Zel's Fleet Blog - BX, Jag XJ-S, Sinclair C5, Mercedes 208D & 230TE, AC Model 70.

[CitroJim]

Zel, I don't know if it'll fit but I have a good injector from a single-point Saxo Throttle Body...

[xantia_v6]

If the lambda signal indicates that the system is running closed-loop, then high CO is very likely due to a tired catalytic converter.

[RichardW]

If the lambda signal indicates that the system is running closed-loop, then high CO is very likely due to a tired catalytic converter.

That was my thought, Mike. 4% CO is about what I would expect an uncatalysed engine to make. If you've got an IR thermometer you can check the cat - it should be significantly hotter on the outlet than the inlet.

[Zelandeth]

If the lambda signal indicates that the system is running closed-loop, then high CO is very likely due to a tired catalytic converter.

That was my thought, Mike. 4% CO is about what I would expect an uncatalysed engine to make. If you've got an IR thermometer you can check the cat - it should be significantly hotter on the outlet than the inlet.

Not a bad thought to check that. Definitely need to check the lambda signal first as there's visible smoke if the engine is revved from idle so I reckon it is running richer than it should be anyway. Then will have a look at the temperatures of the cat.

[Zelandeth]

Interesting. Think we may be homing in on the issue. Looking at the terminals for the lambda sensor on the ECU, D6 is the ground terminal and D7 is the signal. D6 is floating resolutely at 1.5V above ground...this should show as no more than 0.1V according to the manual...

If there's a 1.5V offset being applied to all the lambda readings no wonder it's struggling!

EDIT: Having done some further reading showed that terminal D6 should be a solid ground, going via the ECU then straight to the main injection ground point, bolted to the side of the block on the coil pack bracket.

Tying this terminal to ground directly (I'd nothing to lose by this point really!) did change the behaviour of the car, it's now running even *more* rich, and idling like a carb car with the choke left on. No sign of a proper lambda signal, the sense line is just stuck showing 0.7V...again that voltage appearing from within the ECU rather than the sensor itself it seems.

I think we're at the point now that given I know it had other issues as well (by virtue of the dead check engine light) that I really need to track down a known good ECU before I can chase this any further. Realistically I'm just going to be chasing my tail otherwise.

...It's *entirely* possible that I've just successfully found the fault which took the donor vehicle for most of my injection kit off the road originally!

[Zelandeth]

Okay, so Niva ECIs don't seem to grow on trees...In fact they're very thin on the ground. The fact that they were a relatively rare failure mode (the extremely heavy conformal coating on the PCB meant that they actually tended to survive a dunking when offroading) means that few people bothered keeping them around as spares from breakers just because there wasn't much of a market for them.

Have found one actual Niva one for sale over in Germany - albeit with the seller asking north of £170 for it, which for a scrap pull from a 1995 vehicle is frankly ridiculous. For a brand new unit I'd be willing to pay that, but for a used one which may or may not work it's a bit over the top. £50 or so I'd be more willing to go for.

A theory springs to mind however.

I know for a fact that the injection system was bought in off-the-shelf from GM. So it stands to reason that the same physical ECU is probably used in other vehicles - just with a different map flashed onto the "personality module" which just plugs into a slot at the top of the unit. This brings me to this...

The one used in the Corsa B looks physically the same...in so far it's a silver box with the same dimensions, same connectors and same removable lid on the unit where the aforementioned module with the vehicle data on it slots in, and poking around reveals that it also appears to use ALDL based comms just like the Lada one, even though I've not managed to track down a clear listing of the connector pinouts yet...I have to wonder what the odds of this being a simple solution are...

[CitroJim]

Zel, sorry to hear that it appears the ECU has succumbed

Rather than trying to source a replacement how about looking at knocking together a Megasquirt? I know they're really intended for multipoint injection but one such may offer a solution...

[Zelandeth]

A Megasquirt ECU would certainly do what I need it to, however the problem with the Megasquirt approach (at least it was when I first started looking at this project) was twofold:

A: They're eye-wateringly expensive - seem to recall the figure I was looking at was round £300 when I started digging around.

B: I'd need to find the data to set the thing up - realistically getting the car to a specialist and faffing about with rolling road setup is most likely going to be involved to get the best out of it. I did look around to see whether there were any maps floating around online that could just be downloaded and used as they were, but didn't see any at the time - again, this is going back well over a year now so I may need to do some more digging.

[Zelandeth]

Well, for the sake of £15.99 plus postage, I've grabbed a Corsa B ECU. It should be pretty obvious if I pull the case off if they're radically different.

The manual even states "to allow the same unit to be used in multiple models of vehicle, a unique engine calibrator is fitted..." which adds credence to my theory that they're identical units save for the label on the front and the module plugged into the slot at the top.

This little blue thing I reckon is the important, Lada specific bit...

...and out of the case.

Back of the board. Not much here save for a shedload of bypass caps.

For those into their electronics, here's a closeup of the three ICs in the plug-in module.

If this isn't a PROM of some description I'd be extremely surprised...

I'd say this was a resistor network from the package style alone...

Not so sure about this one...

I reckon the issues I'm having are buried somewhere within the plethora of semiconductors down by the connector sockets, sadly they all seem to have custom part numbers on, which makes it hard to tell what they're meant to be to allow meaningful testing. I hate it when manufacturers of complex devices do that. Yes, HP, I'm looking at you with regards to your 80s era test equipment too.

[Hell Razor5543]

HP also make a nice sauce as well!

[Zelandeth]

HP also make a nice sauce as well!

Argh!...Tea/Keyboard interface error detected...

[Hell Razor5543]

HP also make a nice sauce as well!

Argh!...Tea/Keyboard interface error detected...

[CitroJim]

If this isn't a PROM of some description I'd be extremely surprised...

Yes, I'd say so too


I'd say this was a resistor network from the package style alone...


Yes, I'd agree

Not so sure about this one...


Not one I've seen but it may be a protection device to eliminate spikes and noise on the incoming signal wires before they reach the ECU proper...

First port of call has to be bad/dry joints Zel...

And then poor earths within the ECU. Also try reseating those three plug-in ICs

[Zelandeth]

Already on that track Jim. The quality of the solder work in there is actually astonishingly good. There really aren't any even vaguely suspect looking joints.

Reseating those ICs has already been tried...I did wonder if it was something that daft as it wouldn't be the first time, sadly I'm not that lucky it seems.

Tracing out the ground paths definitely isn't a bad idea...Will pull the heatsink off when I get a chance and see what we can find.

[Zelandeth]

Important safety tip to any of our locals in MK...Avoid Bradwell Common Boulevard today!

Apparently the generally accepted wisdom of it being a good idea to let tar set before letting traffic drive on it, or it being a good idea to erect signage where you're doing road repair work is beyond our local council. I've just finished frantically scrubbing that lot off before it had a chance to fully set. Had managed to get pretty much the entire nearside of the car, plus roof, bonnet, windscreen, rear window and spoiler. Because it was on a roundabout, so the nearside front wheel just sent a bunch flying straight up in the air.

Hacked off doesn't even begin to cover it. Especially having worked in the field for several years, if one of my crews had been caught doing this work I'd have been skinned alive...never mind the crew themselves!