THC Stopped Working - Both THC Enabled and THC Active not illuminated

Today's update.

Went back into the shed and started it up and THC decides to work today. Perhaps there was a bug on startup, I'm not quite sure.

 

snowgoer540 wrote:

I am not familiar with the nightmares you had setting up your current config, if it was install related or hardware related or both. A friend of mine had freezing issues with a Gigabyte Brix with the J1900 processor, I even bought one (and sold it shortly thereafter) to try to work out what the issue was and never made it anywhere, just doesnt like linux I guess (in all seriousness there was an issue logged in intel's known issues list, but I cant recall what the exact issue was, rest assured they arent going to fix it ).

So far I have tested
- Windows XP Home edition
- Windows 7 Home Premium
- Dell Optiplex 990 (currently in use)
- Gigabyte GB-BACE-3160 

All computers had the same issue where they would set off the:

"Watchdog has bit! (set the .has-bit pin to False to resume)
Unexpected realtime delay on task 0 with period 105000"

The gigabyte mini pc would freeze all the time before I could even launch anything.

I am still not sure if the issue is related to the computer or if the error is coming from a power connection (checked every connection at least 5 times) or the Mesa 7i76e.

At the minute, the Optiplex only occasionally shows the message. 1/3 startups it will come up at the start of opening PlasmaC (within the first 5 minutes), and when it does this I may need another 3 times turning everything off and on before it comes good.

The times when it does not come up in the first 5 minutes, I can usually (9/10) run it for hours without an issue. 

This week I have been on a hot run and only encountered the issue 1/7 days, othe weeks it is 5/7 days. It is completely random.

Does anyone have any ideas?

I would be happy to send the gigabyte to anyone in Australia who would like to test it, and I would also be happy to pay for a computer with ethernet connection that someone has successfully installed PlasmaC or QtPlasmaC on and ran with their system.
 

little_sparky wrote:

So far I have tested
- Windows XP Home edition
- Windows 7 Home Premium

? I dont follow this part

The gigabyte mini pc would freeze all the time before I could even launch anything.

I had similar experience with the Brix.

Does anyone have any ideas?

In searching a bit to see if there was anything I could suggest to try, I see you have gone through a lot of the options with PCW.

In this post:
forum.linuxcnc.org/plasmac/39317-watchdo...-bit?start=10#171559

He asked for the results of (later corrected to below in the last post):
But the thread seems to have died without the results.


He also suggested:

On possible thing to check is if a slower servo thread helps, so you might try 2 ms
(2000000 ns) This will require changing all the PID P terms to 500

Have you tried this? This is necessary when running PlasmaC on a raspberry pi 4, and has no noticeable effect on performance.

The Dell 990 uses a 82579LM Ethernet chip, and I have seen some reports of this
being buggy and unreliable for realtime not sure if this is the issue here

If it is an issue an add-in Ethernet card should fix the problem

Also a possible solution.

Sparky, there is an address here you can send your brix to here.  www.vmn.com.au/
I'll try and get it going for you when I have a free moment. You can also contact me through that site via chat or the contact us page. 

I ran a Brix J1900 for a long time on my spaceship and replaced it with another PC when Mint refused to run Python 3 in master.
I now have it running again using Debian unstable and the linuxcnc-uspace package in that version of Debian.

I also have a Brix 3160, probably the same model you have which I also had running on the spaceship. Its now in daily use using Linux Mint on my Packing bench. It ran Windows 10 for a while but I trashed that OS after it started to blue screen every time I printed something. 

Sometimes the J1900 would freeze but never when running a program. Only when idle.. This has been powered up for a month now with no bad behaviour.

Just note you cannot use the internal wifi card, pluck it out and get a USB wifi dongle. I use an Asus nano.
The latency is around 15 usec with the 5.16 kernel which I found gobsmacking. Loading a program or opening a program will see a single spike out to about 110 usec which is still perfect for a mesa card.

I'll post a quick screen shot in a sec..

Gigabyte GB-BXBT-1900

little_sparky post=238966 userid=24582
So far I have tested
- Windows XP Home edition
- Windows 7 Home Premium
? I dont follow this part

My bad, these are the operating systems I found on the side of the computer and thought they were the models, they are a couple of old (2000-01) 32 bit desktop computers that were no good.

In searching a bit to see if there was anything I could suggest to try, I see you have gone through a lot of the options with PCW.

In this post:
forum.linuxcnc.org/plasmac/39317-watchdo...-bit?start=10#171559

He asked for the results of (later corrected to below in the last post):

Code:

halcmd show param *.tmax

Towards the end of the thread I started to get some 'ok' results and was able to get by for a while with the level of performance I was achieving. Eventually I ran into some new-ish problems and started a new thread. I should have probably linked the two together.

Code:

[code][code][code][code]One possible thing to check is if a slower servo thread helps, so you might try 2 ms (2000000 ns) This will require changing all the PID P terms to 500 Have you tried this? This is necessary when running PlasmaC on a raspberry pi 4, and has no noticeable effect on performance.

[/code][/code][/code][/code]

I will definitely give this a go.

Code:

[code][code][code][code][code][code][code][code][code][code][code][code][code][code]The Dell 990 uses a 82579LM Ethernet chip, and I have seen some reports of this being buggy and unreliable for realtime not sure if this is the issue here If it is an issue an add-in Ethernet card should fix the problem Also a possible solution.

[/code][/code][/code][/code][/code][/code][/code][/code][/code][/code][/code][/code][/code][/code]

Are you suggesting replacing the 82579LM with another 82579LM, or replacing the current with an aftermarker ethernet chip, something like  this