6i25-7i77 Comunication errors

Hi,

I've been struggling with this issue for at least 4 months now, the machines runs perfectly for two hours, (Plus-Minus), then comes up with Mesa errors 13, 14, and sometimes 4.
It's to do with serrial communication, and I've had permanent problems with this. I've also seen more posts here on the forum about it, which reliefs me that I'm not the only one struggling with these issues.

Has anyone else experienced this problem and came to a solution

What version of LinuxCNC are you running? I have one machine that has sserial errors due to noisy electrical cabinet and for now I'm running 2.6.x.

JT

This indicates that there is noise picked up (Greater than a TTL threshold = ~1.4V)
in the DB25 cable between the 6I25 and the 7I77, either from a ground loop or
induced noise from nearby cables.
Re-routing cables and checking carefully for grounding issues may help

Also if you have a noisy electrical environment, using the shortest possible cable from the
6I25 to the 7I77 will help.

If you are running LinuxCNC 2.7.X, this is made worse by a error reporting bug
in the hm2 driver that makes single transient sserial errors fatal

We are working to fix the driver bug and we also have a firmware modification that
improves FPGA RX noise immunity :

freeby.mesanet.com/filtered.zip

for a 7i77 unzip this file and run:

sudo mesaflash --device 5i25 --write 77f.bit

and then either power cycle the machine or run:

sudo mesaflash --device 5i25 --reload

to load the new firmware

Hi Peter,

I followed all your instructions, I'm still getting the following errors:

Dmesg output:
[17140.402340] hm2/hm2_5i25.0: Smart serial card hm2_5i25.0.7i77.0.0 error = (13) Communication error
[17140.402345] hm2/hm2_5i25.0: Smart serial card hm2_5i25.0.7i77.0.0 error = (4) Extra character
[17140.404397] hm2/hm2_5i25.0: Smart Serial Comms Error: There have been more than 1 errors in 10 thread executions at least 200 times. See other error messages for details.
[17140.404405] hm2/hm2_5i25.0: ***Smart Serial Port 0 will be stopped***
[17678.356929] hm2_5i25.0: dropping AnyIO board at 0000:03:00.0
[17678.356937] hm2/hm2_5i25.0: unregistered
[17678.356970] hm2_pci 0000:03:00.0: PCI INT A disabled
[17678.357008] hm2_pci: driver unloaded
[17678.360488] hm2: unloading
[17680.591119] RTAI[math]: unloaded.
[17680.609471] SCHED releases registered named ALIEN RTGLBH
[17680.671755] RTAI[malloc]: unloaded.
[17680.770462] RTAI[sched]: unloaded (forced hard/soft/hard transitions: traps 0, syscalls 0).
[17680.773428] I-pipe: Domain RTAI unregistered.
[17680.773538] RTAI[hal]: unmounted.

I don't know what to say anymore.
I've supplied external 5V supply to the 7i77, I'm using 7i77isol for each 7i77,
I've done the mesaflash with the 77f.bit filter, I checked my voltages, and monitored them, both 5V, and 24V is stable, the 3.3V on the 6i25 is also stable, I'm using the DB25 Cables provided by Mesa, ... I've really tried my best on everything, and have a strong feeling that there is something wrong with the latest 6i25/7i77 batches.

I have also moved the Computer, Cables and 7i77x2 away from all high voltages.

I have used many sets like these, and never had this many issues. We have more than 50 sets of these in the field, but the last two batches that we received, just cannot work for more than 2 hours, then these errors occur.

I have been struggling with this since last year October, and towards end of November, I decided to get busy with other things to get my mind off these things, because it just pulls me right down.

please help.

The main thing that changed is that the error handling in hm2 driver in
linuxcnc 2.7 has a problem with sserial errors. It ends up treating single
errors as fatal. As a first step, I would suggest reverting to linuxcnc 2.6
until this issue is resolved.


If you have sserial errors using the 77f.bit file it does mean that you have
enough noise coupled into the DB25 cable that you have at least 1.4V of
noise of longer than 200 ns duration.

Okay,

I have now diverted back to 2.6, and will have to try it out tomorrow, when I'm back at the factory.

Thanks for the reply. I will most likely give feedback tomorrow.