question

This is the circuitry of the pcb that's in the DB9 housing (99% confident).
It's a 4 layer pcb (not needed IMHO) and some connections have been hard to trace (vias under pads are a big no-no in the industry for example, but have been employed here). The orientation of the diodes is guesswork, due to the unit I have here being damaged. How they are depicted is most likely going to work to create the power for the ADuM and the RS232 voltages from the DTR/RTS pins of the DB9 connector (not convinced this is a good/proper design - comment by the EE that helped me), but it's not working from +5-12Vdc then. It would need -12Vdc to run per the logic.. anyhow, I'll get back to this once I get a new cable and can look at it with an oscilloscope while still in working order.

For deploying this in the field it's just not properly engineered enough though (things like the isolation that is being created by the ADuM is being screwed up by too close traces/vias somewhere else on the board), I'll have to look elsewhere for a solution I can trust to run reliably.

Did some more testing and then discussing with the EE.

The original plan of the whole circuit seems to have been to run the ADuM from the TX pin of the computer alone. Changing TX (RS232) from logic 0 (+??V) to logic 1 (-??V) would via D8 draw capacitor C5 down to store the negative level for Q2 that the RX of the VE.Direct signal needs to be recognized by the computer (couple times back and forth and it's possible to drive something with this).
From that very same negative level the LM555CM was supposed to make the positive voltage to drive the ADuM and also Q1 so the VE.Direct device can recognize the computer signals (change settings on the VE.Direct device).

But the LM555CM seems to have been too power-hungry for this and they cut the wiring and instead got the positive power from the DTR pin of the RS232 interface. This is done by the Victron Connect App upon starting (or anytime the 'refresh' button is hit).. it does take control of any COM port available on the machine and it sets pin 4 (DTR) to +??V (logic 0).
So now they got a source for powering the ADuM and funnily enough the LM555CM is not needed anymore, which is prob why this circuit didn't make sense, as it wasn't doing anything purposeful due to the missing connection between it's pin 1 and C1 and R3 (see image above, light blue line). No idea why it's being populated still..

The only thing that definitely is wrong is the VE.Direct FAQ and it's suggestion to be able to supply 12V on the DTR and/or RTS pins from a normal power source. The diode D6 and then further the ADuM are incapable of dealing with this.
You should not supply more than 6V to either DTR and/or RTS if you want to keep your cable functioning.
That a RS232 interface with it's +/-15V works is due to D6 being able to suck up some heat as the source is high ohmic and not a lot of current comes out of it.
But any other power source with more than 6V will overwhelm D6 and then kill the ADuM chip.

PS: also found a drawing mishap, D2 the diode from 1-3 was wrong in the older drawing.

PPS: we're designing our own level shifter/isolator circuit now. It will be driven by +12V (as the Ethernet<>RS232 converter needs a power supply anyway). If there is interest I can drop the ecad files once ready here or on one of the git places.

This is the pcb board from which above circuit is from..