Problems with increased electrical load causing fueling to lean out

I've noticed that when my fans come on the Lambda values lean out by 2 to 4 % I then found out that this happened with any high electrical load so the same happens with head lights, fogs and even heater fan. To try and combat this currently I've wired the fuel pump straight off the battery, fitted an fuel pressure sensor (no change in fuel pressure), fitted a new turbosmart FPR (stock one was actually faulty anyway but still didn't change anything) and fitted a new Link Can lambda wideband to replace my analoge unit incase it was a analoge voltage offset issue, which turned out to make no difference.

What I have noticed is that when back probing probing the injectors they are showing a lower voltage than the ecu when the fans ect are on. can someone tell me if this difference is likely to be the cause of my problem?

The display on my Head unit is a direct Can feed from the ECU so its the same voltage the ecu is seeing. I've messed on with injector dead times but as there is a descrpency between the injectors and ecu sometimes, altering the deadtimes doesn't help.

This is on a Toyota Celica GT four with a 3SGTE rev 3 engine & standard 540cc injectors (the same as the later spec MR2 turbos)

Hey,

Sometime its because the injectors share the same voltage supply (+12v) with other circuits. Or parasitic voltage drop caused by wrong wire sizing, corrosion in a connector, etc...

I'd be looking at where the 12V feed for the injectors is coming from. If the supply voltage to the injectors is different to what the ECU is seeing then the ECU can't correctly account for dead time and you will definitely see variations in the AFR. To test this I'd see if you could temporarily supply 12v to the ECU and injectors direct from the battery.

Thanks for the replies, I've stumbled across another post on this forum from someone else having a similar problem. I've now feed the injectors & resistor pack from the EFI relay so now the ecu and injectors are seeing the same voltage. Makes me wonder how many other GT4s and MR2's have this problem.

Is there a way to correctly set up the injector deadtimes? I've got the values from Link and from Denso (which were similar) put i'm still getting a bit swing so was wondering if theres a technique that can be used to check if the values correct.

If you are using resistor packs, it's highly unlikely that any documented deadtime info will account for this. You probably need to measure it yourself if you want it to be accurate.

Can you setup the ECU to batch-fire the injectors? If so, one technique is to tune one site for Lambda 1.0 in sequential, then switch to batch fired (ie, two injector firings per cycle). If the Lambda value goes down (richer), then you know the deadtimes for that voltage / pressure are too short. Or if Lambda goes up (Leaner), then the deadtimes are too long. Change the deadtimes, re-tune for Lambda 1.0 with the new deadtimes running sequential and repeat the test.

Would that be group/staged ont the LinkG4X?

Could this be done at idle or would it need to be at a different load site?

It looks like that Help Browser shown in your screen shot has the explanation of each mode, and that you should be able to do the test I was describing, using Multi-point Group, with Injection Rate set to 1/2 Engine Cycle. Firing the injectors every 360 degrees is what it would look like. Hopefully your wiring / firing order will allow this.

Idle is fine, but I would suggest any idle control or closed loop fueling should be turned off. You want the engine to stay right on one cell or put the same fuel value in surrounding cells so if the RPM increased or decreased slightly it would still be trying to use the same fuel table value.

Thanks David, I'll have a go at this today. I though the deadtimes were off but I had convinced myself they shouldn't be changed as Link told me the deadtimes were set really close from the factory and with me being new to this I took that as being gospel

I've given this a try and think i've got somewhere near and if I'm right the numbers in the base map were way off. Am I right in thinking that by swapping from sequential to multi-point i.e. 2 injection events per cycle, that the error in deadtime will be essentially doubled compaired to sequencial?

Exactly right. If the deadtime is correct, then the fueling should deliver the same amount of fuel in one pulse vs. two pulses (each with half the fuel).

You can still tune a car with the incorrect deadtimes, but you will "bake in" any errors in the fuel map. And when you try to add "10% fuel" you often won't get exactly 10% and have to make a few tries. Accurate dead times allow better compensations and accurate fueling changes using math (so it will be better with the VE model if you choose to change Target Lambdas.

I understand why you would disable Closed loop fuelling but could you expalin to me my you would disable idle control?

It would depend on the idle control. Some use an additional air-bypass or might use large values of ignition retard which could skew lambda readings. If your idle control just locks onto a specific RPM and you have a steady lambda reading (not noisey), then I would have no problem leaving it enabled.

Yeah I does have a small bypass valve but that pretty much goes down to a low duty cycle and stays there. The ignition idle does the bulk of the work and the Lambda readings are pretty solid with very little swing at all.