On Tuesday 06 October 2015 19:56:00 Peter C. Wallace wrote:

> On Tue, 6 Oct 2015, Gene Heskett wrote:
> > Date: Tue, 6 Oct 2015 16:44:41 -0400
> > From: Gene Heskett <[email protected]>
> > Reply-To: "Enhanced Machine Controller (EMC)"
> >     <[email protected]>
> > To: [email protected]
> > Subject: [Emc-users] joint following errors by the kiloton
> >
> > Greetings all;
> >
> > I gave up on that crashomatic box, and went out and found me one of
> > those Dell 745's with an Intel dual core running at 2.6GHz, and
> > 4Gigs of DDR-2.  But I couldn't move that drive, no PATA connectors
> > on the mainboard.  So I had to install fresh, which went rather
> > swimmingly, taking perhaps a half hour after I'd put my only new
> > SATA drive in it, a 2 Terrabyte Seagate I had more or less reserved
> > for amanda's useage.
> >
> > I of course installed some of my personal pets, like mc, sshfs,
> > unclutter and synaptic.
> >
> > I let the package manager update aroun 350 packages including
> > switching our repo to master-rt, and about mid-morning there was
> > another update to the 2.8.0pre, which I automatically installed.
> >
> > But at that point I was still hacking up low profile brackets for
> > the 5i25 and its p2 breakout.
> >
> > Getting that going, I was amazed that there was not a
> > /home/gene/linuxcnc directory made by the installer!  Is that a bug,
> > or is that a function of running pncconf the first time?
> >
> > I got sshfs working, and copied the 5i25.zip over to the new box,
> > unpacked that and installed the 2 files my 5i25 card needed.
> >
> > And installed boot.ko, needed to start Jon's pwm servo amp.
> >
> > And convinced amanda I needed to recover my /home/gene/linuxcnc tree
> > so I had all my configs and gcode back.
> >
> > And I've been the last 4 hours trying to get the machine to
> > succesfully home.  I don't care how slow you set the speeds in the
> > trajectory or axis# sections, it will throw a joint following error,
> > either while cruising along at 10 IPM (last week it could move 70
> > IPM except for Z & it could do 45!) or in making the stop when a
> > home switch has been hit. Its also moving some faster than any
> > setting in the .ini file.
> >
> > Somewhere along the line it WILL throw a joint following error.  On
> > axis X, or axis Z.
> >
> > Latency test has been run, and the servo-thread jitter is about 40
> > microseconds.  isolcpus=1 is in effect and working.
> >
> > Can this be an sms problem?  Or is this something all aglay in the
> > 2.8 development branch?
> >
> > What is the best way to switch it back to the current 2.7.0 & see if
> > that fixes this?
> >
> > Thanks.
> >
> > 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)
> > Genes Web page <http://geneslinuxbox.net:6309/gene>
> >
> > --------------------------------------------------------------------
> >---------- _______________________________________________
> > Emc-users mailing list
> > [email protected]
> > https://lists.sourceforge.net/lists/listinfo/emc-users
>
> Hardware stepgen following errors usually come from the servo thread
> jitter
>
> If the built in position mode stepgen driver is used, its quite
> intolerant of servo thread jitter and goes bonkers if you have
> significant servo thread jitter and you dont bound its acceleration.
> The normal fix for this is setting the stepgens max_accel parameter to
> about 20% greater than the machines max_accel setting.
>
> This works acceptably in most cases but if you need better servo
> thread jitter tolerance, using the hardware stepgens in velocity mode
> and using a PID loop for position feedback (from the stepgens feedback
> counter) works much better
>
> In this case you set the PID parameters thusly:
>
> P = 1/servo period ( that is P= 1000 at 1 mS servo thread )

No direct feed via FF0?

> FF1 = 1.000
>
> FF2 = number of seconds between HM2 read and write (probably about
> .00005 or 50 usec on a PCI system)
>
>
> MAX_ERROR = ~.0005
>
> In addition to PID, in drastic cases, a config that incorporates the
> DPLL component can deal with latencies up to a few hundred usec.
>
>
> Peter Wallace
> Mesa Electronics
>
These were originally set by PNCConf, and I have not touched the p.i.d 
values that were assigned.  Latency jitter on the servo-thread is less 
than 50000 ns, or 50 microseconds.  The box I took away was an amd, and 
ran in the 30000ns range, with virtually zero problems as long as the 
box hadn't stopped or crashed.  Which was x times daily.

I was just watching the axis f-error with a halmeter, and it seems to be 
a time * speed that builds up as it moves until it finally trips, and if 
I run it a few inches and stop, the .5xxxxx reading at the stop point, 
bleeds off like a leaky capacitor, reaching its minimum value in about 
10 to 15 seconds.

I don't recall ever seeing it behave like that before

X & Y P's are set to 50, Z's P is 75, and all I's and D's are set to 0.0, 
as were FFO and FF2.  

I'll go set it up as you recommend & report back.

Thanks Peter.
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)
Genes Web page <http://geneslinuxbox.net:6309/gene>

------------------------------------------------------------------------------
Full-scale, agent-less Infrastructure Monitoring from a single dashboard
Integrate with 40+ ManageEngine ITSM Solutions for complete visibility
Physical-Virtual-Cloud Infrastructure monitoring from one console
Real user monitoring with APM Insights and performance trend reports 
Learn More http://pubads.g.doubleclick.net/gampad/clk?id=247754911&iu=/4140
_______________________________________________
Emc-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-users

Reply via email to