Magical disappearing 7i84 issue

Hello all.

I’m struggling to figure out a fault somewhere in my mill system. I also have just a few days to do this before my first decent contract using it starts (you know, the one that’ll pay me back for several years’ worth of building machines to learn how) and am tearing my hair out a little trying to fix it.

My mill runs using mesa 5i23 - 7i48 - 7i52 - 7i84. In the last couple of weeks it developed a fault where at any time, linuxcnc would show a run of SmartSerial errors (CRC, Communications error, Too Many Errors error) and I’d have to restart lcnc. For a day or two it was down to giving that error a couple of seconds after lcnc startup, or refusing to start lcnc at all, showing that the first pin of the 7i84 as not found.

I’ve done extensive tidying up of the wiring to try to remove doubtful areas, and had a little time where I’ve been able to operate the machine as normal, but now I can’t get it to start at all, just the startup error list with 7i84 pin not found.

The 7i84 is plugged into COM 0 of the 7i52: CAT5 cable wires from left to right (oriented with the 5V power plug at lower left) are: Brown/white, Brown, Orange/white, Orange, Green/white, Green, Blue/white, Blue.  

Power terminal on the 7i84 has 3x GND wires and 5x 24V wires running straight to a 24VDC PSU (I was using fewer wires and a jumper before, but am trying to eliminate possible causes here). Just now, no other wires are connected to the 7i84.

sudo mesaflash -- device 5i23 -- readhmid shows me the expected list, with one sserial port on pins (TXDATA0 is on I/O 45, RXDATA0 - is on 46)
sudo mesaflash -- device 5i23 -- sserial shows 5 lines about the port itself but nothing about the 7i84.

So, my 7i84 seems to have disappeared!
I have on hand, a spare 7i84 card and 2x spare 7i52 cards. Swapping these out has not made any difference to the errors.

I’ve tried generating a different config using PncConf, that uses the 7i52 and 7i84 (to try loading a different firmware onto the 5i23) and lcnc will start with no 7i84 pins used, but when I assign a 7i84 pin to e.g. x limit, lcnc will not start, 7i84 can’t be seen by the system.

I’m at a loss…it was working this morning, even made some chips! But then a smart serial error came up, and I’ve not been able to start lcnc since.

I’ll happily post hal, ini, bit, pin files etc. or any photos if they’ll help but am not sure where to start and this is already a wall of text.

Any ideas?

Argh.
Andy

As a guess, I would check the 7I52 5V power, low/flakey 5V could
cause marginal communications, CRC errors etc.

Also I would check with a generic hal file that has no 7I84 pin
access and check 7I84 operation by watching the green or red LEDs
(so you can verify operation while swapping channels,cables etc)

Thanks for the quick response, PCW.
External 7i52 5V power was showing 5.041V; I've switched it over to cable power, neither works.
Started the generic hal file I described above generated by pncconf. With the 7i84-related pin commented out, lcnc starts; with it enabled, it doesn't start.
I get CR2 and CR8 on, steady, yellow.
CR6 is on, steady, red.
At lcnc startup (whether 7i84 pin is commented out or not) I get a single green blip from CR7 and then back to CR6 on, steady, red.
Previously CR7 was flashing green in a sequence on startup, then iirc staying green, but now only red.

As a possible fallback - I've tried swapping out both 7i52 and 7i84 boards and it doesn't seem to change the behaviour - I have a 7i80-HD. Would you be able to build a bitfile for my current setup but based on that instead of the 5i23? The config line was:
loadrt hm2_pci config="firmware=hm2/5i23/5i23_fmd.bit num_encoders=14 num_pwmgens=7 num_stepgens=2 sserial_port_0=00xxxxxx"

 A single green blink probably means TX from 5i23 to
7I84 works but RX fails

Did you try other 7I52 sserial channels?
(in case it's a cable/5I23 bad pin issue)

Make sure all sserial channels are enabled  for this test
(In the hal file: sserial_port_0=000000 for example)

I can make 7I80 firmware that matches, which 7I80HD
do you have (-16 or -25)?

 

I tried a whole other 7i52 board, but my firmware uses up the rest of the 7i52 if I understand correctly, so there's no space for other channels.I probably don't understand that correctly though!
Just to check: hal file loadrt line ends with: sserial_port_0=000000
readhmid output is attached as a text file
sserial cable is wired as described above (I made another in case it was that, no change)

Do you mean that I could plug the terminal block into other positions on the 7i52 (COM 1-5 instead of COM 0) and have it work? what else would I have to change to make that work?

Emergency fallback card is a 7i80-HD-25

I'm just looking at updating the 7i84 firmware per this thread: forum.linuxcnc.org/27-driver-boards/4433...es-mesa-7i84?start=0

Good idea/bad idea?

Thanks!

Can you post the readhmid printout here?

If you cannot communicate with the 7I84 you
cannot update the firmware (and if comms are flakey
its probably not a good risk)

Ach, sorry, forgot. Here as attachment:

 

This should be close:

 

My guess is that either the 7I52 cable or 5I23 pin P3.45
is bad. Pins at the end of the flat cables tend to be the ones
that break  first and also those are the pins that connect first
/disconnect last when cables are plugged/unplugged (which
presents an ESD and voltage differential hazard)

A quick check of the I/O pin health is to power down everything,
remove the flat cables and run this test:

sudo mesaflash --device 5i23 --rpo 0x1000   # P2
sudo mesaflash --device 5i23 --rpo 0x1004   # P3
sudo mesaflash --device 5i23 --rpo 0x1008   # P4

You should expect a result of 0x00ffffff in each case

If there is a bad pin on the 5I23s P3, another possible fix
is to move the 7I52 to P4 (and make firmware for that arrangement)