mx594

iTrader: (4)

Let me start at the beginning.

I have an 03 G35 coupe 6MT. About 8 months ago I put together a custom turbo system based around an HKS GTRS twin turbo kit and JWT twin intercoolers. I am running a CJM stage 0 fuel return system with a Walbro 255 pump and Deatschwerks 600cc injectors. Stock block.

At first, I was using a UTEC and an AEM wideband to tune the car. It ran great in boost and made awesome power at 7 psi. However, I always had trouble keeping the transition from stock ecu to UTEC smooth with varying weather conditions. I also had this strange problem where I would get a really lean spot right about 2400 rpm when under light/medium throttle. No matter what I did to the MAF offset in the 0% column on the UTEC it would never go away. I tried adding a ton of fuel at 2250 and 2500 rpm (the UTEC has breakpoints every 250 rpm) and it never made a difference. The car would buck when cruising at light throttle if I was in the 2250-2500 rpm range. Taking off in 1st I could watch the gauge go extremely lean for a brief moment as I passed through 2400 rpm. And this is where my struggles begin.

First I thought it was a problem with my MAF sensor location (blow through) so I moved it a little further downstream to make sure I had a good 9” or so of straight pipe in front of it. I also put the stock MAF housing back in, instead of using the pipe I had fabricated with the MAF adapter welded onto it. Neither change had an effect.

I chalked it up to the fact that the UTEC MAF voltage offset alone was probably not sufficient to compensate for the larger injectors and expect perfect driveability. So I went and got an Osiris Tuner. Osiris allowed me to have much more precise control over the vacuum region of the engines operating range, while letting the UTEC continue to control fuel and timing in boost. I planned to remove the UTEC once I got a PMAS MAF sensor and was comfortable enough with the Osiris software to tune in boost. The first thing I tried to do was tune out the 2400 rpm lean spot. I added a bunch of fuel around 2400 rpm and light load. To my surprise, it seemed to have no effect. Things started to get really strange when I used Cipher to log the O2 compensations that the stock ecu was applying. According to the logs, the A/F compensations looked like a big “V”, centered right at 2400 rpm. They would start out in the 90’s, then dip down to 75 at 2400 rpm, then shoot back up over 100. Essentially what I thought this was telling me was that the stock ECU was actually creating its own lean condition, then recognizing it and fixing it. I figured that this may have been due to the fact that I have an 03 and the stock O2 sensors are narrow-band. I also tried adjusting the MAF curve in the voltages were I was seeing this happen, but that only caused problems at other RPM’s. On top of that it was almost impossible to see what my A/F ratios were doing because logging the narrow band sensors was pretty much useless and there wasn’t a good way to compare the separate data from the AEM wideband sensor to the Cipher logs (The Uprev software does not support 3rd party widebands at this time).

So at this point I was frustrated with the UTEC and Osiris, and decided that I was going to finally do it right so I bought a Haltech and installed it a few days ago. So now the stock ecu is completely out of the picture (as far as fueling is concerned anyway). I bought a base map from Hal which got the car up and running right away. In fact I didn’t even have to make any adjustments to get a stoich idle. I must also say that the Haltech ECU manager software, the USB hookup, and the ability to log the AEM wideband output were all a welcome relief compared to my previous setup(s). However to my dismay, the 2400 rpm lean spot is still there! I have made somewhat of an attempt to add fuel to the low load areas of the VE table around 2400 rpm but again it doesn’t seem to help. I even tried to do the no-load tuning that Haltech recommends, where you rev the engine at idle and tune the no-load column, but when I try to hold the engine at 2400 rpm, I have to add a ridiculous amount of fuel to keep it from going lean…like an additional 30% to the VE table which is nuts.

I am starting to realize that this might be some kind of mechanical problem. Either I am getting more airflow than the computer(s) think I am getting, or I am getting less fuel than commanded. I doubt it is an airflow issue because I have had the same problem with two completely different control strategies (MAF and MAP), plus the MAP sensor is connected directly to the upper plenum and shouldn't be affected by anything except a vacuum leak in the lower plenum maybe...

That leaves the possibility of a fuel problem, which could either be fuel pressure (pump, return) related or injector related.

Any thoughts?

Chris@FsP

iTrader: (4)

Log the IDC and see if it drops at 2400. If not, and it still runs lean in that area, it almost has to be a mech/airflow problem. I wonder if there could possibly be an issue with the intake cam timing solenoid? Try moving the wideband sensor to the opposite bank to see it the same thing is happening over there.

mx594

iTrader: (4)

Alright this is getting more and more frustrating every day. I finally got a chance to mess around with the car again last night in an attempt to figure out what is going on with this lean spot. I did some logs while driving in 2nd gear from about 1500 rpm to 3000 rpm at light/medium throttle. After reviewing the logs I couldn’t find anything that looked out of place. The injector time doesn’t dip in that spot, the intake cam advance is following the target closely, the MAF sensor voltage doesn’t spike (neither does the MAP reading). I don’t get it. The A/F goes all the way off the scale at 18:1 briefly, then comes back. This is actually centered at 2250 rpm, not 2400 like I thought.

Later I tried to do the no-load tuning again, just by sitting in the driveway and revving the engine and trying to hold it in a zone center, then tuning the VE table. This was a disaster. I can hold the engine at 2250 rpm, and it will just sit there and stumble with the A/F gauge pegged at 18:1. No matter how much I add to the VE table in the zone that the engine is in, it will stay at 18:1. The battery voltage and fuel pressure stay constant, so I know that’s not the problem.

The really weird thing is that while I am trying to rev the engine, I can hold it at about 2000 rpm just fine and the A/F looks good…then when I gradually apply more throttle it will gradually rise to 2250 and stop. As I keep applying more throttle, instead of revving higher it will just keep getting leaner, until it finally “breaks through” and the revs jump up and it gets through the lean spot. So basically I can hold it at 2250 rpm with two different accelerator pedal positions – as I increase the throttle it will get leaner instead of revving higher. 2250 rpm is right on the border of two cells (2000 and 2500). Thinking that I may just need more resolution in that area, I added a 2250 rpm row and tried to hold the engine in the center of that cell while adjusting the VE, but again even after adding like 20% to the VE table in that cell, I still have a lean condition.

I guess the next thing will be to either swap the wideband to the other bank and see if I have the same issue or get a dual wideband system from Haltech. It seems like it almost has to be an injector issue, but it’s weird that it is so rpm specific.

I also noticed that the internal MAP sensor on my Haltech is a bit off. It reads about -1.5 in-hg when the engine is off. I checked it with a hand vacuum/pressure pump though and it seems to be pretty accurate at -20 in-hg and 20 psi.

Dynosty

iTrader: (27)

If all of the inputs/outputs of Haltech are responding properly, then it must be mechanical.

As for the map sensor, atmosphere is not always 0. You can use weather.com to find current pressure for your region, or a good thermometer will include baro as well.

mx594

iTrader: (4)

Thanks Hal. Is it possible for a fuel injector to fail at a specific RPM (or more generally, a specific frequency)?

As for the MAP sensor, you are probably right. I wasn't too worried about it because I verified its accuracy at multiple different boost levels.

In the Haltech, why do I see different numbers when I display the Injection Time Channel(Inj Time) and (for example) the Injector 6 Injection Time Channel (Inj6 Time) simultaneously? Shouldn't they be the same???

gmx

I have the same problem but not at 2400 it is at 3050 rpm it will get too lean no matter what I do to the map. Its strange because it only happened at 3050 rpm and between
-8 psi to -2 psi but in boost and every where else is fine.

I tried everything checked all the sensors, the only thing that I haven’t checked yet is the injectors but I'm not sure yet.

if you found away to fix the problem Please post it here .

mx594

iTrader: (4)

I moved the wideband to the other bank and I get the exact same readings.

Dynosty

iTrader: (27)

It is possible for an injector to have a problem at a specific place. If everything electronically is staying consistent and the problem remains, it leads me to think it is a mechanical problem. I can't see fuel pressure dipping at once specific rpm and coming back after that, but have you monitored it?

Regarding the injector time difference, don't forget the injector dead time if you compare fuel base to actual injector time. Feel free to log all of them and email to me (raw, .csv format) along with your current map and I'll have a look at it.

mx594

iTrader: (4)

I had my girlfriend slowly rev the engine up to 2250 rpm and hold it there while I watched the fuel pressure gauge on my FPR (damn drive-by-wire). There was barely any fluctuation in the fuel pressure, maybe .5 or 1 psi at the most, which was probably due to changing manifold vacuum.

You are absoluteley correct about the difference in displayed injection times. The difference between the two was about .4X ms which is exactly what the injector dead time is set to. I can't believe I didn't think of that!

So in summary:

• My Injector duty cycle does not drop (I will log it again, just to make sure)
• The intake cams are following their target advance values quite closely, according to my logs
• My fuel pressure does not change during the lean condition
• My MAF and MAP readings do not dip during the lean condition
• Both banks are suffering from the lean condition, which makes it hard to believe that it is an injector issue

I think I will try adding a 2250 rpm row again and see if I can't get it to at least run stoich in that zone, no matter how screwed up the fuel curve may look. Is it possible to just have a big bump in the VE curve at a specific RPM? Everything I have learned tells me that the VE curve should be smooth, with no sharp lumps or dips.

The air intake on the passenger turbo is about twice as long as the one on the drivers side turbo (JWT style, except with two separate filters and a post-turbo MAF). Could that possibly have an effect???? I am not ruling anything out at this point!

str8dum1

iTrader: (11)

have you tried the older firmware just for grins?

mx594

iTrader: (4)

No. I didn't bother because I had the same lean spot with the UTEC and even with the Osiris. So I am positive that it is not a problem related to the Haltech.

mx594

iTrader: (4)

I finally got a chance to do some more street tuning this weekend and the lean spot continues to plague me. I attached a sceenshot of a log in which I accelerated in 4th gear from about 1000 to 4000 rpm while trying to keep the engine at -17.5 inHg. The orange trace is the air fuel ratio. You can see that there are three distinct lean spots, the biggest one being centered at 2400 rpm. The A/F goes all the way up to about 17.7:1. You can see how I have adjusted the VE map (Fuel-Base) to add a significant amount of fuel at this spot, and yet it does nothing. All it does is make the car run rich just before and just after the spike. I can feel all three spots without even looking at the gauge becuase the engine loses power.

The lean spot is primarily at high vacuum. If I perform the same exact "pull" but keep the engine at -5.8 inHg I don't have any problems...the VE numbers in that column are smooth and progressive. The higher the vacuum gets, the more pronounced the lean spot gets.

I switched the wideband to the other bank and had the same phenomenon.

The manifold pressure does not spike
The injector pulse width does not drop
The MAF sensor voltage does not spike

I did notice that intake cams go from 0 advance at 2000 rpm to 15 degrees at 2500 rpm, but that doesn't explain why it only happens at high vacuum.

Ideas?

str8dum1

iTrader: (11)

my car has the same 2500ish rpm issue actually. Mine isnt leaning, but it definitely hesistates/bucks when going thru those load cells.

Sharif couldnt tune it out either. Super annoying actually because 70mph is ~2500 rpms in 6th gear. So the car would continuously hesitate as the cruise control would modulate the speed around 2500 rpms and hi vacuum....

argh. Add that to the cam phasing problem that other 2003 users are having makes me wonder whats up with 2003 G35's

mx594

iTrader: (4)

It is definitely some kind of mechanical issue becuase I had the same behavior with the stock ecu/UTEC. The only difference now is that I can datalog much easier with the Haltech.

I think I have proven that it is not a result of a sudden increase in airflow to the manifold, because both the MAP and the MAF sensor agree that the flow rate continues to increase at a more or less constant rate through the lean spot. Unless the engine is getting additional air from somewhere else...lower plenum leak maybe?

Could the cam phasing have an effect? If the intake cam advances, does it allow more complete filling of the cylinder? Wouldn't this be refelected in the airflow (MAP, MAP) measurements though?

Is it possible that the lean spot is actually a rich spot? I heard once upon a time that if you are rich enough, an O2 sensor can actually read lean...is that true?

If it is a fuel problem is almost certainly has to be fuel pressure related, since I am seeing the same issue on both banks. Maybe I will put a line on my FPR and tape the gauge to the windshield so I can watch it while I drive.

mx594

iTrader: (4)

Bump. Look at that screenshot! That can't be normal!

gmx

same thing here

mx594

iTrader: (4)

gmx -

Is that the same thing? I have three distinct spikes and they are very symmetric. What RPM is that spike at? What vacuum?

Does anyone know how the injectors are actually driven? Does the current come directly from the ECU? The Haltech sends the signal directly to the injectors right? I am just wonderiing. When you log the injector pulsewidth you are really just logging what the Haltech is outputting and not necessarily what the injectors are actually doing right? Is it possible that something in between the injectors and the Haltech is causing them to do something other than what they are being told to do? Can I measure the ACTUAL pulsewidth of each injector?

I guess the next step is to monitor the fuel pressure as I drive and make certain that it is not changing. Then it must be the injectors...

Dynosty

iTrader: (27)

The injectors are constant-powered via the oem power distribution. The Haltech drives the injectors by pulsing the ground. It has direct control of this, assuming copythrough is not on in those cells. You can verify copy through is not on, if the fuel table shows numbers in those cells rather than the letter 'c'.

You can measure actual pulsewidth with a scope on the injector wires.

Have you tried setting cam timing map to 0 and seeing if this makes a difference? Have you traced cam timing?

rcdash

iTrader: (18)

I wonder if this is something off with transient throttle response. This is with the throttle slowly rising and then holding steady but the AFR goes lean and then comes back on its own, correct?

Or if you hold at 2400 rpms, does it stay lean? If so, you can just tune it out (add fuel), correct?

mx594

iTrader: (4)

I am not using copythrough anywhere in the VE table. The only place I am using copythrough is on the timing in vacuum.

When you say OEM power distribution, you mean that the injectors see a constant 14V while the car is running? So in theory the voltage I log with the Haltech is the same voltage that the injectors see? I have logged the voltage during the lean spot and it doesn't change. Do you think it is possible that the voltage at the injectors is dipping at these spots and causing a momentary increase in latency?

I have tried adjusting the intake cam target timing so that it advances more gradually (sooner) but it did not make a difference. Will it hurt anything if I set the whole area of the intake cam table to 0 and re-test?

I have also tried turning off the copy-through on the ignition timing and adjusting it so it doesn't advance from 15 at 2000 to 35 at 2500 like the stock map does, but this didn't help either. (Sidenote - when you use copythrough on timing and the Haltech displays the 3D graph in white, are those the actual values for the lowdet timing map from the factory ecu, or are they just built into the ECU manager software as generic representation of stock ignition timing?)

I might try monitoring the voltage at the injectors while I drive.
I might also try hooking up a scope and measuring the actual pulsewidth.

Could it be an alternator problem? A ground problem?