Is there a way to reset the card while the GUI is running?

Hi all,

I get joint following errors and errors finishing read while bench testing the motors I'm planning on using, running them off a 7i96s. I haven't had the time yet to troubleshoot the causes of those while building my machine, but having to shut down and restart the GUIs (I've been switching between gmoccapy and qtdragon) after clearing the card via terminal is a bit annoying and am wondering if there's a way to clear the red light without having to exit the GUIs.

Thanks! Ive been searching for a solution since yesterday but so far haven't found a discussion on that specifically

What do you mean by clearing the card?

If you get "error finishing read errors" you can try a much lower
servo thread rate (say 2 ms or even 4 ms) when testing

Ultimately the real time issue on the host needs to be resolved

Hi PCW, thanks for your reply..

I've actually had the servo period set to 3ms after bumping it up from 1ms and I still get the errors but not as frequently.

Re clearing the card: when one of these errors happens, the red light comes up on the card and nothing will respond without shutting down whichever GUI, clearing the error via terminal (sudo mesaflash --device7i96s ---addr 10.10.10.10 --reload), and then opening the gui again.

At the moment and after a fair bit of searching that's the only method I know of that works to clear the errors, it just slows everything right down having to close and reopen the GUI after. The command won't work with the GUI running.

A red light simply indicates a watchdog timeout which is expected
if communication is lost

Restarting LinuxCNC should clear this. If not it likely means
you have some communication or latency issue that needs
to be addressed.

What are your ping times and what host hardware do you have?

(real time errors at 3 ms is pretty bad...)
 

Hi PCW, thanks again

Ping on 10.10.10.10 ranges from 0.092ms to .257ms

Running a Broadcom BCM43225 card and an RTL8111/8168/8411 PCIe ethernet controller

On an older desktop with an AMD A8-3820

You can set the second network adapter for Mesa and check if that one is better at ping times, might be worth the time and effort.
I have machines that are on 24/7 for months (power cuts still occur) and never loose connection to Mesa board.

Hi Tommy, thanks for the reply..

I think they're the same card, I just listed whatever the screen showed me with an lshw command.

Are those ping times bad?

No they are not, but some network cards have some issues with coalescing despite having good latency, so it is worth a try.

For the heck of it I just ran a latency test until I got another watchdog trigger and the max latency was 1025314

You'd think having the servo period set to 3ms would be enough! But somethings keeps making it drop. I'm starting to wonder if it'll be something as trivial as replacing the network cable with some high quality option

It could be cable related but much more likely to be a host issue of some kind.

Does disabling the WIFI make a difference? (some MINIPCI/PCIE WIFI cards ruin Network latency)

Watching servo-thread.tmax may give you an idea of whats going on (its in units of clock cycles on X86)