Injector scaling

[richardpalinkas]

Have a question regarding scaling injectors, just fitted a walbro 525 and fuel pressure reg to my car, set base pressure at 4 bar as required, injectors are a bit of an unknown so I’m working off what the data was previously in the car, but what I am experiencing is good cruising afr on wideband, almost spot on under wot to redline, but I get a lean spike under decel, which I have decel fuel cut disabled. It’s not all the time, but if I’m coasting along, lambda is 1, but if I apply throttle, say 25% and come off throttle, I get 1.15lambda that slowly creeps back down to 1. Under all cruising I’m sitting close to commanded, maybe slightly rich. That being said, it did the same before upgrading fuel pump from 460 to the 525, dead times are the same, injectors I believe are sonic performance 1000cc. The sonic performance offset table is a lot different to the offset that was originally in my car. Can offset cause the random lean out under some decel applications?

[Roland@pcmtec]

Increase your minimum pulsewidth. 

Also log fuel pressure on decel with a seperate gauge. You might be getting pressure fluctuations.

Also check the narrowband voltage, it might not actually be lean and it's just your wideband, is it in the tailpipe or in a bung? 

How are your LTFT at cruise / decel? If it's above 5% you will get random lean/ rich spikes in transients. 

Finally how does it drive, does it feel lean and miss/splutter if you back off and get back on it straight away? It might not actually be lean. 

[richardpalinkas]

I’ll try to increase the min pulse width and see how it goes. How large a value can you go before going too far? I’ve seen most inj min pulsewidths 0.272 but factory injectors run at 0.9, just so I know if it’s getting too much. 
 

il check what the o2 sensor is doing. Wideband is mounted in the rear o2 sensor port just after the cat. 
 

I haven’t been logging ltft yet as I’ve been mainly looking at the wideband and disabling ltft. Also can you log ltft with pcmtec ? I’ve been having to log with for scan the ltft

 

car drives fine, I just hate looking at the wideband when it’s anything above 1. Gets me worried, even though I know on decel it’s probably not hurting anything

[Puffwagon]

Decel doesn't matter, so just leave it as is and don't worry about it.

[Roland@pcmtec]

Min pulse width is a peg. So if you cruise at say 0.5ms and you set it at 1ms. You'd be injecting double the required amount of fuel and the car would be massively rich. 

So what is too much entirely depends on the injectors and what your true min pulsewidth is. You'd need to datalog decel when this lean event occurs it and see how low it goes. If it's lean only when it's at this very low pulsewidth then increasing it will likely solve your issue. 

The whole purpose of min pulsewidth is make the car run richer when the injector starts dribbling and causing misfires (eg decel and possibly the cause of your lean condition). 

LTFT is included in the default layout file. Load that and it will be in there. 

[Roland@pcmtec]

2 minutes ago, Puffwagon said:

Decel doesn't matter, so just leave it as is and don't worry about it.

Also this. If this car runs fine then you are optimising something that doesn't really matter. 

[richardpalinkas]

So more data logging/ adjusting and what I have done seems to have solved the issue. 
 

I found dead times for my injectors, so to start with I started from scratch. Put in the newly acquired dead times which were quite a bit bigger than id1000 or the data that was in the vehicle, which resulted in a steeper low slope, high slope was similar to what I had with the old injectors. But random lean under low rpm and off throttle is now next to gone. 
 

however... with both sets of injector data I also have an issue which I can’t seem to resolve. At wot, as the boost curve comes up to max boost, approx 3500rpm, my fuel is always richer than my commanded lambda table. After this period, maybe to about 4000-4500rpm, fuel is back to commanded lambda and holds steady until redline. But rich wise, if I’m commanding 0.75 lambda, it goes to about 0.72 lambda, then comes to my desired fuel map. 
 

ideas?

 

thanks

[Roland@pcmtec]

That isn't a very big error and something that most tuners simply fudge out. A few possibly causes.

Your peak injector pulse width will be at peak torque. I assume your peak torque is at 3500rpm so you could possibly just adjust the high slope to correct this.

Another possibility is it could be simply blow through due to camshaft overlap giving you a false lean which you have corrected for at lower rpm. If you set your camshaft overlap to 0 degrees across the entire rev range does the lambda also hold flat (regardless of error). If it holds flat then you can simply tweak the speed density overlap adder tables to fix the issue.

As you can see they drop the camshaft overlap back to 0 at 3500 rpm, so it is plausible this is your cause.

A way to determine if your high slope is truly correct is to drop your boost pressure to spring pressure, does the car still run the same lambda?

A way to determine if your speed density needs adjusting is to add 20 degrees over overlap across the rev range, does the car still run the same lambda? Do the opposite and run -20 degrees of overlap. Make sure the overlap holds flat across the rev range by datalogging it, if it is not flat then the test won't be accurate/useful.

Finally you could do what basically every tuner does and fudge the speed density table where this error occurs. This would take 5 minutes and get you a flat lambda curve but without first testing the lower boost and camshaft overlap adjustments may simply be fudging and error that has resulted from somewhere else in the tune.

[richardpalinkas]

I have already tested the spring pressure as this is the only way for me to road tune without wheel spin in 2nd gear,  then put back in my higher boost and check on 3rd gear on the highway. Lambda is the same boost varying.
Will check the cam overlap and see how that affects it

[Milanski]

My

battery offset at 13.5V (deadtime) is 0.00118s.

low slope is 0.0032lb/s

Mim pulse width is 0.3ms

Logging is see my InjPW is around 0.5ms on decel.  On idle im at 0.9ms.  Is the PW value that is being dataligged excluding the deadtime? As time = slope/fuelmas + offset i would expect to be seing p.w. at least 1.18ms.  

[Roland@pcmtec]

16 hours ago, Milanski said:

Logging is see my InjPW is around 0.5ms on decel.  On idle im at 0.9ms.  Is the PW value that is being dataligged excluding the deadtime? As time = slope/fuelmas + offset i would expect to be seing p.w. at least 1.18ms.  

Correct, deadtime is ignored everywhere and added at the final step only.