ECU's, interfaces and baud rates

[pentium]

So for a few years I've been trying to get data out of my 92 Tracker's ECU to diagnose emissions problems. The first problem was that it predates OBD I being standardized and has its own proprietary interface which is basically a single wire bidirectional TTL-232 line running at 7812 baud. You have to build an adapter to attach a proper RS-232 cable to that and the schematics for such an adapter does float around the web.



You do need to make a few modifications however, notably upgrade the 78L05 with an L7805CV and a jumper to switch from either serial power or 12v power from the car but the adapter isn't rocket science to build.


Actually, this is the second version I've built. The first variant was so dangerous it blew up my first ECU.
So there IS a little bit of rocket science in here... >_>

Anyways I can test the adapter out on the bench and it seems to be working. Data from the laptop passes through and gets converted (and echos back because that's the way the schematic has things wired) and when the ECU data pin is attached we have a 4.1v constant MARK and on data transmission the adapter pulls the line between 4.1v and 0v to represent data.


That's me pressing the spacebar at 110 baud.

Thing is that when I start my monitoring software (an older version of RhinoView before he locked down the software to only work with his $160 interface) all it reports is bad data being received (and it will do this even when the ECU is not attached so god knows if it's not liking the echo or if the count in general is for whenever the ECU never responds to a command). That I can tell the ECU never transmits anything as-is and is waiting for a command to start streaming data.
I don't know if this is because of the UART in my CF-25 toughbook being incompatible with the oddball baud rate, the TTL-232 signalling not being properly formed, or a problem with the adapter itself.

[redwire]

Re: ECU's, interfaces and baud rates

Looks like a good build, second time around is always better Some ideas from my fuzzy head:

I would run a PC terminal program, like TeraTerm etc. and with echo off and no ECU connected, you should be able to see what you type in. Like a loopback test looking at the echo. You could scope the RS-232 lines, I usually put one of those 10uF charge pump caps in backwards.

The Rhinoview software might be looking for CTS or DSR. Try jumper DTR (DB9 pin #4) to DSR (DB9 pin #6) and CTS (DB9 pin #8) to RTS (DB9 pin #7). It might also test for an RS-232 echo (=converter connected) if it is finicky about connecting the converter then ECU after launch, or looking for an RS-232 idle line all SPACE instead of the usual MARK -ve.

I wouldn't expect software errors with nothing connected. Then I would look at your laptop, OS and the Rhinoview software if is uses COM DLL's for 16550 UART access. Win 7 has the HAL and many options for running in compatibility mode if you can try the software on another PC.

[stj]

Re: ECU's, interfaces and baud rates

have you looked at the ELM website?
there is loads of data on voltages,protocol and signaling.

i hope you know thatit may be pwm encoded.

[pentium]

Re: ECU's, interfaces and baud rates

I would run a PC terminal program, like TeraTerm etc. and with echo off and no ECU connected, you should be able to see what you type in. Like a loopback test looking at the echo. You could scope the RS-232 lines, I usually put one of those 10uF charge pump caps in backwards.

That is what you see in the scope image above. The returned echo to the terminal means that data is accurately passing through the adapter and then passing back to the terminal.

The Rhinoview software might be looking for CTS or DSR. Try jumper DTR (DB9 pin #4) to DSR (DB9 pin #6) and CTS (DB9 pin #8) to RTS (DB9 pin #7). It might also test for an RS-232 echo (=converter connected) if it is finicky about connecting the converter then ECU after launch, or looking for an RS-232 idle line all SPACE instead of the usual MARK -ve.

That I can tell for the first part of this, Rhinoview is pretty indiscriminate about what the DTR, CSR, RTS and CTS lines are doing. I can hang a scope off when everything is attached and I can see it trying to do something but it happens too quickly for any of my analog scopes to see properly. It WILL however immediately complain if you try to start the tool without the adapter plugged in (so I guess it only checks for a loopback and never checks again to see if the adapter is still present)

My toughbook runs Windows 98se so a lot of the UART still I believe is pretty hardware oriented. As for what Rhinoview does I can't get a lot out of the developer. He pulled or removed a lot of information regarding what the software does after paid a bunch of money to get access to the Suzuki protocol documentation. I assume he did that so people can't reverse engineer the software and cause legal problems.

[stj]

Re: ECU's, interfaces and baud rates

the interface looks like shit, he should have used the newer max chips or a HIN232 with .1uf ceramic caps and a 7407 buffer instead of all the transistors.

[pentium]

Re: ECU's, interfaces and baud rates

Hey, I can't say I'm skilled enough to completely review the design and totally build it better. I had a schematic, I had some knowledge on what SHOULD be modified and I tried. When I rebuilt the adapter I recycled almost everything but two blown diodes and a fake MAX232(and yes I was testing components prior to reusing them). I wasn't willing to wait several more weeks just for parts to roll in from china.

[stj]

Re: ECU's, interfaces and baud rates

fake max232 is good.
max232 from maxim was very old, it uses 10uf electrolytic capacitors for the oscillator stepup system

the newer maxim chips and the "clones" by texas-instruments, harris etc use a faster oscillator and only need .1uf ceramics to run.

i try to only use harris HIN232's in my designs.
they are pin compatable other than the smaller caps.

even better, use max3232's
they use .1uf caps and they are dual-voltage - they run from 3.3v or 5v

you can even get them ready-built for less than the cost of parts!!!
http://www.ebay.co.uk/itm/271645198031

[budm]

Re: ECU's, interfaces and baud rates

The circuit does work when it was built in 1995!, you can update all the parts you want but it is not going to fix the protocol/communication problem. It is like saying you have to use Windows 8 because it is newer than win 98. I still use lots of old equipment with old interfaces in my labs here at work, you are only dealing with 7~8K BAUD rate.
If you do the search 'ALDL TO PC INTERFACE' there are lots of info out there.

[stj]

Re: ECU's, interfaces and baud rates

i was just thinking that it's using far more parts than it needs to.
more parts = more chances of a mistake.

and of course we all have max chips or compatables - they are in all settop-boxes.
i must have over 50 in my spares i salvaged.

i'm posting my stocklist because it references the voltage and capacitor size needed against each part - some may find it handy when using salvaged chips.
forget the left column - it's just the number i have.

--------------------------------------
level converters

1 x hin232cp 16pin dip 1u 5v
1 x icl232ipe 16pin dip 1u 5v
2 x max220cpe 16pin dip .1u 5v
1 x max232cpe 16pin dip 1u 5v
1 x max3223cpp 20pin dip .1u 3v3
1 x max233cpp 20pin dip int 5v

1 x max232eewe 16pin wide-soic 1u 5v
1 x max232cwe 16pin wide-soic 1u 5v
1 x max3232cwe 16pin wide-soic .1u 3v3

2 x adm237ljr 24pin soic 1u 5v
2 x hin237cb 24pin soic 1u 5v
1 x max211cwi 28pin soic .1u 5v

2 x hin202cbn 16pin soic .1u 5v
1 x st202ec 16pin soic .1u 5v
1 x st232c 16pin soic 1u 5v
2 x sp202ect 16pin soic .1u 5v
2 x sp232ecn 16pin soic 1u 5v
4 x max232(ti) 16pin soic 1u 5v
1 x max232a 16pin soic 1u 5v
1 x max202e 16pin soic .1u 5v
6 x max202 16pin soic .1u 5v

5 x hin207eca 24pin sop .1u 5v
1 x st207ec 24pin sop .1u 5v
1 x max207cag 24pin sop .1u 5v
2 x max3221cae 16pin sop .1u 3v3
1 x mp243ec 28pin thin-sop .1u 5v
3 x max3243cai 28pin sop .1u 3v3
1 x max3243ecai 28pin sop .1u 3v3
1 x max3238cai 28pin sop .1u 3v3
1 x max211cai 28pin sop .1u 5v
1 x max213ecai 28pin sop .1u 5v

--------------------------------------

[stj]

Re: ECU's, interfaces and baud rates

found this.
http://www.fixkick.com/hacking/hacke...ap-hacked.html

[pentium]

Re: ECU's, interfaces and baud rates

God dammit Badcaps not telling me of new posts.

That adapter pictured near the very bottom of that page has basically all the components of my adapter, but in a different form factor.

I'm confused though on what you were saying stj.

fake max232 is good.
max232 from maxim was very old, it uses 10uf electrolytic capacitors for the oscillator stepup system

the newer maxim chips and the "clones" by texas-instruments, harris etc use a faster oscillator and only need .1uf ceramics to run.

So, swap back in the fake MAX232?

Same for you budm.

The circuit does work when it was built in 1995!, you can update all the parts you want but it is not going to fix the protocol/communication problem. It is like saying you have to use Windows 8 because it is newer than win 98. I still use lots of old equipment with old interfaces in my labs here at work, you are only dealing with 7~8K BAUD rate.

So, it WILL work, but we need the right baud rate and that can be done? You mean that it's possible my UART is incompatible?

i was just thinking that it's using far more parts than it needs to.

The other common design I saw was terrifying.
*Linked, because Tripod is a butch.*
There's little to no protection to my lines should something go wrong.

[stj]

Re: ECU's, interfaces and baud rates

what part and make is the fake 232?

what i'm saying is you need to use the right caps for a specific chip.
some want 10uf electrolytics, some want 100n ceramics!!
some dont need the caps, but those cost $$$ and i dont have one.

[stj]

Re: ECU's, interfaces and baud rates

btw, i found a page that says the data lines must be 1.4v - that's a bitch to get right.

[pentium]

Re: ECU's, interfaces and baud rates

Well it says MAXIM in a font different from all me real MAX232 chips and just like all my others it's just a MAX232CPE. I'm slightly curious though why these capacitors would not work but I guess usingng .01's for 10uf might cause problems. Fine, I'll order in some proper caps.

[stj]

Re: ECU's, interfaces and baud rates

cant you scavange some .1u ceramics?

[pentium]

Re: ECU's, interfaces and baud rates

No. My current living conditions meant that I can't stockpile PCB's for depopulating. I have to order new and wait a month for parts to ship in.

[pentium]

Re: ECU's, interfaces and baud rates

BuMP for closure.

I finally dug up a USB to serial dongle that uses an FTDI FD232BL chipset. This is in the same family as the chip used in the official RhinoPower USB adapter so once the drivers were loaded it was plugged into my serial adapter and I was able to successfully connect to the ECU.

[joshnz]

Re: ECU's, interfaces and baud rates

BuMP for closure.

I finally dug up a USB to serial dongle that uses an FTDI FD232BL chipset. This is in the same family as the chip used in the official RhinoPower USB adapter so once the drivers were loaded it was plugged into my serial adapter and I was able to successfully connect to the ECU.

good work.

[srhofmann]

Re: ECU's, interfaces and baud rates

The reason that the cable works is that it contains an internal clock generator. Most FTDI chipsets allow for a fractional divisor of the clock rate allowing more "legal" baud rates. Most older computer UARTs use an external clock and integer only divisors. The chipsets, drivers and user interfaces weren't smart enough to say, that's an invalid baud rate. Valid clock rates can easily be determined by taking the highest baud rate and dividing by integers until you get close to your target baud rate. If you want to eliminate the USB cable you could try setting a clock rate of 7680 on your laptop, assuming the built in RS232 port supports 115200. Any speed within +- 5% should probably be tolerated.

[budm]

Re: ECU's, interfaces and baud rates

What are you using for that GUI to show parameters on your laptop?