Better datalogging using the stock ECU

[S2central.net]

Properly mad stuff Dmitri - well done that man...

[prj]

Very impressive, can you tell me what the little black box on top of the ECU is please? I'm guessing it must be the emulator you use for tuning.

Yeah, that's the emulator.
Moates Ostrich 2.0

[prj]

Properly mad stuff Dmitri - well done that man...

You've got mail Sir

[Val]

2) The fastest I can read is on average one sample every 210 milliseconds on Group 0. So that's 4.76 samples per second.
A bit faster than VAG-COM. It gave me 4.3-4.4 samples per second.
On group 1+, it is about 228ms, so about 4.39 times per second.

Is this with 1 or 3 groups (by 4) logged in the same time?
Anyway, looking good

[prj]

Is this with 1 or 3 groups (by 4) logged in the same time?
Anyway, looking good

No, I am logging only group 8 in this instance.
I re-wrote in assemblly inside the ECU what it shows in group 8, so I can get 12 different values into one group at the same time, instead of just 4.

You need a custom KWP 1281 implementation for that though (which I wrote in windows).
This entire system consists of 2 parts:
1) ECU software
2) Windows computer software

One can not function without the other.

And because the baud rate is so low and due to KWP 1281 limitations, you can't really log more than one group. You'll get an update rate of 1hz, which is useless.

[prj]

An example with some parameters.


There are more loggable parameters of course, but no point to put them all onto the same picture.

[prj]

Revisiting this because the KWP1281 protocol has a lot of overhead, is slow, and when logging you don't really need a "TCP" style protocol, more a "UDP" fire and forget style protocol... so all the data integrity and handshaking does not create huge demands.

The problem is that sometimes when mapping / tuning KWP1281 disconnects. Be it VAG-COM itself or my custom logging driver.
There is some noise or error on the K-Line and then connection just drops...
Then it takes ages to re-establish the connection, and obviously your pull is ruined.

So what I did is, I made a virtual switch inside the ECU's software (can be made to a hardware switch), which either lets normal diagnostic routines run, or my special routine.
This special routine, when activated, replaces the entire monstrous ECU diagnostics routine and simply spams the K-Line with data as fast as possible.

I used a simple protocol with two sync bytes, and then data, so the client can sync onto the data stream.

I need to field test this (so far works only on the bench), but I am getting speeds of around 50 samples/second.
This compares very favorably to 4.2-4.5 samples/second I got before on the bench.
Not to mention, that if a byte is dropped or mis-received, just one or two samples will be lost, as the ECU will just keep going and the client will re-sync to the data stream.
Pictures coming soon (tm).

[Nuvo]

That sounds like a very impressive development. Good luck with the field testing.

[EspenW]

50 samples per second sounds awesome and will make the data much better suited for things like the Virtual Dyno program. I seem to get an average of 3 samples per seond out of the standard Vagcom, and it drops as low as 2 samples per second at high rpm.

[prj]

50 samples per second sounds awesome and will make the data much better suited for things like the Virtual Dyno program. I seem to get an average of 3 samples per seond out of the standard Vagcom, and it drops as low as 2 samples per second at high rpm.

I knew you would be interested
If all is successful, I might patch your tune someday.

It is pretty easy to make some activity in VAG-COM kick it into this special mode, but to get it out of there would require restarting the car.

I just change a byte in the binary with my emulator, which toggles between normal diagnostics mode and fast logging mode.

[Nero]

Just had a first look at this topic.

Really like it. Quite impressed again with our Motronic and your skills!

[prj]

Routine that populates the buffer in all of it's glory:

  1 $NOMOD51
  2 $INCLUDE(80C515.MCU)
  3 ; Entry point, 90 85 22 is replaced.
  4 MOV DPTR, #08700h
  5 CLR A
  6 MOVC A, @A+DPTR
  7 JZ valuespam
  8 ; If the byte is not altered, do normal logic
  9 MOV DPTR, #08522h
 10 LJMP 555Ah
 11 ; If byte is altered spam...
 12 valuespam:
 13 ; Set up transmit mode to 9 bit UART with 9600 BAUD
 14 MOV PCON, #080h
 15 MOV SCON, #0E8h
 16 MOV P2, #00h
 17 
 18 ; Check if a buffer transmit op is already in progress
 19 JNB IEN0.4, fillbuffer
 20 ret
 21 
 22 fillbuffer:
 23 ; Leave room for length
 24 MOV R0, #0AFh
 25 ; 1, 1st magic sync byte of 0xFEFD
 26 INC R0
 27 MOV A, #0FDh
 28 MOVX @R0, A
 29 ; 2 - RPM, 0x3A
 30 INC R0
 31 MOV A, 03Ah
 32 MOVX @R0, A
 33 ; 3 - LOAD, 0x3F
 34 INC R0
 35 MOV A, 03Fh
 36 MOVX @R0, A
 37 ; 4 - Voltage, 0x36
 38 INC R0
 39 MOV A, 036h
 40 MOVX @R0, A
 41 ; 5 - IAT, 0x37
 42 INC R0
 43 MOV A, 037h
 44 MOVX @R0, A
 45 ; 6 - Coolant Temp, 0x38
 46 INC R0
 47 MOV A, 038h
 48 MOVX @R0, A
 49 ; 7 - TPS, from ADC
 50 INC R0
 51 MOV DPTR, #0BE00h
 52 LCALL 017CFh
 53 MOVX @R0, A
 54 ;8 - Timing, from memory
 55 INC R0
 56 MOV R1, #05Fh
 57 MOVX A, @R1
 58 MOVX @R0, A
 59 ;9,10 - IPW, from memory
 60 INC R0
 61 MOV A, 05Eh
 62 MOVX @R0, A
 63 INC R0
 64 MOV A, 05Fh
 65 MOVX @R0, A
 66 ; 11 - Vehicle speed
 67 INC R0
 68 MOV P2, #01h
 69 MOV R1, #070h
 70 MOVX A, @R1
 71 MOV P2, #00h
 72 MOVX @R0, A
 73 ; 12 - Maf table number, memory
 74 INC R0
 75 MOV R1, #0E8h
 76 MOV A, @R1
 77 MOVX @R0, A
 78 ; 13 - Maf table cell, memory
 79 INC R0
 80 MOV R1, #0EAh
 81 MOV A, @R1
 82 MOVX @R0, A
 83 
 84 ; Add length
 85 MOV A, R0
 86 SUBB A, #0AFh
 87 MOV R0, #0AFh
 88 MOVX @R0, A
 89 
 90 ; Initialize transmission
 91 SETB IEN0.4
 92 MOV SBUF, #0FEh
 93 
 94 RET
 95 END

I wonder how many people on this forum can read this

[Rocco]

Routine that populates the buffer in all of it's glory:

  1 $NOMOD51
  2 $INCLUDE(80C515.MCU)
  3 ; Entry point, 90 85 22 is replaced.
  4 MOV DPTR, #08700h
  5 CLR A
  6 MOVC A, @A+DPTR
  7 JZ valuespam
  8 ; If the byte is not altered, do normal logic
  9 MOV DPTR, #08522h
 10 LJMP 555Ah
 11 ; If byte is altered spam...
 12 valuespam:
 13 ; Set up transmit mode to 9 bit UART with 9600 BAUD
 14 MOV PCON, #080h
 15 MOV SCON, #0E8h
 16 MOV P2, #00h
 17 
 18 ; Check if a buffer transmit op is already in progress
 19 JNB IEN0.4, fillbuffer
 20 ret
 21 
 22 fillbuffer:
 23 ; Leave room for length
 24 MOV R0, #0AFh
 25 ; 1, 1st magic sync byte of 0xFEFD
 26 INC R0
 27 MOV A, #0FDh
 28 MOVX @R0, A
 29 ; 2 - RPM, 0x3A
 30 INC R0
 31 MOV A, 03Ah
 32 MOVX @R0, A
 33 ; 3 - LOAD, 0x3F
 34 INC R0
 35 MOV A, 03Fh
 36 MOVX @R0, A
 37 ; 4 - Voltage, 0x36
 38 INC R0
 39 MOV A, 036h
 40 MOVX @R0, A
 41 ; 5 - IAT, 0x37
 42 INC R0
 43 MOV A, 037h
 44 MOVX @R0, A
 45 ; 6 - Coolant Temp, 0x38
 46 INC R0
 47 MOV A, 038h
 48 MOVX @R0, A
 49 ; 7 - TPS, from ADC
 50 INC R0
 51 MOV DPTR, #0BE00h
 52 LCALL 017CFh
 53 MOVX @R0, A
 54 ;8 - Timing, from memory
 55 INC R0
 56 MOV R1, #05Fh
 57 MOVX A, @R1
 58 MOVX @R0, A
 59 ;9,10 - IPW, from memory
 60 INC R0
 61 MOV A, 05Eh
 62 MOVX @R0, A
 63 INC R0
 64 MOV A, 05Fh
 65 MOVX @R0, A
 66 ; 11 - Vehicle speed
 67 INC R0
 68 MOV P2, #01h
 69 MOV R1, #070h
 70 MOVX A, @R1
 71 MOV P2, #00h
 72 MOVX @R0, A
 73 ; 12 - Maf table number, memory
 74 INC R0
 75 MOV R1, #0E8h
 76 MOV A, @R1
 77 MOVХ @R0, A
 78 ; 13 - Maf table cell, memory
 79 INC R0
 80 MOV R1, #0EAh
 81 MOV A, @R1
 82 MOVХ @R0, A
 83 
 84 ; Add length
 85 MOV A, R0
 86 SUBB A, #0AFh
 87 MOV R0, #0AFh
 88 MOVX @R0, A
 89 
 90 ; Initialize transmission
 91 SETB IEN0.4
 92 MOV SBUF, #0FEh
 93 
 94 RET
 95 END

I wonder how many people on this forum can read this

you wonder? you are mad and crazy Dmitri it looks Japanesse for me

[Nuvo]

I wonder how many people on this forum can read this

I can follow most of it, although I'd need to dig out the reference manual for a few processor specifics. I know x86 assembler better.

[prj]

I can follow most of it, although I'd need to dig out the reference manual for a few processor specifics. I know x86 assembler better.

Nice. I started with x86 as well, writing hacks for online games.
Processor is a 80C535. Open IDA Pro and get cracking, there is lots to do... hehe.