pcmhacking.net

Oh yeah and bonus points for being able to automate sending a keep alive message at regular intervals and also RX filtering. What the AVT can do writes on the book on good interface capabilities. See page 73.

It looks like this adapter has existed since 2009, but based on a PIC processor previously. http://www.electronics-lab.com/obd2-elm ... -with-usb/ the pic shows the err pic version and the link is to the same page as the first post here. The pic source code no longer seems to be available and it looks like he used to sell the boards but nothing is listed for sale currently.

Ahhh!! I wondered why the name seemed so familiar! Came across that design before.

Maybe hes just been working on it slowly over the years.

Im getting a couple of these made up to see what they can do. They arnt cheap to make, but the code compiles with no errors or warnings so will be interested firstly to see if it does what it says on the tin. Someone on the hack-a-day article said their fab shop complained that the suplied gerbers are missing the bottom silk screen layer in the zip, but i read that as no printing on the bottom. The fab house i use didnt have any problems with it anyway.

antus wrote:Im getting a couple of these made up to see what they can do. They arnt cheap to make, but the code compiles with no errors or warnings so will be interested firstly to see if it does what it says on the tin. Someone on the hack-a-day article said their fab shop complained that is missing the bottom silk screen layer in the gerber zip, but i read that as no printing on the bottom. The fab house i use didnt have any problems with it anyway.

What made it expensive? was it because its surfacemount or the IC's are dear?

I dont have time to build and they charge per smt solder point for assembly. So its a 48 pin main ic to begin with and then the rest. Lots of solder points. Plus initial costs for cutting a solder mask for this case a quantity of 3 boards. For a small quantity like this its not worth the expensive machine setup costs to gain the efficiency and lower price of fully automated assembly (nothing ive done to date has worked out worth it). Through hole tech (THT) is more expensive than surface mount (SMT).

There are J2543 cables in many forms, including expensive options like drewtech mongoose pro GM 2 and cheaper options like MDI clones from China. All of these are expensive also, and the J2543 spec is overly complex and hard to write for. A cheap device like the above which does support 4x and large packets would solve the hardware problems. The hardware looks capable and the software is GPL.

Yes this is true.

My boards arrived today, but so far no good. The manufacturer has installed transistors instead of 5v regulators. No smoke (it survived I think, but there was heat) so i'll need to source some replacement regs and repair by hand before I can try again. You can see IC6 is a BC547 rather than an L78L05... I havnt spotted any other issues yet, but there may be....

There was also some confusion at the manufacturers about the orientation of the main chip. It turned out the picture on the obd2 board shot in the first post was incorrect and they noticed and didnt duplicate the mistake! I reported this back to obddiag.net and today the pic they have on their site looks correct.

A cheap device like the above which does support 4x and large packets would solve the hardware problems. The hardware looks capable and the software is GPL.

Yes, it is capable. I hacked a proof of concept in to it with success but it was messy. In the end I decided I wasnt happy with the code quality. The elm protocol sucks because you have to send 2 bytes for every byte of data. I was able up size the buffers to about 180 chars with no significant changes which allowed for about 80 bytes of actual VPW data after converting it back to binary on the interface before VPW transmission. It was an OK proof of concept, but needs a closer look and a cleaner implementation. Modifying the firmware to support a protocol more similar to the AVT was something I didnt have time to do. The timing requirements were pretty straight forward, though. The firmware was not stable with a buffer larger than about 180 bytes, but it totally would be possible if it was written properly.