[SOLVED] EMC stops functioning but GUI continue

Ok, so hitting the machine again this morning. I found it to do the same thing. I'm adjusting PID so I have the FERRROR and MINFR.. set high, about 1 inch. I can jog the axis smoothly for a while, until the dog bites, or until the servos and DRO become unresponsive and only the virtual tool moves when i jog.

John, I looked over the page, and excuse me If i'm missing something, but I'm not quire sure what to do with that instruction. I'm fairly new to ubuntu and brand new to EMC.

I'm assuming motmod is how i reset the watchdog? has_bit pin to FALSE? but i'm confused how to get "into" motmod.

Do i enter the command "loadrt motmod" in term?

If the servos and DRO become unresponsive with no watchdog (which they often do according to the dmesg), then what is happening? My amplifiers seem as happy as pie.... The servos will jog happily until they oscillate violently or suddenly rapid in one direction. It all appears quite random which also confuses me. Most of the time they will jog just fine, but then something goes wrong e.g. quiver, or sudden un-commanded movement, the system becomes unresponsive, and most of the time dmesg will say i have been bit. occasionally they will just stop being responsive after normal motion with no sudden or odd movement.

The machine was only assembled yesterday.

Thanks for being so helpful John and Peter, its contributers like you and this forum that will make EMC grow and further bolster the impact of open source GNU software.

btw, john, nice wing.

Thanks John. I don't think my amp is tripping. The feedback error is 1" and I'm often able to drive it for 14 inches or so, back and forth before the thing freaks out. The status led on the amps is green the entire time. I've seen these amps trip from excessive load, but they haven't tripped on the emc setup yet.

I ran the latency test for about 10 minutes that yielded the 16k and 10k results. The computer is brand new.

What bothers me, is that even though I set the dog timeout to something ridiculous to get it out of the way, the thing will still stop being responsive after a while, sometimes its 30 seconds, sometimes its 5 minutes.

I threw a meter on the output from the 7i33 to see if i was getting any output when it does this, and there is none. I get a happy +-10V when its running, but when it freaks out there is no output....which is what the watchdog does, i think. its possible, i suppose that if watchdog bit and i saw it dmesg when it happened, that it might not tell me again of its still biting if it hasn't been reset, which may be why I didn't see it in DMESG after a closing and reopening EMC.

I'll be out for a week now when i go back to school, but I'll be slapping the thing around again this saturday (30th) when I'm back.

Thanks,

Every time you re-run EMC, the watchdog will be reset
so the dmesg when it fails will indicate if the watchdog has bit

"The computer is brand new" this is not necessarily a good thing for latency...
I would run the latency test for at least a couple of hours and poke about on the machine
while its running to rule this out.

It could be something unrelated like a EPP communication error (though the driver should report any EPP timeout errors
so I dont think its this.)

thanks, Peter. I'll give it good beating and watch the latency this saturday. what kind of values am I looking for that are good or bad?

So...i'm suprised that watchdog resets after emc is restarted. That may mean the restarting having no effect is indicative of my problem...since when i open EMC sometimes the problem continues, and sometimes it doesn't...

maybe I'll post pictures of the machine this saturday as well.

I noticed the dmesg is full of USB stuff, but I have been ignoring it thinking its just internal power supply processes that I shouldn't heed, but perhaps they are important. Do you think it would be better to give it a dedicated PS instead leeching off of USB?

thanks,

Worth a try, maybe you are at the limits of the USB power probably better than external supply would be to tap off of the PCs 5V
or just power your encoders off the external supply

ok, back to work. I'm going to run the latency tests again for a while as i hook up the limit and home switches. I'm taking my values in relation to this page....
wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Latency-Test

I'll let you know what the results are. I'll also fish around in the computers power supply see if i can find a good 5v or two to sneak out....

allright, according to the page i just listed my latency values are well within operating standards. HOWEVER>>>!!!!

I did a little experiment, which yielded surprising results...i'll post those momentarily.....

so, here is the deal. The x axis will work flawlessly for an indefinate amount of time. however, the Y axis causes all the faults. I discovered this by disconnecting the components one by one until the system was reliable and repeatable. When i discovered it was the Y axis that caused us grief, I began to investigate further. what i found was shocking. If i connect only the encoder, JUST the encoder, with all other wires un-plugged, no tangible connection to the amplifier whatsoever, and turn on the the amplifier, the act of turning on the amp makes the Y axis freak out. There is no physical connection to the two minus the shared ground in the case and possibly EMI. I'm not really sure how to explain this... any input?

thanks,
A