Technical Anyone had any success with an ELM323/327 interface

[thunderfox5]

Mine seems to be the Bosch ME7.2.1 Euro 2 version...

The program is currently being written in C#, like my Car PC interface. I tested it this afternoon for the DTCs, there's a video here:



It's in portuguese though, but looking might prove interesting!

 

[danhans115]

That looks very cool! language isnt a problem most things I find interesting seem to be written in German and with some websites I tend to get the jist of it.

Ive written something very simple in VB .net. I started in that language as Ive been using BASIC for years and I have a single board computer that runs Windows CE5, which I am going to fit inside the chassis of a connectnav+ unit. I just need a connectnav chassis now...

The ECU in my car is Bosch ME731H4, so the requests should be the same. Im just waiting for my wife to get home so I am allowed out

 

[_Dejan_]

Im start on "OBDII" plugin for CarPC application which will be writen only for stilo, because stilos are mostly not OBD compatible... But I will use CAN protoccol and not ELM...

 

[danhans115]

Ive just tried doing the ReadLocalDataByIndentifier, and it works but havent had much luck with the interpretation.

Ive beeen using some of the definition files from sanderqv's unidiag program (which does what we have been discussing with a simple level shifter serial circuit).

The prob is the scaling... for example

I used the ME731 definition as a reference (attached at bottom).

You send

>21 03 (odometer)

and recieve

61 03 11 03 9A

strip of the stuff we arent interested in and the definition file tell us that 3 bytes are recieved, and the scaling is X*0.1 = Kms

So for my car above 11 03 9A = 1115034 * 0.1 = 111503kms ( about 69284 miles), but my car had 80,510 miles on the clock.

Looking at the other unidiag definition files the scaling differs significantly between applications but the requests do not. They are grouped together by ISO code and neither of our cars seem to be covered by what definition files there are.

I suppose you could use your odometer reading and work out an approximate scaling yourself, but it wouldnt help you with the others.

My maths isnt exactly fantastic so I might have the calculation wrong, but it said unsigned and I treated it as so.

any ideas?

Dan

 

[danhans115]

Im start on "OBDII" plugin for CarPC application which will be writen only for stilo, because stilos are mostly not OBD compatible... But I will use CAN protoccol and not ELM...

Most diesel ones arent yes (05- on.. in most cases hehe), but all petrol stilos are EOBD compliant.

 

[thunderfox5]

Yeah I knew that program from Sander... He's been doing an excellent job it seems! The reason why I opted out of using that cable (the KKL interface) is that the ELM is one of the most mainstream applications out there! Most people will definitely be using it for diagnosis, instead of the KKL, and there are absolutely no solutions for it, apart from the limited and expensive ItaliaCheck.

Also, Sander seems to be going on a "trial and error" basis to figure out things such as ECU specs, the scaling, the sensor data... Trial and error should be avoided, and a standard should be found! Makes me wonder how the FIAT testers work all the same, and SanderQV's program doesn't. There must be a key to this riddle (and certainly doesn't rely on the uncontrolled KKL interface).

I'm pretty happy we've made it this far! I'll try that command for my Punto right now, and see what it does, then we can figure out a workaround for it by means of comparison (the Punto's odometer marks in kilometres).

 

[danhans115]

The KKL is very difficult to code easily for unlike the elm, where you dont have to worry about the timing and quirky changes in baud rates. The flexible ELM327 with its commands makes home programming very simple.

As for standards, well I dont think we will find one much beyond we have. At the point you we are at you are using kwp2000, in which Fiats used as set out in that document we are referring to, beyond that it was up to them what was what.

I would have thought the examiner will read the ISO code from the ecu (1A 97 etc) and use the correct scaling parameters accordingly for that application.

The documentation is available to certain developers like Bosch, Snap-on etc for large amounts of money so they can develop their own equipment. So the best way would be to log communication beween the vehicle and some of their equipment. Ive do not have any contacts who would allow me to do anything like that, although I did think about doing it secretly.

The more information that can be out in the open the better... something with features like vagcom for the fiat would be fantastic.

Dan

 

[thunderfox5]

Okay, just gave it a go using the ME7.2.1 definitions... Doesn't seem to make much sense out of some, makes more sense out of some others...

Here's what I got, and what I interpreted:

18 00 FF 00    # read DTC
reported:    58 00
meaning:    No DTCs

21 03        # odometer km (about 153000)
reported:    61 03 03 1F BC
meaning:    20473,2 km - X*0,1 - wrong data

21 38        # airflow rate (not known)
reported:    61 38 00 1B
meaning:    2,7 kg/h - X*0.1 - ???

21 36        # engine load (idle)
reported:    61 36 04 AE
meaning:    28,0% - X*0,75/32 - believable

21 3B        # vehicle speed (0 km/h)
reported:    7F 21 12
meaning:    Negative response? Because it was stopped?

21 30        # engine speed (about 800)
reported:    61 30 03 52
meaning:    212,5 - RPM X/4 - just X would be believable (850 rpm)

Logs and txt:

From what I've just learned, the 7F 21 12 response to the vehicle speed request, according to table 5.2.2 of the KWP2000 specification, means Sub Function Not Supported. That doesn't make any sense, speed should somehow be supported I mean, even the dashboard relies on this function...

 

[thunderfox5]

Seems I don't need to create an error in my car anymore... I got a treat as I updated my software:




Still don't get any sensor data though... This sucks

 

[SanderQV]

Hello All,

Some remarks:

Yeah I knew that program from Sander... He's been doing an excellent job it seems!

Thank you .

The reason why I opted out of using that cable (the KKL interface) is that the ELM is one of the most mainstream applications out there! Most people will definitely be using it for diagnosis, instead of the KKL, and there are absolutely no solutions for it, apart from the limited and expensive ItaliaCheck.

The KKL cable gives full control, but the timing is (very) difficult.

Also, Sander seems to be going on a "trial and error" basis to figure out things such as ECU specs, the scaling, the sensor data... Trial and error should be avoided, and a standard should be found! Makes me wonder how the FIAT testers work all the same, and SanderQV's program doesn't. There must be a key to this riddle (and certainly doesn't rely on the uncontrolled KKL interface).

Why do you think it is 'trial and error'. Almost all information in my .def files comes directly from the examiner software, so I would assume it is correct...

I'm pretty happy we've made it this far! I'll try that command for my Punto right now, and see what it does, then we can figure out a workaround for it by means of comparison (the Punto's odometer marks in kilometres).

Considering the Odometer, the difference between the dashboard and motronic odometer can be upto 15%, because of different calibration.

If people are interested in .def files for certain types of Alfa/Fiat/Lancia, one could always ask in the unidiag yahoo group (I don't read this forum regularly)....

Kind regards, Sander.

 

[thunderfox5]

Hey Sander! An honor to hear from you "in person"!

I have recently joined the Unidiag group and have been examining those def files in order to figure out how my Punto works. So far I seem to be having some success, based on the EDC15 and Bravo definitions, definitely made some big progress referring to those files!

I was suspecting the .def files' formulas were calculated on a trial and error basis because of that... Some values aren't coincident with the real values on the car, however it helps a lot to know that they came from the testers, this way I know that they are correct for certain!

Congrats on your work, man! Wish me luck for my punto as well

 

[thunderfox5]

Hey look! RPM are working

 

[danhans115]

Ive made a lot more progress with my abarth, most of the scalings are the same on the ME731 as the ME31 so I can use the values from that.

 

[Mike63]

I think it's been established that Gendan make one of the better tools:
http://www.gendan.co.uk/product_ELM327.html
Anyone tried a Chinese copy?
http://cgi.ebay.com/ebaymotors/ws/e...0201239692&viewitem=&sspagename=STRK:MEWAX:IT
Since ELM want $33 for their chip it seems unlikely that it is genuine, but my experience (when in Shanghai) of Chinese copies is rather hit and miss. However I trust their ability to crack a PIC chip! The typical sort of problem is they might crack a 323 and claim it is a 327.

The specs on this suggest it's a 327:
http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=290338187793&ssPageName=STRK:MEWAX:IT

Of course it is just possible that Gendan is getting their boxes made in China, and these are all the same kit. Someone said they bought a Gendan for £40 off ebay but I can't find the genuine article.

 

[sixpotman]

Mike search on ebay item number 150365755175

Gendan unit for £40 including deliver. The softwares their
light version but there is other available that work ok with it.

Bought one myself.

John

 

[danhans115]

I think it's been established that Gendan make one of the better tools:
http://www.gendan.co.uk/product_ELM327.html
Anyone tried a Chinese copy?
http://cgi.ebay.com/ebaymotors/ws/eBayISAPI.dll?ViewItem&item=370201239692&viewitem=&sspagename=STRK%3AMEWAX%3AIT
Since ELM want $33 for their chip it seems unlikely that it is genuine, but my experience (when in Shanghai) of Chinese copies is rather hit and miss. However I trust their ability to crack a PIC chip! The typical sort of problem is they might crack a 323 and claim it is a 327.

The specs on this suggest it's a 327:
http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=290338187793&ssPageName=STRK:MEWAX:IT

Of course it is just possible that Gendan is getting their boxes made in China, and these are all the same kit. Someone said they bought a Gendan for £40 off ebay but I can't find the genuine article.

They cant claim an ELM323 is a ELM327, the PIC chips are completely different (16F core vs 18F core).

I bought a chinese copy from ebay for something like £22 in the UK - the ad siad v1.2a but it was infact an ELM v1.0a. It just had v1.2a printed on the PCB. The v1.0a is pretty useless for anything other than kwp2000 stuff and needs modifying to run at speeds other than 9600bps. I ended up buying the gendan one and swapping the other one for something.

I would say gendan are getting their interfaces made in china as the PCB is almost identicle to that of the cheaper interface with a few minor differences to the silk screen. They just come with better quality cables and in a nice box.

I also got mine from Gendan via ebay for £40

 

[sixpotman]

So are Gendan using a genuine Elm Chip or a Chinese Clone?

John

 

[danhans115]

So are Gendan using a genuine Elm Chip or a Chinese Clone?

John

Clone I would say. Given the price of the ELM chip from ELM electronics (£15-20 depending on quantity etc), I really cant see them using the genuine article.

If the manufacturer has managed to defeat the code protect mechanism on the latest v1.3a and get the code from it then their clone would work in exactly the same way as the original, so it shouldnt be a problem

If you want an ELM327 which definately has the real deal inside your best bet is to buy one from scantool.net, but they are about twice the price of the Gendan at least.

 

[sixpotman]

Just checked back to when I bought my gendan unit
I have an email stateing that Gendan use genuine elm
chips and at present are the 1.3a version

John

 

[danhans115]

If you are that bothered answer ask ELM electronics, for gendan to get the chip at a cheap price would mean that they know them by name. I wouldnt always trust everything a supplier tells you.

Most of their Gendans cheaper equipment is re-branded stuff from the same handful of far-east suppliers, so I cant see why the ELM would be any different.

So long as it works..