 400 steps per rev
Using 10 micro steps
Hi Arturo
All my wiring is tidy, signal wires are not close to power wires
The results are very consistent surely with noise i would be getting Inconsistent results.
The board revision is 3.1 if i remember correctly
How could i test for noise
Phil_H
-
Phil,
I would like to send a new Breakout Board so we can rule this out. Please send a me a message with information of your inital order so I can locate it.
Thanks,
Arturo Duncan
-
Hi Arturo.
Thanks for the kind offer, it seems my email is playing up, i can receive but not send at the moment.
So will try later.
Phil_H