Soko,

not sure if the standard cable connects pin 1 on the OBD port side, but the DB9 connector has GEP_7 on pin 1 (see OVMS schematics). So you can route that to an EGPIO input (take care to add a voltage divider on the way). So that would need a simple hardware extension / modification, but is an option.

Sorry, can't help you on the VCDS stuff.

In my scan example, I initially was going to make an artificial example, then changed to using my Twizy live responses as an example. So that should have been 79b instead of 700.

Regarding typing the command 256 times… you know we've got a powerful scripting system on board? ;-)

Of course you can also write a C++ function to do the same by calling the CAN API directly. You just cannot use the polling framework, as that's bound to specific rx IDs. But you can simply read all CAN frames and look for responses without filtering by ID first.

A general purpose OBD scanner working that way could well be a valuable extension to the RE toolkit.

Regards,
Michael


Am 31.07.20 um 17:40 schrieb Soko:

Heya,

I'm back from my 2 hr VCDS session... not satisfying at all...

Car-off polling issue:
VCDS lists me 19 ECUs (with the ignition on) which includes 09 "Zentralelektrik" (which Michael recommended), 19 "Diagnosteinterface" and 61 "Batterieregelung" which stay responsive for a period of time after ignition is off. All other are not responsive no more with no ignition...
But: Once the car is locked not a single ECU is responding :( . So there is nothing to poll which will respond in any state of the car.
The only option is the +12V on pin 1. Is pin 1 somehow connected to OVMS via the standard OBD adapter?

Sniffing VCDS (+CarScanner App) commands:
Here I'm somehow confused now. I've tried to sniff the header/txid of ECU 61 (Betterieregelung) as the list of sharkcow hasn't it listed yet. It seems that the header is not transmitted, neither in VCDS nor in CarScanner. Only the PID goes to the OBD port. To understand the transmissions I've tried in CarScanner to swap between two dashboards/value-readings. One for SoC (=22 02 8C) of ECU 8C (header=7E5) outside-temp (=22 F4 46) of ECU 01 (header=7E0).
I couldn't find any 7E5 or 7E0 in the sniffed data!? Why? How is the ECU addressed then? I've also think of any rule to get from the ECU number (i.e. 8C, 01) to the header, but couldn't find any...

Scan for OBD devices:
Thanks for the info & details Michael. But is there a chance that I'll find an OBD device which VCDS doesn't know about (or doesn't list in the software)?
If yes: Can I somehow put this manual request in C++ code instead of using the shell? Then I don't have to type this manually 256 times (by using a loop) ;)
Also, is there a typo in your email? You write "...sent to ID 700, type..." but the command says "79b" and later you write "...there is a device listening at 79b...". So is ID 700 the device 79b or did you mean ID 79B?

thx for you patience with me

Soko

On 31.07.2020 14:09, Michael Balzer wrote:
Soko,

you don't need to use the poller to scan for OBD devices. You can simply do manual transmissions and look for responses.

To send a manual request, simply use the can tx command. Example:

OVMS# can can1 tx standard 79b 02 01 00 00 00 00 00 00

This means: 2 byte OBD request sent to ID 700, type 01 = get current data, from PID 00. PID 00 is supposed (OBD-II) to reply with a four byte bitset of the device's PIDs 01-20 (hex) support.

Send this with can logging enabled without filters, so you can see all responses. A response could look like this:

V (591792) canlog-monitor: 1596194327.716686 1R11 7BB 03 7f 01 11 00 00 00 00

This tells you:
OK, let's try to change the session (command 10) to the ISO 14229 standard "extendedDiagnosticSession" (value 03):

V (2085712) canlog-monitor: 1596195821.642256 1T11 79B 02 10 03 00 00 00 00 00
 V (2085722) canlog-monitor: 1596195821.659737 1R11 7BB 03 7f 10 12 00 00 00 00

No luck. You can now begin trying all possible session codes. Or, in my case, I already know the device uses C0 ;-)

V (2095662) canlog-monitor: 1596195831.598560 1T11 79B 02 10 c0 00 00 00 00 00
 V (2095672) canlog-monitor: 1596195831.609308 1R11 7BB 02 50 c0 00 00 00 00 00
We're logged in -- nice! :-)

Let's try the PID read again (do this within 60 seconds or the session will log out, unless you send a "tester present" frame):

V (2530842) canlog-monitor: 1596196266.780867 1T11 79B 02 01 00 00 00 00 00 00
V (2530862) canlog-monitor: 1596196266.797961 1R11 7BB 03 7f 01 11 00 00 00 00

Still no luck. So this tells you, the device does not support standard OBD-II PIDs. How about a UDS DTC request (19)?

V (3034142) canlog-monitor: 1596196770.085839 1T11 79B 02 19 01 ff 00 00 00 00
V (3034162) canlog-monitor: 1596196770.103517 1R11 7BB 03 7f 19 12 00 00 00 00

No luck. Let's try command 21 (read enhanced data by 8 bit PID) at PID 00 next:

V (2737632) canlog-monitor: 1596196473.567437 1T11 79B 02 21 00 00 00 00 00 00
V (2737712) canlog-monitor: 1596196473.649927 1R11 7BB 03 7f 21 12 00 00 00 00

Nope… maybe PID 01?

V (2744182) canlog-monitor: 1596196480.120911 1T11 79B 02 21 01 00 00 00 00 00
V (2744262) canlog-monitor: 1596196480.203461 1R11 7BB 10 34 61 01 13 88 13 83

Houston, we've got a response :-)

You would now check if this response is a single or multi frame type, then try to get all data and then try to map the data to something you would expect to be stored in that device.

For the standard UDF commands, you will need a copy of the ISO 14229 standard.

Regards,
Michael

_______________________________________________
OvmsDev mailing list
OvmsDev@lists.openvehicles.com
http://lists.openvehicles.com/mailman/listinfo/ovmsdev


-- 
Michael Balzer * Helkenberger Weg 9 * D-58256 Ennepetal
Fon 02333 / 833 5735 * Handy 0176 / 206 989 26