Axis 2.9.0~pre: jogging moves the tool cone, but not the steppers

I'm assuming that you're interested only in the two files, attached. Let me know if you need more.

That look basically OK so I would go back to what tommylight said or watchdog issues.
that is, if you can actually jog the axis (after telling linuxCNC its homed) that means the stepgens are working (and generating valid position feedback since you are not getting
following errors) but you are getting no motion which suggests an enable or watchdog
issue

I understand. However, the enable signal works as expected. And, I'm not getting bitten by the watchdog. Note that I've doubled the polling period to avoid this.

Also, one would have reason to expect that if pncconf/test works, that the configuration should also work with axis. That doesn't seem the be the case in my instance.

More thoughts?

If the actual position values are changing in the backplot
(check under view that "Show actual position" is selected)
and you are not getting following errors when jogging,
the stepgen hardware is working so the problem is further
down the pipeline, so either a WD bite, step timing, wiring
or enable operation are the most likely culprits.

Roger all of the above. The numbers in the backplot change in accordance with the movement of the tool cone.

Although I do have problems with an occasional watchdog bite, none has occurred during these tests.

I have checked the signals with an oscilloscope. As I expected, the enable signal is working and consistent with turning on the drivers (which you can hear when it happens), and, of course, the LED indicator on the drivers that tells you that they are enabled. It changes state in accordance with the state of the button on axis. However, neither the direction signal nor the pulse signal changes state. These two, as well as the enable signal, work properly with pncconf which moves the stages perfectly.

It seems a software issue to me.

Communications between the 7i96s and the computer, as judged by the lights flashing on the ethernet port, seem to be working fine. For your reference, ping reports delays as follows:
rtt min/avg/max/mdev = 0.142/0.337/0.875/0.086 ms
It's a mystery to me why the watchdog does occasionally bite, but I wanted to save that issue for another post.

Next!

I would probably bet on a WD bite, That's the only thing I can think of
off hand that can make (mode 0 = step/dir) stepgens still work
(software and feedback wise) but not change the direction pin
when jogging different directions

Well, let's see. If it were a watchdog bite, then at least one of two things should happen. They are:

1) a report of the watchdog bite should pop up on the axis screen; and

2) the LED CR6 should be illuminated.

Neither of these events occurs during my testing.

My conclusion is that it's not a watchdog bite that causes this problem.

That isn't to say that I don't have problems with a watchdog bite. But, it's a different and separate issue.

I'm using 6400 pulses per revolution of the step motor with a 5 tpi thread. The reason for microstepping so much is that it reduces mechanical resonances and therefore increases accuracy and reduces the possibility of skipping a step.

I note again that I'm using an oscilloscope to measure pulses. There aren't any. They are present and operate correctly with the test in pncconf.

One of the things that I understand is that pulse generation is actually accomplished by the FPGA circuit, not by the software. So, if the FPGA is receiving the commands to generate the pulses correctly, they will be generated. They aren't.

I have confirmed that this configuration works with the test in pncconf. FWIW, this is identical to the configuration I was using in with my 5i25 I/O board. I used the high number of microsteps for the same reason then.

Your turn!

Here's a quick update. Lost power for half the day after the last post. Then other things took priority.

Today I tracked down the cause of the occasional watchdog bite. It was the OS interacting with the swap file in some way. Eliminating the swap file (which I don't need) eliminates the problem with the watchdog bite.

I've changed back to the 1ms servo loop period. No problems at all,

EXCEPT: as before, no joy in being able to move the stages in axis. They work fine in pncconf. The tool cone moves ok though. FWIW, I note that jogging with the keyboard in pncconf doesn't work. Only with the mouse. Ahem, there are still some of us who prefer to use a keyboard!

FWIW, I updated LinuxCNC from to the version 6 Sep 2022 version from github. It makes no difference.

Attached are the configuration files.

Your turn!