Caddy intermittent Canbus issues

[picowight]

Working on an 04 Caddy ostensibly to get the folding hard top working , however it has a history of (very) occasional complete sudden shutdown, I scanned it after an occurrence a few weeks back and it was obvious from a plethora of U codes that the CAN network had gone down.

Working on it yesterday and accessory mode on, engine off, car has a network hissy fit, entering a new power mode of "limbo" neither on nor off!
Scoped the CAN from a BOB ... yuk!

Battery out over night,next day car is normal but CAN but has the dominant state being pulled down to about 1.3-1.7 volts when there is no traffic.

This was running today, notice the voltage drop out as well.

I split the network DLC, BCM, VCIM/5 other modules but it appears to occurr on both sides, less visible on the heavily trafficked side but still being pulled down. I scoped the 5 module side in acc mode and waited until they all went to sleep

As it is on both sides of the split network, and affects both H + L I'm think whatever generates the CAN carrier signal ?
Ideas welcome!

[liviu2004]

Do you use ac coupling?

[picowight]

No, not on a CAN signal. Excuse the ancient version of Pico it's on my Linux desktop and it has not been updated in years!

Topology, red shows disconnect.

Going back through the waveforms I note that when the error was occurring (first very distorted pic above), after I split the network and still scoping from the DLC it did come back to a normal signal, but the messages just looked like a heartbeat rather than comms - though I guess there was only two modules on the network at that point.

Clearly significantly different to the before the disconnect (though a pull down is still evident-) but it did not fix the power mode issue.

The car starts, runs and drives fine, I am thinking that whatever is pulling down the voltage is what is causing the intermittent issue. When I split the network - at a connector, I also disconnected the suspension module (ESP) as I thought that might be the culprit and tested the network in either direction from it's plug.

[picowight]

So day 1 CAN bus completely unuseable
day 2 as above with the recessive voltage being pulled down
Day 3...

Recessive voltage is stable but about 20% of the data packets are at a higher voltage. So that points to an issue with one of the modules. I would say at this point I had a charger on the car and this was with the ign on, however the results were the same on acc where the previous captures were made. Also the same with the engine running.

[muttnjeff]

Intriguing thread, been following along and am curious as to what you find out.

[picowight]

Well I found out some very interesting things.

The "bad" waveform in the first post .. isn't! This is what GMLAN looks like when it's in sleep mode 1.4-3.6v exactly. See Fig 7 https://www.vehicleservicepros.com/serv ... techniques

Then there appears to be a sleep mode bias voltage of about 1.5v that is being put out onto the network even when the engine is in ACC or Running. GM owned Allison, see Fig 3 "known good" in this article https://www.vehicleservicepros.com/serv ... 0-no-start

So it appears that some part of the system is intermittently putting the car into sleep mode when it should not (and will not wake up) and keeping it there even after starting. I'm making a plan to look at the ignition/run button (keyless entry and start) and the wiring to each module that tells it to wake up/sleep. The BCM is the ignition mode decider and the Remote entry module is the back up, Midchells gives me the logic but I have not looked at it yet.