Re: [Emc-users] Jitter, latency , rtapi error

[Gene Heskett]

On Wednesday 07 March 2018 22:05:40 jeremy youngs wrote:

> Doubtfull, if the interface is wired so the + terminals are connected
> to the 5 volt buss, and a logic zero at the output then becomes the on
> signal, tied to the 4060's - terminals. It will likely run just fine
> on faster signals. The diff is that if the logic signal is tied to the
> + terminal, the pullup from the interface to turn on the opto's in the
> 4060, isn't near as strong as the logic zero sinking ability. I have a
> variety of drivers, and none of them are running as slow as the
> choices in stepconfig. Set the polarity of the signals in the hal file
> so that the step signals are negative for a logic one. You may want to
> check that with a real scope.
>  Invert pins to sink instead of source , more current handling
> ability. Run the other script in terminal, for 30 minutes copy report,
> paste here . Will do , that will take a bit of fanagalling as no
> internet at the shop , so be patient ( should have tomorrow evening or
> so ) , you said run something, can I run linuxcnc?

I believe thats what I wrote.

No net at the shop? Scary, how do you update?. All my machines are net 
connected, but the only firewall is dd-wrt reflashed into the router. No 
one has gotten thru it in a decade, one of the internets best kept 
secrets.

>  I believe the on board video is Intel , don't Know if 915 . It gives
> the same rtapi error regardless of video used.
> The old motherboard ate ram at a pretty decent rate

One thing I've noted is that 4GB of dram seems to be a sour spot, and 
gets into swap very easily. 2GB or 8GB is much less of a problem, with 
memory allocation being much more tightly controlled at 1 or 2 GB than 
at 4 GB. I have an r-pi 3b, with only a gig of ram, and with 7 days of 
uptime, is only into swap about 7 megs and it has done a few hours work 
in that 7 days. Had a power bump last week else its uptime would be 60+ 
days.

> I figured the caps 
> were bad and a reconditioned Mobo was less than the caps , I figured
> the Mobo would help the rtapi , no such luck. I about have this y
> ballscrew contained , still not the final revision but it will be
> better than before , and I will get to use in stock screws so that's a
> plus.

Contained? please elaborate. "stock screws" generally are metal to metal 
threads and waste 90+ percent of the motors power in frictional losses 
compared to ball screws.

> Also is there a tutorial on smi ? Perhaps YouTube? I will myself but I
> have been puzzled about this and isolcpus , it's kinda my
> understanding that those were for previous versions of rtai .
> Thank you for your time Gene .

I believe there is a discussion about smi on . 
Might have to search for it.  Whether its still valid I'm not positive 
and will assume the final word is in the wiki.

isolcpu's has I believe been deprecated, I don't know that I am using it 
on any of my 4 current machines. OTOH, I am running 2.8-pre on all 4 
machines. There could be some diffs between the latest 2.7 and 2.8-pre.

> --
> Check out the vibrant tech community on one of the world's
> most engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users



-- 
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 

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Jitter, latency , rtapi error

[Gene Heskett]

On Wednesday 07 March 2018 19:23:32 jeremy youngs wrote:

> Hello fellas ,
> I have this optiplex 2.8 GHz . I'm trying to figure if I should pitch
> it and try something else. I have reviewed the docs on latency and
> after 24 hours running I get what I think are good numbers, a Max
> jitter of 8882 and a service thread of 27405 . By comparing the given
> examples in the docs, I would think this machine would do rather well.
> It has a new Mobo , a pcie graphics card , and a pci epp card . It is
> rosewill and the graphics is an ati iirc.
> Latest bios available, no hyperthreading option, only disable multiple
> CPU core ( makes no difference)
> I have always gotten rtapi error: unexpected realtime delay on task
> one . 

I have seen that. Generally associated with the use of ati or nvidia 
closed src video drivers neither of which is allergic to shutting down 
the IRQ's when they think they are busy.. I have 2 of the optiplex 
myself.  One is a utility box I keep for programming mesa cards, the 
other is running my G0704 via a 5i25 mesa card.. The onboard video on 
both is i915, an intel video, not ati, but thats this particular pair, 
other optiplexes may not have intel video.

I've had relatively little of that sort of problems running the linux 
version of the i915 driver.

> I have my steppers set to kelling 4060? ( The only kelling 
> setting in stepconfig) these are the slowest of them all. I have
> eliminated optos . I get good performance from the machine and don't
> really want to punt. I've done the upgrades in hopes of solving the
> rtapi error. I've installed new cmos battery as well. Do I need even
> slower step settings?

Doubtfull, if the interface is wired so the + terminals are connected to 
the 5 volt buss, and a logic zero at the output then becomes the on 
signal, tied to the 4060's - terminals. It will likely run just fine on 
faster signals. The diff is that if the logic signal is tied to the + 
terminal, the pullup from the interface to turn on the opto's in the 
4060, isn't near as strong as the logic zero sinking ability. I have a 
variety of drivers, and none of them are running as slow as the choices 
in stepconfig. Set the polarity of the signals in the hal file so that 
the step signals are negative for a logic one. You may want to check 
that with a real scope.

It might be illuminating to launch "linuxcnc -lENTER" from a terminal 
screen as that will log anything suspicious to the terminal screen. 
Leave it run, preferably doing something, for half an hour, then quit 
and copy/paste the screen to another message here. That will give us an 
idea of how serious a problem you have.

-- 
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 

--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

[Emc-users] Jitter, latency , rtapi error

[jeremy youngs]

Hello fellas ,
I have this optiplex 2.8 GHz . I'm trying to figure if I should pitch it
and try something else. I have reviewed the docs on latency and after 24
hours running I get what I think are good numbers, a Max jitter of 8882 and
a service thread of 27405 . By comparing the given examples in the docs, I
would think this machine would do rather well.
It has a new Mobo , a pcie graphics card , and a pci epp card . It is
rosewill and the graphics is an ati iirc.
Latest bios available, no hyperthreading option, only disable multiple CPU
core ( makes no difference)
I have always gotten rtapi error: unexpected realtime delay on task one .
I have my steppers set to kelling 4060? ( The only kelling setting in
stepconfig) these are the slowest of them all. I have eliminated optos . I
get good performance from the machine and don't really want to punt. I've
done the upgrades in hopes of solving the rtapi error. I've installed new
cmos battery as well. Do I need even slower step settings?
--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users