Exhaust gas leak. Suspect manifold

[Telf 224 2 Cars]

Ok so now we are at a situation where we have a digital system operating in a mode that is contrary to the ideal (grouped versus sequential). We have a sensor input that is disabled. Now the ECU is software driven so unless it reverts to a failsafe mode it is still essentially running in sequential. The same remains true. It is causing an issue at low rpm but not at high because sequential doesn't matter at higher rpm.  I don't know if the software has a reversionary mode ( doubtful) but this simple error could be the root of the whole mess.

[jackherer 543]

2 hours ago, Telf said:

So looking at the diagram my loom is built exactly like yours. Do I need 4 separate injector signal lines from the ECU? Does our fuel system differ between the MI16 and the 8v ( as in mine is a common fuel rail ). I'm wondering if I've blindly followed your diagram and missed a vital point.

Incidentally I was looking at the K3 manual and page 58 shows it wired as yours is with two pairs of two injectors so that's 100% as valid for the K3 as it is for the M3D.

[jackherer 543]

No, it's running in grouped because that is how it is wired. I know what you are saying but the only possible difference here is the timing of the pulses occurring during the combustion cycle. Not the amount of fuel injected unless it was either doubled or halved! Dave Walker doesn't think it will have any affect and I'm inclined to believe him but also prepared to be proved wrong.

[welshpug 1,645]

I would imagine if its not seeing a cam signal it would default to batch fire, I'm not especially familiar with Emerald though, but all the other sensors will have a programmable value to use as a default if they get no signal.

 

it does sound like purely that you jad a few air and exhaust leaks that played havoc with the lower rpm load sites, coupled with some unknown as yet issue with the serial connection.

[jackherer 543]

6 minutes ago, welshpug said:

I would imagine if its not seeing a cam signal it would default to batch fire

I put Paul's ECU in my car and put my map in it but it wouldn't fire up until I changed the injector setting from sequential to grouped.

 

And when it did fire up the plugs were fouled really badly and it ran terribly until I put new plugs in then it was fine.

 

So it seems to make a huge difference even without a cam sensor, Dave Walker said it would double the fuelling.

 

I find it odd that the ECU software lets you configure it to not use a cam sensor yet still lets you choose between grouped and sequential injectors elsewhere in the settings.

 

But the bottom line is when you have two wires coming out of the ECU and each wire is connected to two injectors they are firing in groups whatever the ECU tries to do with them. And if you then put it on a rolling road and put values in the map based on a wideband in the tailpipe it is running grouped fire injection whatever the settings say on the screen. The only thing that could possibly be different is the timing of the pulses in the combustion cycle.

 

In theory I should be able to change the setting from sequential to grouped then effectively double all of the fuel table numbers using the injector scaling function and it should end up running the same...?

 

The serial connection problem has to be electrical noise, the same ECU and the same laptop are fine when using my car/wiring harness/engine.

[Telf 224 2 Cars]

I think you should change it to grouped , double everything and see what happens  I bet the problem goes away. Like I said it may be working in grouped because of the wiring bit I bet the ECU s still trying to fire them sequential which changes the pulse time. As I said all the info online says less than 2500 rpm which is where the fault is showing

[Telf 224 2 Cars]

Also you can do it safely because we both saw it running and it was in load site 1 and occasionally 2 . So switch it to grouped and see if it runs with the present figures. You are saying it might but like pig. If that's the case it proves there is a difference between the modes be it pulse width or whatever. Change the values in load lines 1 and 2 and see what happens

[jackherer 543]

OK I will give it a go.

 

However all the stuff you're seeing talking about the effects of sequential below 2500 are not relevant because it is NOT injecting sequentially due to being wired in groups.

 

Remember the standard jetronic injection fires all four injectors at once without issues, I really don't think it will make any difference but I will try it and let you know!

 

5 hours ago, Telf said:

the ECU s still trying to fire them sequential which changes the pulse time.

No, it might be changing the pulse timING (i.e. when it starts).

 

The pulse time (i.e. duration) is defined by the number in the map that John typed in with your car on a rolling road with a wideband in the tailpipe

 

And the  firing pattern is fixed by the wiring, the injectors are wired together in pairs so they have to fire in pairs.

Edited September 17, 2019 by jackherer

[jackherer 543]

There are some developments with this.

 

The problem with the laptop disconnecting was very time consuming to figure out. I temporarily bypassed the main power feed to the ECU directly to the battery which solved it but when I tried permanently replacing the wiring either side of the main relay the problem returned. In the end I put a 'scope onto the main power feed and there was an obvious regular spike. Then I noticed the tacho converter right next to the main relay... It wasn't plugged in when it was being mapped at Emerald which explains why their PC stayed connected and mine didn't. With that unplugged the laptop stays connected indefinitely with the engine running and using all of Paul's original wiring.

 

Regarding the rich AFRs I have sealed several small leaks in the exhaust which has stabilised the numbers on the wideband. The excessively lean figures at idle were indeed false readings caused by oxygen leaking into the exhaust., with it sealed the idle mixture is now a bit rich at around 13-13.5. The trouble spot at 1500rpm up to about 2000/2500 is looking even richer than it was, the worst point at about 1600 RPM that was in the elevens is now apparently about 10:8:1.

 

So I think it was probably mapped with some exhaust leaks that were causing a falsely lean reading so the fuelling was increased to compensate.

[Thijs_Rallye 87 4 Cars]

Good to hear there is progress. What I was wondering about, is there a way to apply lambda feedback to the ECU? Might be worthwhile installing a permanent wideband in the downpipe to keep the AFR under control. A spartan isn't that expensive tbh.

[Andy 105 1 Cars]

42 minutes ago, Thijs_Rallye said:

Good to hear there is progress. What I was wondering about, is there a way to apply lambda feedback to the ECU? Might be worthwhile installing a permanent wideband in the downpipe to keep the AFR under control. A spartan isn't that expensive tbh.

Yep. The Emerald has a neat closed loop fuelling algorithm so fitting a wideband is a good call . I use the AEM Uego but have used Innovate in the past . It would be a long term solution, but does not necessarily pinpoint the problem now. Both fuelling maps are identical at light throttle over the rpm range and, critically, when it was first mapped, no nasty fuel smell,

 so something changed subsequently .

Good detective work on the Tachometer converter Keiran . I am surprised . I used the same setup on my last car with that converter with no difficulty , but it was a long way from the ecu or the laptop cable, so perhaps I was just lucky 

[Andy 105 1 Cars]

11 hours ago, jackherer said:

There are some developments with this.

 

The problem with the laptop disconnecting was very time consuming to figure out. I temporarily bypassed the main power feed to the ECU directly to the battery which solved it but when I tried permanently replacing the wiring either side of the main relay the problem returned. In the end I put a 'scope onto the main power feed and there was an obvious regular spike. Then I noticed the tacho converter right next to the main relay... It wasn't plugged in when it was being mapped at Emerald which explains why their PC stayed connected and mine didn't. With that unplugged the laptop stays connected indefinitely with the engine running and using all of Paul's original wiring.

 

Regarding the rich AFRs I have sealed several small leaks in the exhaust which has stabilised the numbers on the wideband. The excessively lean figures at idle were indeed false readings caused by oxygen leaking into the exhaust., with it sealed the idle mixture is now a bit rich at around 13-13.5. The trouble spot at 1500rpm up to about 2000/2500 is looking even richer than it was, the worst point at about 1600 RPM that was in the elevens is now apparently about 10:8:1.

 

So I think it was probably mapped with some exhaust leaks that were causing a falsely lean reading so the fuelling was increased to compensate.

You could bite the bullet, save the existing map, alter the fuelling at light throttle and see if that (a) improves the afr and (b) reduces or removes the fuel smell. If it is just running rich around those cells, a reduction should fix both (a) and (b) 

[Telf 224 2 Cars]

Kieran , I reckon it's worth doing this I can drive and you can alter it on the fly. I can come over early tomorrow as I'm on a late so we could potentially fix the problem 

[Thijs_Rallye 87 4 Cars]

If you save your current map I don't see why you couldn't try, you can always revert back. Check what regions of the map you hit when it smells badly, deduct 10% in that area and try again if it has improved.

 

 

[jackherer 543]

OK I'm up for trying it, I might be busy tomorrow though, I'll let you know later today.

 

One minor problem is the lack of tacho will make it hard to hold it at a steady RPM...

[jackherer 543]

With regard to lambdas the problem I see there is that some of the exhaust leaks in question were at the gaskets either side of the manifold so even a downpipe located sensor would have seen false lean readings and added excess fuel to compensate making the problem worse.

 

The same is true for the mapping sessions, the manifold fitted then was cracked so even if they'd used a downpipe mounted lambda instead of their tailpipe probe at Emerald I think they would have still seen false readings and made it too rich to compensate.

 

We could connect my wideband to the Emerald so it can adapt now that the leaks are fixed but I don't think the tailpipe sensor reacts fast enough so we'd need to get a boss welded into the downpipe.

[Telf 224 2 Cars]

If we removed the wires from the relay base we can just use the spare flying loom wires to extend the tacho loom and dump it on the back seat

[Tom Fenton 1,536]

If you are planning to try and correct the map on the road, disable the acceleration enrichment while trying to do anything steady state

 

You will find its very limited what you can achieve, as its hard to keep the map in any one position for long enough, other than the very light load areas.

 

I'd say you are better to fit a wideband properly in the exhaust, wire the feedback into the ECU, then use the adaptive mapping function and allow it to suggest an amended table itself as you drive it.

 

Ultimately you will probably end up back on the rolling road however.

[jackherer 543]

2 hours ago, Telf said:

If we removed the wires from the relay base we can just use the spare flying loom wires to extend the tacho loom and dump it on the back seat

I tried isolating it as much as possible but nothing I do stops it disconnecting the laptop.

 

it's weird because if I run an isolated +12v wire to the ECUs pin 28 directly from the battery the laptop behaves but if I reconnect the ECU to the main relay and run an isolated +12v wire to the tacho adaptor the problem remains.

 

I ran a new wire from the tacho converter output to the brown multiplug and kept it away from all other wiring but that made no difference.

 

The only two ways I can keep the laptop connected are either with the tacho converter totally unplugged or with an isolated +12v wire run to the ECU from the battery.

 

2 hours ago, Tom Fenton said:

If you are planning to try and correct the map on the road, disable the acceleration enrichment while trying to do anything steady state

OK noted.

 

2 hours ago, Tom Fenton said:

You will find its very limited what you can achieve, as its hard to keep the map in any one position for long enough, other than the very light load areas. 

Yes, it was pretty hard holding it at fixed points with the brakes just to read the lambda numbers but trying to hold it at specific load sites while trying to co-ordinate with a passenger with a laptop is going to be tricky, especially as I still can't get the rev counter and laptop to work simultaneously.

 

It's definitely going back on a rolling road at some point.

[Telf 224 2 Cars]

Right ok so it gets put back together and we go with the session as planned on Wednesday then

[Andy 105 1 Cars]

Ah. Not sure I meant you to get this complicated. All I was suggesting is that you connect the laptop and with the ecu powered by the engine not running, you adjust the fuel cells at the 2000rpm to 5000rpm and 0% to 12% by, say 10% . Save those values on the ecu map ( having first saved the unedited map to the desktop ) and then just drive the car . You will soon know if (a) it still stinks  of fuel and (b) whether the fuelling has gone too lean, in which case it will not pull as well . You will not know what the afr is, but you will know if you are heading in the correct direction. .

 Oh, one more thought. On the K6 , there are two live feeds to the ecu( I cannot remember the PIN numbers but easy to look up) I assume that Paul has simply wired both of these feeds to the main ecu relay ( that will be ignition switch controlled ) 

[Telf 224 2 Cars]

So after a week of much work by Kieran and a less than successful time at a local garage we have arrived at the conclusion that the exhaust system needs replacing and the MAP adjusting. 

 

Kieran has supplied me with a stainless split/2pipe down pipe so now I need the rest of the exhaust. 

 

I'm wondering if to go for a complete stainless system that a local place will manufacture at cost as yet unknown or fit a mild system off ebay. 

 

See pic below... At £160 it seems reasonable. Any body had any experience with these systems or general advice as the best way to proceed?

[Telf 224 2 Cars]

Update the as yet unknown cost for the mid and rear section is £375-£425

[jackherer 543]

Sportex is cheap nasty tat...

 

You should be able to get a stainless system made for less than that but it will be loud, they always are.

[Telf 224 2 Cars]

ISC is cheapest locally . Other than that I guess a standard exhaust from a decent fitment place