1987 AMC Renault GTA 2.0 - Gone to a new home.

Bit of progress, digging through some very badly scanned PDF files from about 2003 I found the pinout of the ECU's plugs- the later Renix model ECU's had a single Amphenol connector with 50-odd pins all in one, and this is heralded as the single-point injection model over here, whilst the twin-plug model is the multi-point.

Well, upon removing the board a while back to repair the throttle stop motor control circuit I took a look, and indeedy, yesiree, they have used a multipoint circuitboard, torn the FETs off that would control the other 5 injectors and left a single one.

Man, this is why I love the efforts of car manufacturers during hard financial times. Not.

It's almost like working with a BL product- each serial number for each car with its' own unique configuration. No two cars are exactly alike!

It's cool and dry enough that I can now bring my oscilloscope outside with me (earlier in the year anything brought from inside with the aircon on would immediately get a liberal coating of condensation) so I can begin to probe the plugs on the back of the ECU and actually know what to expect...

--Phil

It's a steep learning curve, is this electronic fuel injection.

As one of the aircraft engine-men I used to work with said: "I can't be having with all these modern boxes of electrickery, it's all voodoo and squiggly-Amps!", referring to the digital traces you had to do to troubleshoot anything more complex than a carburettor.

Despite growing up in an era of squiggly-Amps, there's a lot to learn. This ECU, despite being fairly old in design by modern standards is quite comprehensive in what it can do. If it logged data and was adjustable, it would do as much as a newer Megasquirt.

It's all good in theory, but the electronics have aged, the wiring is French and maintenance here on the electrics is minimal to non-existant at best. They were (and still are) a very misunderstood vehicle. The failure of one sensor is enough to cause the vehicle to stop running (Crank position, MAP sensor imparticularly) or alter the fuelling so much it'd run like a bag of nails.

Ironically, there is no warning light on the dash to indicate to the user (Unlike OBD-II models with the check-engine or MIL light) that a failure has occurred. On the whole people tend to ignore it, but at least it'd be addressed after a year as check-engine light on is a) a safety inspection fail and b) if you get pulled over and the cop notices the light on the dash, a ticketable offence.

I'm beginning to understand the ECU's program, how it takes squiggles in on some wires and makes fuel and spark happen on the other end but when it throws its' toys out of the pram, without much in the way of diagnostic tools it can be mighty frustrating.

It DOES have serial data out to let you know what it's doing, but I haven't got as far as getting anything to be able to capture that information and make sense of it...

Vive la France and all that

haha

--Phil


PS At least I can laugh at it. The ECU nearly got pulled out and thrown on the floor the other evening.

After a conversation with my father he suggested from the symptoms that it's both fuel and ignition.

I went plugged the ECU back in and started the car up.

It did its usual. I went got my neon tube spark tester to see if the spark was even getting down the wire at all. Blink blink blink. Got a flashing light.

I at that point have my hand resting against the spark lead, and rev the engine.

POWPOWPOWPOW from my right hand to my left (not good, I know!)

The spark leads seem to be breaking down under load.

Swapped the kinglead out with no4 lead. Some of the problems went away and the engine developed a slight misfire under load. Still present, and still stinking rich.

Unplugged the lambda sensor... Took kit for a short spin down the road. Hesitates a little still but I got into second and up to maybe 15mph. Lots better.

Get paid Friday, gunna get a new set of leads for it.

Lambda sensor will probably have to wait.



Stuck my scope onto it to see if the knock sensor is putting out. It is. Despite being all smashed up.

Changed out the plugs, same trick, though the current plugs are fairly pale on the electrodes. Odd.

--Phil

I lied, pictures FTW.



;D

Somewhat retro laptop that I'm going to try to hook up and get my GTA to talk to.

No playing videogames on this beast lol

(I'm surprised I managed to hack it to run Windows 95 to be honest with you, probably my best achievement of my university days)

--Phil

Only the French

Shame about the stall in progress, look forward to an update after Christmas

I think you pretty much hit it bang on the head with your first comment

I think the truth is more boring: It's a standard that never really made it. 31250 is the data rate for MIDI bus traffic amongst other things, and 62500 is just twice that.
The standard serial port speeds I think came about the same as most other speed standards, in as far as "What's the fastest speed we can push down this line and not get errors?" Probably find that 0.2MHz difference is the safety buffer-zone to prevent errors from occurring. After all, you can stick that signal down 50 feet of untwisted, unshielded (edit: my bad, RS232 states the wire is shielded) wire and expect it to work.

But yeah

It's French, and I kinda expected something to go amiss.



I might manage to squeeze a pittance of an update in before then. Looks like we are going to be flying out to the UK on the 23rd so there'll be a 2-week hiatus either way.

--Phil

I read that. We're supposed to be flying in to Heathrow on Friday morning.

--Phil

Parked up for now, on vacation to see the folks.

the wife stipulated I am not to be bringing lots of parts for it home- last time that was the cause of our baggage being opened for inspection as it was full of balljoints and trackrods etc...

Merry christmas, y'all.


--Phil