12 Volts Where There Shouldn't Be Any
Part 1: Tuesday Evening — Voice Input, a Van Loom, and a Nebula
Two weeks ago I started repairing a friend's VW T5 Transporter. Locked keys, drilled lock, two garages that made things worse. The loom had been cut through. Both window mechanisms were broken. I was building one working door from two broken ones, splicing the loom wire by wire — clean solder joints, heatshrink on every connection.
That work is done. The loom is repaired. The door is mechanically rebuilt. But the electrics still won't talk. Windows dead. Central locking dead. The door module is invisible to the rest of the van.
Time to stop soldering and start diagnosing.
What Delphi Said
With the physical repair done, I plugged in the diagnostic laptop. Three fault codes told the whole story:
| Code | Description | Status | |------|-------------|--------| | 01336 | CAN bus, comfort | No communication — Intermittent | | 01330 | Convenience control unit | No communication — Intermittent | | — | Driver door module | No communication — Permanent |
Code 01336 is the interesting one. It means the comfort CAN bus had dropped to single wire mode — the bus normally uses a twisted pair (CAN-High and CAN-Low), but it can limp along on just one wire and ground if the other is damaged. That's the "intermittent" — it was clinging on, then giving up entirely.
The driver door module showing permanent loss of communication means it's completely off the bus. Dead to the network.
Following the Wires
On a T5, the comfort CAN bus runs on two wires:
| Signal | Wire Colour | Dash Connector Pin | |--------|------------|-------------------| | CAN-High | Orange/Green | Pin 27 | | CAN-Low | Orange/Brown | Pin 28 |
These run from the J393 convenience control unit (behind the lower dash panel, driver's side, above the pedals) through the body harness, through the rubber boot gaiter at the A-pillar, and into the door to the door control module.
I unplugged the door module connector and probed the body-side pins with the multimeter.
First test — continuity across the CAN lines. The meter went to AUTO and sat there. Open line. No path through. That confirmed the break.
Second test — DC voltage on the power pin. 11.74V. Good — the 12V supply through the loom was intact. The power wire survived whatever happened. The thinner CAN wires didn't.
Third test — and this is where it got interesting — DC voltage across the CAN pins. 11.74V.
That's not right. On a healthy CAN bus with the ignition on, you'd see roughly 2V between CAN-High and CAN-Low. Seeing full battery voltage means one of the CAN lines is being pulled to 12V. That shouldn't happen. Ever.
Pulling the J393
The J393 convenience control unit lives behind a trim panel above the driver's pedals — four T20 Torx screws and the panel drops down. The module is a black box made by Delphi, part number 7H0 937 049 AA.
I pulled it out, cracked the case, and inspected both sides of the PCB. No burn marks. No charred traces. No obviously blown components. The board looked clean.
But here's the test that mattered: with the J393 unplugged, I went back to the door connector and measured across the CAN pins again.
0V.
The 12V was gone. J393 was the source. It was pushing battery voltage out on the CAN data lines through a fried CAN transceiver. The chip had failed internally — latched up, shorting an internal power rail to the CAN output. No visible damage on the outside. Just quietly dumping 12V into a bus that runs on 2.
The Real Story
This is the part where the backstory changed.
In the first post, I wrote "pay once or pay three times" — blaming the garages for cutting the loom and breaking the window mechanisms. That's the story I was told. Two weeks of splicing wires and rebuilding a door based on that story.
It wasn't the garage that cut the wiring loom. It was the owner. He'd attempted the door swap himself, cut through the loom — almost certainly with the ignition on — and the 12V supply wire arced against the CAN wires as the blade went through. Microseconds of contact. Enough to send battery voltage straight back down the CAN bus to J393 and kill the transceiver chip.
The garages never had a chance. They were trying to fix a door wiring problem, but the damage had already travelled upstream to the convenience controller behind the dash. You could solder every wire in that door loom perfectly — and I did — and the door module would still be invisible. Because J393 was flooding the bus with 12V.
What I Learned
-
Always measure voltage on CAN before assuming a wiring break. An open line and a shorted-to-power line produce similar symptoms (module offline), but the fix is completely different.
-
CAN transceivers can fail invisibly. No burn marks, no smell, no obvious damage. The chip just latches up internally and starts misbehaving. You'd never find it by looking at the board.
-
Disconnect upstream and re-test. The single most useful thing I did was unplug J393 and re-measure. That one test moved the fault from "somewhere in the loom" to "this specific module."
-
Get the real story. If you're told three people tried to fix something and all failed, ask what happened before the first person touched it. The actual cause is usually upstream of where everyone's been looking.
What's Next
A replacement J393 is on order — they're common on eBay for T5s. Once it's in, it'll need coding to the vehicle with Delphi (immobiliser settings, central locking config, window convenience features). After that, a full continuity check on the door loom end-to-end to make sure there's no secondary damage, then reconnect the door module and clear codes.
The loom repair wasn't wasted — those wires needed splicing regardless. But it was never going to be enough. The real damage was behind the dash the whole time.
Pay once or pay three times? This van paid four. And the person who caused it was the one telling the story.
Sometimes the fault isn't where everyone's been looking. It's where nobody thought to check.