[keen]

I ran a two hour 3D job today with all axis running fast and the tool came back to the start fine and did not seem to loose position.

One thing I notice though. the DRO's are slow to enter values - about 2 or three seconds before they appear - sort of sluggish - and when the machine is running they are just frozen - dosnt seem to matter in practise, but it is a bit disconcerting.

Would that be Path Pilot or my computer?

Keen

[bevinp]

Would that be Path Pilot or my computer?

Sorry I don't have PP installed yet but in the absence of responses I offer the following.
It could be the computer since it is likely that LinuxCNC would make screen updates to be almost the lowest priority (maybe just higher than network or USB interrupts). But you shouldn't have a delay when jogging a single axis or MDI-ing one.
Possibility is the monitor driver, or resolution set within PP is too high for the Graphics Processing Unit to cope with. Hopefully someone will give you better advice, There are a few Linux gurus in this forum.

[keen]

Sorry I don't have PP installed yet but in the absence of responses I offer the following.
It could be the computer since it is likely that LinuxCNC would make screen updates to be almost the lowest priority (maybe just higher than network or USB interrupts). But you shouldn't have a delay when jogging a single axis or MDI-ing one.
Possibility is the monitor driver, or resolution set within PP is too high for the Graphics Processing Unit to cope with. Hopefully someone will give you better advice, There are a few Linux gurus in this forum.

Thanks Bevin - I did a few more checks over the weekend.

The problem seems to be only when I run this horrendous code. the PC/PathPilot must dread it ! It is 3D high speed.... masses of commands..... the lines are screaming though, the steppers are screaming away ...poor old pathpilot/PC is asking for paracetomol...coffee....please!.

I need some just watching it. Anyway slowly the DRO's get behind....and eventually stop. But if I slow down the feed, they manage to keep up fine.

No problems at all with my usual 2 1/2 D code. So I guess on that job the computer is at the limit of its processing power and PP is prioritising commands to drive the steppers . I must run that code on my Tormach controller and see if it copes!

Now here is another question....If the PC/Pathpilot cannot stream the code fast enough, does it just slow down the streaming.... but not loose sync?

I am not sure if that is even happening - it seems to be coping with that code - it is not loosing position....but what if the Feed was higher?

Keen

[popspipes]

Keen,

The computer does lag Pathpilot, it seems the larger the program the more lag, I noticed it on zeroing the axes with a large program loaded, it takes considerably longer for them to zero than it does for a short simple program, however I havent had any problems with the programs running correctly.

I am wondering if more memory might help? My Tormach computer has a minimal amount.

[bevinp]

Thanks Bevin - I did a few more checks over the weekend.

The problem seems to be only when I run this horrendous code. the PC/PathPilot must dread it ! It is 3D high speed.... masses of commands..... the lines are screaming though, the steppers are screaming away ...poor old pathpilot/PC is asking for paracetomol...coffee....please!.

I need some just watching it. Anyway slowly the DRO's get behind....and eventually stop. But if I slow down the feed, they manage to keep up fine.

No problems at all with my usual 2 1/2 D code. So I guess on that job the computer is at the limit of its processing power and PP is prioritising commands to drive the steppers . I must run that code on my Tormach controller and see if it copes!

Now here is another question....If the PC/Pathpilot cannot stream the code fast enough, does it just slow down the streaming.... but not loose sync?

I am not sure if that is even happening - it seems to be coping with that code - it is not loosing position....but what if the Feed was higher?

Keen

Keen,
Since the DRO becomes frozen, then it's clear the processor doesn't have much spare capacity. And increasing the feed during complicated 3 axis motions requiring changes with almost every signal to each stepper will likely reduce further any idle time the CPU. It would be nice if the system is self limiting and will slow streaming in time with the "tiring" CPU, but I don't know enough about LinuxCNC to say that.

I did work on updating software for an aircraft simulator (single processor) quite a few years ago and there the priority routine was feeding each [stepper] its instruction without any delay. If the CPU wasn't able to update all the the new values within the cycle duration then some calcs would not be done, or worse, terminated mid stream leaving stored values inaccurate or unchanged. Maybe the same priority here.
.
Put a post on the LinuxCNC forum, Those guys must know exactly the structure and design of the code and the effect of a slow CPU.
Bevin.

[keen]

Keen,
Since the DRO becomes frozen, then it's clear the processor doesn't have much spare capacity. And increasing the feed during complicated 3 axis motions requiring changes with almost every signal to each stepper will likely reduce further any idle time the CPU. It would be nice if the system is self limiting and will slow streaming in time with the "tiring" CPU, but I don't know enough about LinuxCNC to say that.

I did work on updating software for an aircraft simulator (single processor) quite a few years ago and there the priority routine was feeding each [stepper] its instruction without any delay. If the CPU wasn't able to update all the the new values within the cycle duration then some calcs would not be done, or worse, terminated mid stream leaving stored values inaccurate or unchanged. Maybe the same priority here.

Put a post on the LinuxCNC forum, Those guys must know exactly the structure and design of the code and the effect of a slow CPU.
Bevin.

Good plan Bevin - I have just done that....or as a Brit would say: I have done just that.

Another concern is, could the Dell also handle the Bee line code?

Keen

[bevinp]

Keen,
Sorry to be slow in replying. (see your probe thread for my excuse).

Forget about the Brits, think Kiwi. (I actually typed thick instead of think... I'm sure it was a genuine typo and not some subliminal stereotyping. Anyway, I corrected it so you can't take offence now.) Think ANZers.perhaps?

My flash of genius re using bees is exhausted now, as is my brain CPU after that burst of bee-lining brilliance. So your Dell will be spared a heavy additional load.

Bevin
PS. I am sure you will keep your Tormach audience updated on what you learn from the LinuxCNC thread.

[keen]

Keen,
Sorry to be slow in replying. (see your probe thread for my excuse).

Forget about the Brits, think Kiwi. (I actually typed thick instead of think... I'm sure it was a genuine typo and not some subliminal stereotyping. Anyway, I corrected it so you can't take offence now.) Think ANZers.perhaps?

My flash of genius re using bees is exhausted now, as is my brain CPU after that burst of bee-lining brilliance. So your Dell will be spared a heavy additional load.

Bevin
PS. I am sure you will keep your Tormach audience updated on what you learn from the LinuxCNC thread.

Hi Bevin ....Ha! .... typo thick Kiwi - Good thing you corrected it and I am none the wiser...so to speak.

Nothing straight forward that a kiwi can understand to report from the LinuxCNC thread.

But here is a video.

Cheers Keen

https://www.youtube.com/watch?v=HQhjud7dC5I

[bevinp]

Keen,
Good video and reassuring results. So Dell is coping. Well done.
Yes we don't really need to see DRO readings in a busy program. Surprised Tormach support didn't respond to this thread. I hope they are not subliminarily anti-critics.
Chrome tells me "subliminarily" is not a valid word... I guess it is my fault again, after all, who can argue against Google, worth $13 billion.

It's just lowly me,
Bevin

[tmarks11]

Tormach usually does not participate in this forum. There have been a few times (count them on one hand) that an identified Tormach employee has posted in the past year. This is a management choice to avoid getting bogged down in online arguments.

[bevinp]

Tmarks,
Yes I remembered after my post that Tormach policy is not to get involved in discussions about PP running on anything other than their controller, or about user modifications to PP. A very sensible policy.
Bevin

[keen]

Keen,
Good video and reassuring results. So Dell is coping. Well done.
Yes we don't really need to see DRO readings in a busy program. Surprised Tormach support didn't respond to this thread. I hope they are not subliminarily anti-critics.
Chrome tells me "subliminarily" is not a valid word... I guess it is my fault again, after all, who can argue against Google, worth $13 billion.

It's just lowly me,
Bevin

Hey Bevin - I am not sure what about this thread you are surprised Tormach support did not respond to?

Is it that they recommend Tormach controllers only?....or....?

keen

[bevinp]

Hey Bevin - I am not sure what about this thread you are surprised Tormach support did not respond to?

Is it that they recommend Tormach controllers only?....or....?

keen

I had forgotten about their policy of no support and I thought that your problem of lazy DRO could have been very easily answered by them. I was wrong. Just to confirm Tormach's policy, here is their reply to my query re the use of another computer for PP software. It came from Matt Doeppers:
"
For the most part, we recommend that you use our controller for PathPilot ™: 35286 - PathPilot If you want your system be supported by Tormach, then this is the path you would need to take.

If you still wish to try to convert your own computer, please read the attached document.
Please recognize that PathPilot was developed as a stand-alone machine controller, not as a software to be used on a PC (with other PC functionality). We don’t support any customizations to the PathPilot system.

"
The "attached document" to Matt's reply is from their website and probably is where you got the info on minimum computer specs so I haven't attached it here.
Bevin