i have a feeling there is an overload to your laptop... Is there any way that the voltage values being sent to the computer are being saturated due to too large or too small gauge wire with the USB?? Just a thought as I am dumbfounded byt the rest of the talk in this thread (caveman make wheel rounder):D

Tom

Hi Tom,

You should have a serious chat with MR Ohm, before you end up angry and stuck with no where ti go, in some electrickery stuff. :D ;)

I don't have one of these monitors I am merely saying there might be a hangup in the wiring rather than an issue with software

Tom

“Is there any way that the voltage values being sent to the computer are being saturated due to too large or too small gauge wire with the USB??” :confused:

Tom, good one.

Mr. Ohm and I now understand that your exactly worded query must have been intended as a joke in a sarcastic form. :confused: ;) :D

No. I am not using a USB cable anyway, I'm using a serial cable. If there's a flaw, it's either in the interface software or the serial->TTL converter box.

ok... well why would the code change once it got warm?? The only difference you are going to see is that the ECU enters a closed loop state from which it does not draw from the sensors anymore but from the memory... if I understand it correctly

Tom

Yeah, that part I honestly don't understand whatsoever. If I just started the car, it returns data just fine. It'll sit and idle for a while and still return proper data. But as soon as I start driving, the data gets more and more corrupt until it's not even readable. If I then turn the car off and back on again, the ECU is back to normal and sending proper data... until the next time I drive it (or let it idle for a LONG time, as was the case with the time in the Fry's parking lot).

I don't know enough about the ECU to tell you what's different when you're driving. I really don't.

But the Subaru Select Monitor, in the exact same situation, reads perfectly well... :confused: There goes the "broken transmitter" theory...

Could it be ignition or alternator noise riding on the TTL pulses due to the engine running. TTL is a high speed logic that could be detecting the noise as data pulses. As a result your TTL converter could be generating weird data when the engine is running.

The amount or shape of the ignition noise could be dependent on the engine load or speed. If it's alternator noise, then it could depend on the bus load that changes as the battery charges up, or with the change in engine speed.

You might need a low-pass filter in front of the TTL converter to strip those pulses off the real data. Or the noise could be coming from your laptop and being sent on the USB power to the converter. Is the laptop powered by the car's 12V or internal battery?

As Trevor said before - a scope (picture) would be worth a thousand words, in this case. :D

I spent many years chasing Electromagnetic Interference (EMI) problems like this. Lots of luck !!!

it sounds more and more that when your car enters closed loop this is when you are seeing corrupted data... Phil, have any advice on this??

Tom

I'll second Jim's comments.

It would probably be smarter to keep the TTL level signals as short as possible — as close as possible to the ECU. In other words, the MAX232 chip circuit should be right at the ECU plug.

Even if you leave your adapter box as-is, shielded TTL and RS-232 cables may help (if they're not already).

You may be getting additional noise if your laptop is connected to a different power and ground than the ECU (are you using the cigarette lighter jack, perhaps?).


So that USB-to-TTL adaptor (where the chip is inside the USB plug and the long wire is TTL) may not be such a good idea. However, one of the companies has this version that is a USB socket on a little PC board with the tiny chip right underneath it...

http://www.ftdichip.com/Images/MM232R.jpg(img)

That could be attached to the ECU plug, with a USB A-to-B cable to connect to the laptop.

Thanks for the responses guys.

However, I can blow your comments right out of the water... again. Sorry.

The Subaru Select monitor was on the ground beside the car, and had MULTIPLE cables between it and the actual ECU plug. They didn't look particularly heavy-duty, but were thicker than mine (probably because mine aren't coated in anything to keep all the wires together). If anything, the select monitor is even FURTHER from the ECU than my laptop is.

And for all asking, it matters NOT whether I use internal laptop battery or use my inverter. The result is exactly the same in either case.

Good theories, but sadly not correct.

EDIT: Also, I have another way to prove that the theories are wrong.

The TCU sends coherent data. The TCU is right next to the ECU, and is CLOSER to the engine bay than the ECU, and is FURTHER from the select monitor/laptop than the ECU. I rest my case.

EDIT 2: Also, 1994 cars apparently don't exhibit this behavior?

That is bad news. :( The junk log you posted looks a lot like the data that comes out of my 92 all the time. I have noticed that there are lots of FF and other high values. As the binary bits are coming down the wire, the converter is seeing a 1 (2 volts to 5 volts) when it should be seeing a 0 (0 volts to 0.8 volts).

Interestingly, I noticed on my 92 a while back that I got partially good data when the battery was low and all junk after I charged the battery. I wonder if the signal voltage from the ECU is too high?

I don't know why restarting your car would fix it.

Thanks for doing that. Subaru dealers over here are only interesting in selling Imprezas. Their attitude is "buy an impreza or go away". I'd be surprised if they even have the SVX select monitor cartridge.

I assume the select monitor is a TTL device like the ECU, so no TTL to RS232 conversion is being done. This, coupled with the revelation that the Select Monitor gets good data seems to indicate that the problem is in the adapter box.

Strange how it works for the TCU though, and how the problem is intermittant, and only affects some SVXes.

Hi Jim,

Nice to have advice from an expert!

I did wonder if it was ignition or alternator noise, or even Trevor's often quoted discovery of an AC leakage from faulty alternators. However, my 92 has the problem even when the engine is not running, which excludes those ideas.

I understand that for TTL, 0 volts to 0.8 volts is counted as logic 0 and 2 volts to 5 volts is logic 1. So presumably 0.8 volts to 2 volts is "undefined" and could be randomly interpreted as a one or a zero by the MAX232 chip? If so, then less than 1 volt of noise on the line could be enough to cause this problem?

The thing is, the ECU always, without fail, understands the commands from the laptop. It's only the data coming from the ECU to the laptop that gets corrupted.

As we don't have a scope, we need to solve this by experimentation. Shooting in the dark somewhat. Do you think that it would be worthwhile to try and drop the signal from the ECU by a volt or two? How would I go about this? My first thought is a resistor, but what size? Can a resistor pass high-frequency signals without distorting them? My electronics knowledge is very limited.

Phil.

It's possible I guess, I can't see why though.

On a separate note, as you know about engine tuning, would you be able to cast an eye over the fuel and timing tables that I posted on my website yesterday and see if they make any sense?

http://www.alcyone.org.uk/ssm/ecu (down near the bottom of the page)