Problem with a MESA 7i76

Sorry to hijack the thread, but it was so similar and recent it didn't seem worthwhile to start a new one; I have the same symptoms of stepgen 0-3 not working on my 7i76, but no complaints of any kind from my 6i25 (in a new Dell box, replacing the 5i25 from this thread: forum.linuxcnc.org/27-driver-boards/36634-5i25-deterioration ) during LinuxCNC startup.

The only difference is on mine the TB2 block pins alternate 0V and 5V during power-up, rather than being all the same, but with no state changes on PncConf testing or LinuxCNC jogging.

I set PncConf stepgens 4 and 5 to mirror the X axis on stepgen 1, and started running the axis test. Every + step pin on TB2 remained at 0V, every - pin at 5V, but the + step 5 pin on TB3 scoped as the expected square wave that I used to get when the mill was working (all I did was move the electrical cabinet and rewire with routing and crimp terminals!).

I just bought a new 7i76 to minimise downtime, but would you agree that here too it appears that the two TB2 chips are damaged? If so, how much would you estimate it would cost to replace them at Mesa so I could keep this board as a backup? The field voltage and logic all appears to still work just fine (I have ENA lines on TB6 outputs 0-1-2, no spindle control as yet).

If the +STEP and +DIR outputs are high (5V) until LinuxCNC starts and then go low,
that means that they are OK, Likewise, if the -STEP and -DIR outputs are low until LinuxCNC starts and then go High, that means that they are OK.

This suggests the problem is elsewhere (firmware, hal setup etc)

There is no state change upon LinuxCNC startup. + outputs remain low, - outputs remain high. When I use PncConf>Test/Tune>Run for reciprocating travel, the +DIR output oscillates between about 50mV and 80mV, and the -DIR output between 4.915V and 4.920V during direction changes. So there is some signal getting through, it just doesn't appear to be getting amplified/conditioned appropriately?

6i25 firmware should be 7i76+7i85 (I was going to run glass scales, but didn't seem to end up needing to, but it works for a remote MPG box) and currently just running/connected the 7i76.

I'll continue poking around in software, maybe re-build the hal with a new PncConf runthrough and double-check some stuff in the terminal. Worked perfectly previously with no hardware changes, though.

OK I misunderstood. if the output pins do not change state on starting LinuxCNC its likely
that the 74ACT540 and 74ACT541 chips have been damaged (or the series resistor
networks on the outputs have been damaged)

This can happen in a couple of ways

1, Accidental contact of step/dir pins to 12/24V
2. accidental overvoltage on the 5V 7I76 supply (7V will destroy ACT parts)

Number 2 is worse since is will likely damage multiple 5V parts

Hmm... I don't recall anything happening during rewiring that would have caused either of those to take place. The only 5V in is via the parallel cable, all lights and 6i25 seem happy. I'm guessing though from the "74*" that those chips are some kind of shift register or the like, and the serial comms isn't going to notice that an open-loop output isn't working. Perhaps though, since I did see distinct signal patterns on the output, it's the resistors at fault rather than the ICs specifically.

Is there any reasonably doable method of diagnosis I can perform on my end to find out which is more likely? I see the S/D* and RN* screenprint labelling, but no obviously visible issues on the little 8-pin devices.

You can measure the resistance across the networks (if you have a steady hand)
they should be 27 Ohms

The 74ACT540 and 74ACT541 are just buffers