Diagnostic Interfaces

[Rhinoman]

Originally Posted by jtgh

i understand the difficulty in finding the correct baud.
since the Uart is non-return-to zero format. (so a freq ,counter is no help, just a scope is best , measuring the start bit period with a trailing 0 data)

you must have keep sending the commands at all bauds until you found the sweat spot?
lots of patience ,that took.
thanks for doing that !

cheers

Fortunately I have an immobiliser so it wasn't too difficult to get an approx. data rate and work out the message format. The header byte and table location and length I got from the code. Sending random codes on mine wasn't a good idea because there is one message that will erase the immobiliser data.
I've extended the lookup table on mine so I can read the ports and the A/D converter result registers directly.

[jtgh]

Mr Rhinoman:
thanks again for your information.

Experiment 2; -56B40 ECU comm checks.

i have my test jig set up this morning (happy fathers day!) and it is running.
Megasquirt Ing. cas Stimulator, coil simulator I have all sensors.
on the bench I start my virtual engine. verrrooommm . LOL.

i tried +12vdc via 10k ohms to A3 Diagnositc req , pin,
scope on A20(serial data) slow Horz. sweep (not to miss anyaction)

results , stays at 5vdc all time, no wiggle of the UART.

tried with engine running (cas 1000 rpm) and with cas =0 KEYON no start.
no wiggle action form the usart.

I read that only many early ECU's broad cast (not mine) and the newer must be commanded, like you diagnostic page shows.


that being the case, then the baud rate is the key to the door.
I will next try all three using your command set.
160,8192 , and 15126.
I am curious what was the state of the other 3 pins, on your diag. page
A= A9 dutycycle
B= A3 diag req. 10k ohm pullup.
D= A7 test switch pin.

last: immobiliser.. ( wish i had schematic of it or how it interfaces)

Are you saying the Immobiliser ties in to the serial port of the ECU and you just
pulled the baud rate off it during its comms.? that was a good fortune.

I will buy one of those immobilisers next.

if i can get mine going , i am sure i can at least find and correlate your 00 to 1F list of parameters, I have pots simulating all my sensors. and can hold all fixed while varing any single param. that way I can ID it bugger.
and add it to your list.

now to make a serial cable adaptor ( bidirectional serial data type, TTL232)

cheers and thanks again for your help.
jerry

[jtgh]

first is my bench simulater.

next off AKmodules page. I will buy this.

http://www.akmcables.com/serial.doc


off to find an imobilizer unit.

[Rhinoman]

The immobiliser that I have is a factory immobiliser and needs an ECU with the correct code. It sits on the serial bus and exchanges codes with the ECU on power up, if the codes don't match then the fuel injection is disabled.
That cable should work OK although a 7805 is completely unsuitable for an automotive application without a good deal of extra protection.

[jtgh]

rats, custom ECU. ( i will try the 3 baud rates)

yes, that most 7805 is very unsuitable.
i will use the power supply same as my megasquirt or just a simple Zener diode.
Rs resistor, cap , zener and a very low MOV 10v , should be enough to drive max 232
and to handle max load of rs232.
if that dont work will just bipolar transistor driven from zener.


or copy my megasquirt ECU power pack.

[jtgh]

ALDL-names

here are the diagositc names for 95 sidekick and tracker 16v engine ECU.
it might help id any data streaming out.

[Rhinoman]

I've had a look at SilentBobs early 16V 33920-57B30 ECU. This is a 6800 based system with a 63B50 comms IC.
TX and RX clocks are running at approx. 125KHz, which is probably correct (2MHz/16). I am having problems setting up the Logic Analyser, I'll try it again tomorrow. 15625 seems very likely, I originally thought that the SDL was 15625 but I got more reliable comms a little slower. Since then I've revised my hardware and I've got new software on the way, I will redo my tests at some point.

[jtgh]

are you saying the actual uart RX clock is 125khz ?
8250 baud scaler (1 of 3 divisors)
1mgh divided by 64 is 15625, so that would correlate with your old measurements.
but 125khz div by 16 is 7812.5
and 125khz div by 64 is 1953.125

1mgh divided by 64 is 15625, so that would correlate with your old measurements.

I will try them all.

i will scope all my uP pins and try to find a symmetrical clock and then measure it.

i suspect the ECU's91 + use the same params ,for comms up to 95 (usa) , just a gut feeling by looking at what plug in packs are used on a GM tech1 , and a tracker.

cheers,and thanks for looking !!!


late edit, I found my clock to be 12mhz out of crystal into CPU.

[jtgh]

I now have my 91 8v ECU communicating , the live data !
thanks to you Rhinoman, the basic protocol was perfect.

I am at this moment reading my TPS values 02 to FF hex "cool"
as i turn my TPS pot (simulated car) it works perfect.

This is very exciting.
My setup:
1- ECU on bench and 12vdc supply
2- my ECU test jig with all sensor and pots for simulated sensors, baro, and all.
3- old Dell Laptop cpi pentium III-500 , ( any old cheap lap top)
4- special high speed hi accuracy com4 port (pccard/pcmcia) adaptor. $20
5- rs232 to TTL232 bidirectional serial adaptor from AKMcables. $30
6- the program call realterm ver2.0.0.57 $free

I will now correlate all or as many of the parameters in the ECU to my FSM
parameter page. ( the secret to that is my bench jig can freeze all params and allow me to very anything, 1 at a time (mostly).

when that is done a custom program that logs all this data in english.
after that maybe a dash board for the lap top showing all parameters live like the famous Crome program used on Honda's.

The baud rate is 15625 , which is 1,000,000 (master clock) divided by 64.
back when they made the first ECU , the CPU ran at 1mhz, so they just fed that direcly in to the Uart motorola 6850 and it only (a blessing) had 3 baud scalers, 1, 16 and 64 , and as you can see 1 and 16 are useless for shop scanners, so lordy ,they did 64 and the result is 15625.

I noticed this same number is used in the OBD2 protocal too, (legacy equipment supports that , so they used it. even today on new cars)

i will publish my list simular to this. ( not finished now)

ALDL-names


notes, my special rs232 card allows the illegal baud rates as does realterm.
a perfect match.
the "comes-with", IBM clone style DELL UART can NOT generate this queer baud. the ubiquitous N16550a is useless at 15625. Com1 and Com2.

One nice thing ,the realterm proggy will tell you , if the baud is legal.
just plug in the card, and try it, it will say," illegal setting to com1." if true.

OBD1 , for Suzuki, coming soon and free of charge. (cept hardware)

cheers.

[jtgh]

i plan to activate lean burn and this program running.
with my oxy liar box.

here is another man with a liar box.

YouTube - EFIE Gizmo

it just offsets the OXYgen sensor so you can run 15.5 to 1 AFR.
warning you need one more part , another circuit that monitors the TPS, wiper pin.
if the TPS is not moving, then 15.5 :1 if you are accerating then this liar box,drops out.
It must, and i dont think his does.

the hard part is detecting loads, hills and such, you do not want to drive up hills in lean burn mode, "melts things, like pistons"
so use it only for flat crusing. and save big money. some say 30% (40mpg?)

I have a MegSquirt that will do all that and do it all PERFECT !

but this fix is fun too.

cheers.
Lean burn has lots of down sides, I did not say them all here.
i am not selling anything so only facts and theories. no lying thru omission.
eg: melted prescious metal in catcon.
pinging might happen if you get to aggressive. Bad !

As you can see he has a seperate AFR meter and a Scanguage II ,monitoring live MPG and AFR. about $300 worth of hardware there. and common.

I like his demo because he shows , a real car running lean and the proof all at once.
5 stars for a great demo !!