triddle

I've been plucking away with someone else on another forum at figuring out how to talk to the 2006+ CAN based S2000 instrument clusters but there is so little information out there that it seems its all waiting to be discovered. The pinouts of the cluster and ECU are known but we only have a guess right now at a signaling voltage of 5 volt for single wire CAN and a totally unknown bus speed. The project has progressed to the point where the instrument panel is about to get "searched" - that is the full suite of signals that can be passed into it will be to see what makes it blink.

This will be many many times easier if there is some kind of known starting point, especially bus speed. I'd like to see a physical representation of the wave going across the CAN bus to the instrument panel but I don't have access to a 2006+ S2000. Is there anyone who can hook up an oscilloscope to their car and get some snapshots of the communication across that bus at initial power on? Can you transfer the graph to an image and get it on the Internet?

This is actually pretty easy - by sticking some straightened paperclips into the back of the ECU socket and hooking up the test leads to that you won't even have to disconnect or modify your vehicle. Using a scope is safe as well: as long as you don't accidentally short something out you will not damage your car. On a 2006 pin 13 of ECU socket E is for communication to the instrument panel and pins 4 & 5 on ECU socket A can be used for a ground; I haven't checked the pins for other years.

If anyone can get me that snapshot I'll decode it out to binary data by hand and try to correlate it to known CAN protocols. The troublesome instrument panel is about to get cracked open.

SgtB

So the '06 and up clusters are different? Strange. I've got a scope, but I'm in an '01. Sorry.

triddle

They are different - some of the lights and the coolant meter was moved to a single wire using CAN. No one is known to be able to drive the coolant gauge on the 06+ instrument panel. This particular fellow has a 2007 panel in his Eclipse and has everything working but some lights and the coolant gauge. Getting the gauge to work is a combination of figuring out the electrical characteristics on that pin then figuring out what protocol Honda is actually using for communication. We believe the remaining lights are also being controlled over CAN.

CAN is car/factory ethernet basically, then LIN (local interconnect network) and DeviceNet would be the TCP/IP IPX/SPX analogues, though analogies after this get really difficult because the higher level protocols in use are extremely application specific and LIN and DeviceNet have very different signal characteristics but share the CAN packet format. So far we know 5 volts, 1 wire, and probably 10.6kbps across the line. By snooping the packets going on that wire at the binary level I'm hoping I can identify the protocols in use then the panel can be brute forced into submission given enough time.

More information about the 06+ panel than you can shake a stick at: http://honda-tech.com/showthread.php?t=2768013

If the panel is running on LIN this is what communication would look like at a very low level: http://www.youtube.com/watch?v=M1VI9wITmA4...player_embedded

MikeyCB

I would appeal to Modifry and see if he has the info you're looking for. I suspect he might.

ahrmike

Damn electrical engineers!

haha, jk ME rep :]

triddle

I'm apprehensive about contacting Modifry about the 2006+ coolant gauges - I think he gets that a lot and I don't want to add more bother to him. He doesn't support the 2006+ gauges with the ECT fixer. Maybe the bit gods will smile upon me and he'll chime in.

Worse! I'm a Unix administrator/software developer - I write software that does stuff like this for fun:

[media]http://www.youtube.com/watch?v=dNGi-SZ9kGw [/media]

CarbonCrew

Triddle, where are you located? I'd be willing to help if I can.

I do have a 2006 S2k, however I don't have an oscilloscope. I've been look at some of the Parallax USB units in the past, I may just pick one up.

triddle

Thanks for the reply! I'm in the Portland, OR metro. After doing some digging I found a OBD-II transceiver implemented through a PC sound card - using one of those simple devices may actually get an ok enough capture to see what's going on since the signaling rate is believed to be the same. I can manipulate the PCM from the sound card if you can capture it. All though I'd hate to ruin a perfectly good excuse to get a scope too!

I got to the harness going to my ECU today as well - it's not hard to do and there is some extra length in the wires, at least on my AP1. It's not fantastic but it could be much much worse.

TopGear

Ok, I'm really nitpicking here, but I'm a math geek, and I tend to spread the gospel of correct math term pronunciation. Fourier isn't pronounced four-ee-air like you said in the video. It's four-ee-eh (Fourier was french. Fourier rhymes with play).

Don't mind me

</tangent>