AEM CAI:LTFT and IAT Mystery Solved

interesting point though I don't understand why the extension of ground on pipe for MAF can be a problem since it's floating.

That make alot of sense.

If the stock mass is mounted on an insulator, and then mounted on something conductive, but insulated you could have a voltage on the conductor.

You would have thought that if the MAF is grounded to its metal base that this would have bled off any charge. Of course if a radio can get messed up from emf from the alt. then a big antenna like the CAI could soak up alot.

Good find.

well yeh if the floating ground area is huge then it can act like an antenna picking up noise around..

Wow good article. Makes a lot of sense when it comes to nailing down the scaling on my AEM. I have the had the thing for almost three years now and I have been running into a very similar problem. Scaling it should be the same each time no matter what from one map to another, I am finding that ti changes very minutely each time.

Someone over at NASIOC also found out that grounding matters, just recently:

Stumbling/Hesitation Problem - MAF Grounding Error (read everything first) - NASIOC

Different symptoms, but same root cause.

I had to kick myself because I ran into a related problem about a real ago, but never figured it out until I read his thread... My car started acting like I had a vacuum leak, and I'm sure now that it was because the paint that was insulating my homebrew CAI's MAF housing from my chassis finally wore through and started conducting.

subscribed

Interesting find.. thanks for sharing.

I have the same intake and while my fuel trims are pretty stable, I have noticed the illogical behavior of the IAT increasing during a WOT pull. On average, I seem to gain 2 - 4* depending on the duration of the pull. My IAT also seems too much higher than the outside ambient temp than it should be.

Did you notice similar IAT behavior during a WOT pull?

What about the correlation between ambient temps and measured IAT during cruising at a speed that should results in close values? Did it change after your mod?

Was this affecting your A/F Learning D? Do you have a learning view before and after?

I'm having the same problem over on the NASIOC side. I tried two methods:

1.) Isolating the maf using nylon screws and electrical tape along the bottom of the maf

2.) Wiring in a ground to the intake manifold or chassis

Right now I'm experimenting with option 1 as I'm not sure what the long term side effects of grounding the maf like you did will be. It's not the way the subaru does it. The metal plate/screw hole on the maf has continuity to pin #4 and is grounded through the ecu. If you measure the resistance between pin 4 on the maf cable and chassis ground it should be < 1 ohm. However, if you take a normal electrical cable and touch one side to the intake and other to the maf tube (unpainted part), you car will stumble and your fuel corrections will swing, meaning there is a current passing through as NSFW pointed out. This can't be good...

I'm amazed that it has taken this long for this situation to be revealed. I've had an AEM CAI on three cars over the past 7+ years and other than it being bashed every chance people get there is little to no other conversation about it.

But this is a lot more complicated than it appears... the LGT is one body, WRXs and STis previous to 08 are other bodies, and the 08+ WRX/STi's GR/GV bodies a completely different situation altogether. The LGT and others have fenderwells relatively sealed against the engine compartment. But the GR/GV is not... the widebody fenders are totally open to the engine compartment and any normal CAI placed in the fenderwell is nothing more than an illusion of a CAI. This fact led to my fender modifications that allow true ambient air for my CAI's filter to ingest.

When I came across your NASIOC thread I didn't want to pollute it with my own data, and merely interjected my insulating method in case it might help. I then kept quiet because, again, it is Bad Noodle's thread and his issue... if what I found were going to be exposed there then so be it. But that issue and MY issue and findings while apparently connected, are different, so I posted my STi data here. I'm glad you are also working on this from another angle. It has to be good.

Having worked in electronics a lifetime, once revealed, the full impact of the metal-tube-CAI was a big Ah-Ha for me. Grounds are one subject for "normal" stuff we find in everyday life... but a completely different one in other environments, where low voltages and higher frequencies are involved. As some have mentioned, a big "antenna" like that CAI tube in the proximity of such an electronically noisey area will affect the data. Having those sensors now isolated from that antenna, as well as now being inside a grounded electronic shield, provides the best environment possible for clean data.

The impact on LTFT has followed logic here... the lower the voltage the more the influence. LTFT range "A" has been affected the most and "D" the least. But I've only been watching this for the past month, and haven't been in a hurry. In fact, as I said, the MAF Cal impact was an unexpected surprise... after all, I was chasing a different rabbit. Once I saw those changes, too, coupled with changes in temperature of fifty and more degrees as well as transitioning to winter-blend fuels, I've just been patiently and slowly trying to quantify things. I've been happy just to get the temperature monkey off my back. But if what has been happening continues I may finally put my MAF Cal to bed as well.

It is amazing that we have almost simultaneously followed two distinctly different paths to the same unexplored clearing in the forest.

What were your LTFT before and after? Anyway you can post a learning view? I'm just curious to see the difference.

SeeeeeYa, question on your solution as I'm trying to understand what exactly is going on here. You said you isolated the maf from the pipe using the mica wafer, but what about the metal screws holding the maf to the pipe? There should still be continuity from pin 4 on the maf to your ground as the metal screws touch the metal side walls on the maf and the thread in the pipe. So what do the mica wafers actually do? Do they isolate the maf from the radio antenna thing your were talking about? Could you explain that to me in more detail, I don't quite understand the effect. Thanks.

Also, what are your thoughts on the ecu getting using a ground that it wasn't designed to use?

Bad Noodle,

I just set out to isolate the sensor from the intake pipe... thermally. I also didn't want to disturb the physical relationship between the two. The very thin mica transistor insulators fit the bill.

First I removed one screw from the sensor, to see what I had to do with the mica. I trimmed the mica with scissors to fit the application, and opened the hole in it by a minute amount... leaving the hole too small so that the screw and mica were in an interference fit. I slipped the mica under the sensor, placed one on top, and gently forced the screw back into place and tightened it securely. Then I repeated the same thing on the other hole.

Then I used a meter to see how well I'd done, and was pleasantly surprised to find the sensor was electrically as well as thermally isolated. If there had been continuity I'd already planned a fix. It's simple, the same process necessary for mounting certain transistors.

Without touching the tune I drove the car normally. Normal for me means constantly toggling my AP between parameters. At some point my trims suddenly changed. This isn't totally unexpected as they do that, which is one of the reasons I'm always watching them.

This is where I mention a Learning View... when I tried to use my Tactrix I found Cobb had locked the Tactrix out... so I sold it. Now I wish I hadn't because I think Cobb later unlocked it. However, I don't need a LV as I have that info and more available via the AP and/or ATR. I always know what my LTFT and FKL is.

FWIW, I have no knock... or a very rare shift knock. I have no learned timing compensations. That has been the number one goal of my tuning, and it took a long time to get the best performance without any knock. My second priority is a good MAF CAL.. which mostly means not worrying about LTFT range D doing something bad behind my back. My AFR of 11.54 at maximum loads means there isn't a lot of room for sudden unanticipated fuel removal.

For examples; LTFT A has always been -3.xx, which has always seemed wrong. Range B has been stable at either 0.0 or something within a couple tenths of that. C was -1.3 to -1.8, no matter how I played with it. And D, my only real concern, has been minutely modified dozens of times, only to sit at a comfortable number for some time only to change by 3 or 4 digits within a minute on some ride somewhere. Maddening. In the end I quit fooling with A, B, and C. they were close enough... not really, I tried, but there was always a difference between what I did and what I got. As long as D stayed under 1% and was positive, or -0.5 or less and was negative, I was OK with my trims.

You can understand my surprise, therefore, when after isolating the sensor, on the first day's driving my LTFT changed to 1.x, 0.0, 0.0, and 0.0. The next time I drove they changed some, not a lot, but some. I was caught between disappointed and WTF. I began to see what I'd stumbled on and finished the concept with the ground strap.

Now things really came into focus. My logged IAT didn't have a mystery component, and my LTFT stabilized... at numbers I liked. I made a small change in the MAF CAL and it was reflected in the LTFT. My logs have FAR less runniing corrections. LTFT table takes far longer to populate, and then stays there. Checking my LTFT just now the numbers are A 0.5, B 0.0, C -0.1, and D 0.6. I know where the D error is but I'm really in no hurry now.

And I'm aware this is not "scientific" evidence. But science is behind it. As people explore this more will be revealed.... either I'm wrong or a LOT of people are going to reveal similar results as well as other corollarial artifacts.

Given the inferred proposition here, that spurious voltages are being picked up by the MAF/IAT sensor that influences their output, it must be conceived that the OEM installation is also affected, although at a far smaller degree. But, by grounding the metal pipe, and providing a defacto capacitive decoupling via the miica spacers for any high-freq garbage on the sensor liine, we end up with not only clean output signals from the sensors, but also sensors now operating in an EMF free environment.

I still think it runs smoother. Normally this kind of subjective seat-of-the-pants stuff is worthless to me. But sometimes it starts there, and the objective data follows later.

EDIT:

As for why the screws don't touch the MAF ground plate, it is the AEM-supplied screws, I think. They are small and allowed some mica to be drawn into a circular insulator in the small space around the screw, which was my intention with the interference fit. The mechanical strength of the mica allows for tightening the screws very tightly for a secure mechanical installation..

subscribed

As I said in the other thread, I didnt want to mess with a thread hijack, so I will ask about the grounding over here. When I took off my stock exhaust manifold, to put on headers, I removed a very short ground strap from the chassis to the manifold. When I saw this thread I decided to ground the top tab on the AEM CAI to the chassis bolt/CAI shield, as you did, with that strap. We took a small amount of paint off the tab but I did not remove any from the shield itself. Should I? As I think about this I think that I should, take a small amount of paint off around the hole, on both sides of the shield. Right now, I think I am relying on the bolt to be the ground at this point and that doesnt seem like a good idea.

What is your opinion?

Are the wafer boards/isolating the MAF necessary?

Thank you again. I am going to be going in for a tune very soon and wanted to try to eliminate as many potential problems as I can ahead of time.

I'm not certain what you did, but grounding the AEM pipe is what I did, via the new bolt in the AEM tab to the lower shield bolt. First the MAF should be insulated from the AEM's pipe, of course. Otherwise there will be ground loops. Both the insulation of the MAF sensor plate from the AEM mounting surface, and THEN the grounding of the pipe are necessary to replicate my conditions. Insulating the MAF ground from the AEM pipe removes a wide-band EMF entrance, and grounding the AEM pipe provides an EMF-free environment for the MAF/IAT sensor to collect data in.

The changes this brought remain. Fuel trims that were negative are now positive... accounting for the new smoothness I note in day-to-day driving... from more accurate fuel trims if not other ways. The results are subtle, not drastic, but make a positive difference in how I drive and how I enjoy that driving.

Here is an UPDATE... the kind I hate, but necessary.

As I've outlined, I was looking for a problem concerning IAT sensor readings and found reasons to include my MAF sensor. But, some of the conclusions since extrapolated has relied on someone else's finding/conclusions that do NOT fit my 11 STi. Specifically, the metal mounting points of my MAF/IAT sensor are not connected to anything, that means they are not part of any circuit including a ground circuit.

This complicates things in one way, and simplifies them in another. While no ground circuit is involved, thus no ground loops, the observations stand concerning my modifying my MAF/IAT sensor/AEM CAI installation. Given that the sensors have the same voltage supply as the injectors it is probable the supply is dirty. I'd have to get out the scope to verify this, and may, but for now I'm happy for the way things stand, even if the "why" is still uncertain.

Here is some input from my legacygt.com thread: http://legacygt.com/forums/showthre...ms/showthread.php/aem-cai-ltft-and-iat-mystery-solved-p4125907.html#post4125907

I had a chance to try insulating the maf sensor this weekend. I used a slightly different method which might be of interest (as it is less work).

I placed small squares of 1mil Kapton tape over each screw hole on the intake. Then I inserted the maf sensor. With the sensor as a guide I poked a small hole in the middle of the tape. Next I installed nylon bolts through the tape. The size is M4x0.8.

I haven't had a chance to make to make precise measurements yet. What I've noticed though is 1) the hickup I used to have when the rad fans cycle is gone (I had previously assumed it was just the blast of air upsetting the maf) 2) the car stays in lean burn more frequently (I'm mapped to run leaner than stoich under cruise conditions) 3) the engine runs smoother. number 3 is very apparent at a few RPMS as I'm using a single mass flywheel + hardened tranny mounts.

it is too early to tell (only 40km), but fuel economy seems to have improved. after I've driven a couple thousand with this modification I will know for sure.

Click to expand...

Here is some input from the similarly themed thread on NASIOC: Stumbling/Hesitation Problem - MAF Grounding Error & Bad MAF Symptoms - NASIOC

I've installed the IAT to the charge pipe to help with speed density fueling with the MAF still located in a short ram intake. The idle with that setup was inconsistent at best. After reading this thread, I grounded the two pipes to the intake. After tuning, that returned the Five-O 1400's to manufacturer's latencies. Big improvement in driveability.

Click to expand...

Based on nothing but theory I have added a filtering capacitor between the #3 sensor connector wire and ground, to minimize influence from stray EMF, or from the injectors whicih also share this line. I don't see one in the diagrams, and it seems there should be one. Since the voltage there follows alternator output, I'm assuming the usual dirty DC. I'll scope it shortly to be sure.

My MAF Cal is finally, after years of chasing it in circles for the degree of correction I seek, providing me with results that are consistent and acceptable. The relationship between throttle pedal and motor has become a thing of beauty, inviting the use of embarassing hyberbole.

I'm relieved, honestly, to see similar results from others. And I think more is yet to come.

subscribed

First post edited for corrections and updates.