Computer total freezes only during jogging

Hello to the forum.

I have a equal problem in two machines, both are gantry type.
Although I use linuxcnc in several machines (6 machines that work daily) I only have problems in those two machines. And the problem is a dangerous problem that can put in risk the operator.

The problem is that the PC totally freezes during approximately 3-5 seconds while jogging. We only have this issue while jogging (keyboard jogging). It was never registered during a cycle run. For example during a jog (slow or fast) the machine continues to move "uncontrolled" and linuxcnc (using AXIS) freezes the DRO, mouse - it becomes uncontrolled.
Without (or with the user pressing the keyboard) the system gains again control. The actual machine position is not lost! (we are using a 5i25 + 7i77), and no problem is reported by linuxcnc - no watch dog is triggered. In resume the machine behaves like the keyboard key was pressed. Please note that linuxcnc freezes! it is not a real keyboard press.

We started to think what is different in those 2 machines than the other 4, and the only things that I can recall are:

Distance from keyboard/mouse/screen from the actual PC that is on the electrical cabinet (about 3 meters of powered USB extention). The others machines have about 1 meter.
Distance of power cables up to the motors (longer is about 5 meters)


Things that are equal to all machines:

5i25 + 7i77
Hardware is the same on others machines, including the PC that runs linuxcnc and drives that runs the motors.
Important to say that all power cables are shielded (Siemens/Igus cables)
Electrical ground is all made on one common ground (to avoid ground loops)
Motherboard, keyboard
3 servos


Things that we have done to troubleshooting the error:

1) Keyboard change (brand)
2) Keyboard location - directly connected to the motherboard
3) Full motherboard change (different motherboard, different CPU, different graphics card)
4) Different position of motherboard in the electrical cabinet
5) Different path on the power cables for the servo motors
6) Different user
7) We also placed a even longer USB extension and made it turn 2 times around the servo drives to see if the problem got more persistent - and the result was negative (still we got random frezzes, but there were not larger in quantity)

To make things harder we cannot reproduce the error. Machine can work ok one full day, next day it can freeze right on the morning and several times during the same work that it made yesterday without freezing.
But this is driving us mad here. The problem is very real, and it puts in danger both the machine and the operator as the machine sometimes locks at full speed (25 meters/min).

Any idea, or if anyone had a similar issue, please inform. We are out of roots to troubleshooting this.

Thank you

Sounds like something jogging specific has hung the GUI
If this only happens during jogging I would try:

1. Eliminate USB, that is use PS2 only on KB and mouse (a bit hard with modern MBs )
2. Better still use hardware jogging (jogwheel and real time I/O bits)

Continuous jogging any big machine via a non real time interface
(especially one as flaky and prone to long recovery cycles as USB)
seems iffy at best

Thank you,

But I believe that 95% of machines using linuxcnc jog using the keyboard. Big question here is why this happens.
There is no other report of a similar situation? Anyone?

BigJohnT wrote:

emcPT wrote:

Thank you,

But I believe that 95% of machines using linuxcnc jog using the keyboard. Big question here is why this happens.
There is no other report of a similar situation? Anyone?

I've never jogged using the keyboard, are you sure about the 95% figure?

JT

Of course not, but it is more frequent to see linuxcnc machines with a plain keyboard than more specific control panels. If you see:

www.linuxcnc.org/index.php/english/compo...nce-for-each-machine

You can realize that most users use linuxcnc in small machines, therefor I presume, perhaps wrongly, that the control panel is not a priority and having the keyboard that can jog the machine the obvious choice is already there.

More than the number problem is the problem itself. What can be wrong here?

Did you check the kernel log with dmesg for
any unusual events?

I suspect USB since any noise that triggers re-enumeration
can make all USB devices hang for several seconds at a time

Regardless of the specific cause, I believe that continuous jogging via a
non-realtime device is a bad idea

Sometimes a Wireless PEN or a memory stick yes, but the error happens also without those peripherals connected ... or not ??? .... must have 100% sure, and I will post back.
Cables are different among machines (and tested others).

The issue continues to happen without any peripherals connected. I do not have a comparison on the frequency, but it still happens.
The current motherboard also have PS2 connectors so it is very easy for me to still use the same USB keyboard and use a adaptor from USB to PS2.

Would this simple solution skip the possible re-enumeration problem ?

Thank you.

I would use the USB keyboard and mouse through USB to PS2 converters as first choice.

I have had very high latency spikes in the past which I tracked down simply to a USB mouse being moved!

There is an added bonus that on older machines you can access BIOS and the GRUB menu, which you can't with a USB keyboard, because drivers have not been loaded at that stage.

regards