On Sunday 13 December 2015 07:22:52 Ben Potter wrote:

> > And that would mean that direction is changed one step later. Which
> > is
>
> fine, because the next direction change would also be done one step
> later and thus overall distance of travel still is the same.
>
> > How do I invert a step pin in HAL? I did not see anything in
> > "stepgen" section in hostmot2 doc page. Or is it done by treating it
> > as a gpio pin
>
> with the invert_output parameter (with finding out particular pin
> number that corresponds to particular step pin)?
>
> Yes, the relevant lines in each section for me look something like:
>
> setp   hm2_5i25.0.gpio.006.invert_output true
>
> > I would like to thank everybody for the ideas, my current to-do list
> > looks
>
> like this:
> > 0) update LinuxCNC to 2.7.something to use new tp (will provide
> > smoother
>
> execution as the g-code consists of small g1 moves);
>
> > 1) try inverted step pin;
> > 2) recheck steplen, stepspace, dirsetup and dirhold parameters and
>
> increase them, if possible;
>
> When going through my testing for this the first time I cranked all 4
> settings up to 10,000 - the problem still occurred. After I tried
> inverting the output I slowly stepped them back down - for my G540 I
> settled on 700/700/2000/2000.

The Gecko 540 is not in the charts of our wiki, darnit. What is there 
isn't even that fast. See:

<http://wiki.linuxcnc.org/cgi-bin/wiki.pl?Stepper_Drive_Timing>

I take it you do not have any opto-isolation in those 3 paths.  That 700 
ns is pushing an opto's ability to switch that fast.  The BoB I used was 
all opto, and I was warned about its speed by someone (Steve Stallings?) 
on this list. I don't recall which channel now, but I had to remove one 
of them and jumper across it. I think there are probably faster opto's 
about, but what I saw on the scope just wasn't up to getting the job 
done. All the loads that BoB see's are already isolated, so I wasn't 
terribly concerned about losing one stage of isolation.

All my drivers are just 3 types, 2M542's, 2m860 and Pico pwm-servo's for 
2 of the 3 spindles at my place.  And I'm not pushing any of them that 
fast.  But because of that slow BoB, ISTR I am running dir setup and dir 
hold about 1.5x longer than the wiki shows for the LeadShines.  So my 
experience is not with a wide range of drivers, but I will no longer 
consider opto-isolated BoB's because everything I have is already 
isolated. Best BoB IMO, 10ns response times, but its been taken down, is 
the cnc4pc C1G that I have 2 of.  What he is selling now is NOT the 
versions I have. Everything on mine has an led tally, which simplifies 
troubleshooting considerably.

Also, generally, where you have the choice of commonizing the + terminal 
of a driver loads input, you are far better off doing that, and having 
the BoB pull it to ground as the pulldown in most TTL circuits is much 
more robust. Having the reverse setup, where the - terminals are 
commonized, is generally NOT a good idea due to a lesser amount of 
pullup power to light up the opto in the drivers own input.

> I can run with lower timings than that, but am more prone to losing
> steps on heavy cuts, as they are I stall my spindle before I overload
> the steppers. Losing steps is thankfully rare now.
>
> > 3) recheck maxaccel values and do a test with considerably lower
> > values; 4) if the issue is still there - use dial indicator to
> > actually measure
>
> exact error, torture the machine a little more and then post here
> again :))
>
> > Viesturs
>
> Another option that may be possible is putting a temporary encoder on
> the z axis motor - even running it open loop you should be able to
> find out if a discrepancy is still occurring after each change.
>
> Ben

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Some mill pix are at:
Genes Web page <http://geneslinuxbox.net:6309/gene/GO704-pix>

------------------------------------------------------------------------------
_______________________________________________
Emc-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-users

Reply via email to