LinuxCNC for possible use with G540/X-Carve

OK, as suggested, after beginning the latency test, I started eight instances of GLXGears, began steaming a video off the net, also set a webcam to send video out to another computer, downloaded 16 GBytes of data, accessed the HD constantly, switched back and forth between four desktops and a dozen or so windows and kept moving them around, and generally "used" the machine. Oh, and I did a system update. After about 10 minutes, the jitter numbers were: 4778 and 5310. From what I can gather, those numbers are acceptable.

Honestly, I don't expect to use the machine for very much other than running LinuxCNC itself, but that might involve reading G-Code off an NFS mounted filesystem, and it might involve streaming a video of progress to another system on the LAN. And there is every possibility I might want to pop open a browser and look something up or check for something online.

. After about 10 minutes, the jitter numbers were: 4778 and 5310. From what I can gather, those numbers are acceptable.

Looks like a very suitable candidate for software stepping.

I would suggest you take up the offer of the configuration already in use by another user and once you are running can tweak and 'personalise' as you wish.

regards

My latency figures are around the 12000 mark and (so far) I've had no problems with anything I've tried to cut on the X-Carve.
Stepconf tells me the step rate is just over 40,000 steps/sec. (does that mean per axis or is it a total?)

Stepconf tells me the step rate is just over 40,000 steps/sec. (does that mean per axis or is it a total?)

I haven't seen that, but then have not used stepconf for a long time.

The max number of steps is decided by the frequency of the thread and the step type.
Type 0 can run one step per 2 polls of make_pulses, maximum.

What is the BASE_THREAD = figure in your .ini file? ( [EMCMOT] section )

You can use the spreadsheet here work out your figures for yourself
www.linuxcnc.org/docs/devel/html/motion/...rs.html#_no_guessing

Need your stepper driver timings in addition to the max jitter figure.

You might be able to get your base-thread figure down near 30,000 depends upon the other factors, but if it runs OK and does not miss any steps, I would leave it alone

The figure is a total, the number of signals stepgen emits is not reliant upon how many axes are specified, all it does is shift a bit pin value for each axis, if that axis position_fb is less than position_cmd.
The stepper driver uses this to generate the actual step, so the timings of the driver are often a limiting factor.

regards

I'm all for leaving things that work, alone
Not sure where this info is (.stepconf?) so here are all the config files for my setup.

From xyz-carve.stepconf

Those are the figures it is working with

Note however that the xyz-carve.hal file does not use the step timings just leaves them to be set by the thread speed, this is quite normal.
The dirhold and the dirsetup are longer than the figures you provided

All parameters were created by stepconf, I don't yet know enough to manually tweak things.

The ones in the top box were entered by you, you just need to be sure those are correct and in the correct base
Manufacturers sometimes use nanoseconds, sometimes microseconds and sometimes a mix

www.geckodrive.com/geckodrive-step-motor-drives/g540.html

If it works, I should leave it, the main values are the step ones and those are taken care of by stepconf using thread length.

The dirsetup etc might be able to be tweaked at bit, they do appear to be 8x higher than the Gecko spec sheet and over twice what you input.
You can try manually editing the hal and ini files when you are happy doing it.