Advanced Search

The latency will limit your maximum speed but not interfere with basic LinuxCNC operation.

Have you gone through the various latency improvement options:

Disable all speed switching/turbo modes in the BIOS

Disable any power management in the BIOS (C states > C1 etc)

Disable hyperthreading

Set isocpus-L in kernel command line (with grub)
(where L is the last processor so 3 for example with a 4 core machine)
 

Good afternoon.    Well after a few mis-starts,   I am pretty sure I have a good Linuxcnc 2.9.4 install with Debian 12,; on a old Lenovo Thinkcenter  with a core3 processor, and a single onboard parallel port and a 1Tb Solid state HD      I have run a latency test, and the results are not super great, and quite a bit worse today than yesterday.  (yesterday the servo thread was  69200, and the base thread was 32900).   Here is a photo from today; after i opened up some calculators and the video player with no video to play..    Is this going to be a non starter because of the latency?    This computer was a freebee,    i was going to try a old windows XP machine, and i found out last night that would not work regardless,   So this is what i have for now.    If this is going to be a non starter for latency  i would like to know sooner than later  I am just barely capable of basic Linux navigation with a bit of help from google.    I had older Linux version 3 years back that seemed a bit different, I am trying to learn the newer version.       I am not sure i am even testing correctly.   I am have the LinuxCnc on a boot thumbdrive.    and once the computer is booted,  i am starting the Latency test from the Applications menu topmost left screen.    Then the CNC tab, and then Latency test.    As far as i can tell this HAS NOT started the actual LinuxCnc program which i thought was the correct way to go?     If i just run the latency test and do not exercise the machine i get  servo at 9139, and Base thread at 19360     The computer as i said is a freebee.   My IT friend gave it too me and we changed the OS to Linux,   and we then got the LinuxCNC image burned to the thumb drive.          I am not savy enough to know ,   But If we just wiped out the old OS which i think we did,   can or is there still some junk maybe running or causing problems from the old windows OS  or the hard drive,   that can mess with the latency????     When I open LinuxCnc it opens to the configuration selector  ( good?),   But i don't think i am ready for anything else at this time;?   If the Latency can be solved,  then I believe i still need to pick a parallel port breakout board  Correct?     For the breakout board i have been looking at the c10G and C11G break out boards,    I am hoping for a bit of help picking the break out boardI have a few different applications in mind for LinuxCnc, but the first application will be a  3 Axis Lagun mill that used to have a Crusader M control.   the machine went under salt water up to the quill,   the  ball screws seem ok , and the head and quill still work,   but the old Crusader M control and the old DC servos are toast,        I wouild like to put new AC servos back on the mill,  and i have been leaning toward the ClearPath  servo with amp combo, and i wanted to keep to the parallel port for now.    So is the latency going to be a problem?   can it be fixed?   Should i just find another computer?thanks 

That must be related to the source of your spindle-at-speed signal.

Maybe its always false at a zero speed, which will cause issues with
LinuxCNC if you try doing a cutting move (G1 etc) with the spindle off.

Do you need to do cutting moves with  the spindle off?

If so:

If you have a spindle encoder, you can generate a spindle-at-speed signal
by comparing the commanded spindle speed with the actual (from encoder)
spindle speed with the "near" hal component.

If you do not have a spindle encoder you could add some logic to force
spindle-at-speed true when the spindle is off (M5)

Hmmm, second time I'm writing this as the last one seemed to vanish. So sorry if it winds up as a double post

Running whilst monitoring spindle-at-speed
It starts off false
M3 S10000, stays false until reachin the correct speed, then goes true
M5 turns false, and stays false even when the spindle has wound down

It don't know if this is correct, or if M5 is conceptually setting a speed of zero, and therefore spindle-at-speed should be true when the spindle stops

no retro fits on these i was just saying the memory for program storage is what ever the base model came with at the time. sadly the people i work for don't spend the extra money to upgrade that part of the machine when they buy them.

they guys were old school who have retired or moved on and they hand coded everything in the past so bloated cam programs were never an issue before because of this. the one mori is nothing fancy at all but when i first came out of trade school back in 1991 they had that machine at it's still there so it's 35 years old plus