motion.spindle-at-speed behaviour

My spindle cotrolling hardware has an "at_speed" output which comes in on a 5i25 GPO and connects directly to motion.spindle-at-speed.

when I run this:Rapid motion X 0 Y 0, a pause till stable 500 RPM is reached then feed rate motion to X 25. All good so far, but then feed motion just continues to X 50 while the spindle accelerates to 1000 RPM, I was expecting a pause while the spindle came up to speed.

The hardware input and motion.spindle-at-speed are tracking the spindle speed changes, inactive when spindle RPMs are changing and active RPMs are stable.

The HAL man page MOTION(9) says;
I there something else needed to get the man page behaviour?

Also, the man page talks of spindle numbers "spindle.M.<pin-name>" but a halcmd show just shows "spindle.<pin-name>"

Am I missing something?

Ken

andypugh wrote:

Are you reading the manual for the version of LinuxCNC that you are using?

No, which could well be the problem. I was looking at man pages dated 2021-04-15, which is a lot later than Pathpilot 1.

As for the actual behaviour, have you checked if the spindle-at-speed pin goes false at the speed shift?

Yes, the at_speed pin goes false when the RPMs are changing and true when stable again (hal;show).

PP 1 is getting old, Lucid vintage so I've started the move to PP 2 (again). Not hard to do, I can run my mill on a clean install of PP2, just need to change the ini SCALEs. It's making the computer 'nice' and installing all the other things I run on it that put me off.

There are significant (to me) differences between Pathpilot 1 and 2 as well, ".so" versus ".ko" hal modules, it's not "run-in-place", and the "other" sort of kernel. Some of which cause me some confusion.

I'll sit on the at-speed problem until I get sorted again.

Thanks for you assistance with my difficulties.

Ken.

PP2 seems to be about 50% linuxcnc coding and 50% tormachs own coding.

Just a thought, ( TP

andypugh wrote:

Do you know how PP2 and PP2 map on to LinuxCNC versions?
Is PP even LinuxCNC. or is it Machinekit?

Collective wisdom here says it's LlinuxCNC, I know nothing of Machinekit, visted the website first time today.

In much poking around I've never seen a mention of Machinekit.

PathPilot 1 is LinuxCNC 2.6.0~pre and kernel 2.6.32-rtai

PathPilot 2 is LinuxCNC 2.8.0~pre1 and kernel is 4.11.9-rt7 REEMPT

Vmax549
PP2 seems to be about 50% linuxcnc coding and 50% tormachs own coding.

Only the PathPilot UI is tormach's work, I think the rest is pure LinuxCNC. The UI is a substantial piece of work all the same. I only wish they didn't feel the need to strip out all the bits they don't need from both LinuxCNC and the underlying Linux.

Ken.

andypugh wrote:

I asked Tormach tech support. They didn't know :-/

Well that inspires confidence!

I've had some a small amount of contact with them and they were informative and pleasant.

Downloaded Machinekit just for a look, it appears to be LinuxCNC rebranded with a bit of renaming and soft linking.

Ken

Have done a bit more testing of motion.spindle-at-speed. With this program:table motion pauses for a very short fraction of second at the speed change and then continues as the spindle accelerates to 1000RPM.

Added some dwell time to force a pause at the speed change:this works, table motion stops at the speed change, waits for a stable 1000 RPM, then the motion continues. My spindle takes much longer than the 1 second dwell time to change from 400 to 1000RPM, more like three seconds. The pause appears to trick LinuxCNC into honouring the motion.spindle-at-speed input.

Now I wanted to find the minimum dwell time that would result in the correct (documented) motion.spindle-at-speed behaviour.

A dwell of 0.1 seconds doesn't work, neither does 0.15 seconds but 0.2 seconds does.

What's the significance of 0.2 second dwell?

This is with PathPilot version 1, LinuxCNC 2.6.0~pre and kernel 2.6.32-rtai.

I do really have to get my act together and update.