AEM CAI weirdness

Has anyone else out there tuned their car with an AEM CAI? I've been getting some weird results with my car, like the MAF sensor was lying to me, and [EDITED:] back when I posted this, I thought the CAI was the cause.

 

My fuel table is pretty simple. I just took the stock Primary Open Loop fueling table and set every cell with a value lower than 11.2 to 11.2. Put in a floor, basically. But my car kept leaning out halfway through a pull. So finally, acting on a tip from LittleBlueGT and Josh Long (on RomRaider.com, and coincidentally, in my area) I did several pulls in 1st, 2nd, and 3rd gears, starting at 2k, 3k, 4k, 5k, 6k rpm and going to redline, logging MAF, AFR (via PLX's SM-AFR wideband), RPM, knock sum, and not much else.

 

At first I thought it was a MAF scaling problem, so I made a scatter plot showing AFR Error (desired AFR / map AFR) on the vertical axis, and MAF on the horizontal axis.

 

http://i193.photobucket.com/albums/z151/Legacy_NSFW/Tuning/afr-maf-1.png

 

The scatter on the left is open loop. There are rich dips at 174 and 225 g/s. The really weird thing is the vertical scatter near peak MAF. That's not something you can tune out with MAF scaling - one voltage corresponds to a wide range of air flow rates.

 

So I did a scatter plot with AFR Error over RPM...

 

http://i193.photobucket.com/albums/z151/Legacy_NSFW/Tuning/afr-rpm-1.png

 

...and suddenly it all makes sense. Around 5800 RPM, the AFR goes lean almost 10%, meaning the MAF sensor reads low by about that much. But at 4500 and 7000 RPM, the AFR goes rich by about 5%, meaning the MAF sensor reads high by roughly the that amount. I wouldn't have believed it was just a matter of RPM, but the scatter is pretty tight and the number of scenarios I logged should rule out anything else.

 

(For my next trick: a 3D surface showing MAF, RPM, and AFR error. Maybe. )

 

I'm guessing this is caused by the AEM CAI, since that's what houses my MAF sensor. But nobody else has mentioned this sort of thing. Is that because nobody else has noticed, or nobody else has looked? If you've looked, what have you found?

 

I'm probably just going to compensate with the open loop fueling table. Now that I know that the lean/rich variances correlate tightly with RPM, I have no reservations about tweaking the table to get the desired AFRs.

 

I know it's common to fudge the values in the fueling table to achieve the desired AFR, but I had visions of +/- a couple tenths of a point... my WBO2 reads 12.0 when the fuel table reads 11.2. Anyone else seeing errors that high?

FWIW, my knock sum incremented by one (from 3 to 4) during that logging session. Months ago when I saw how lean the car was getting, I backed off the timing to compensate, so it's safe as it is. I suspect it's costing some power though, so I'm looking forward to fixing it.

I never did notice that. But I will try and take some logs for you. FWIW I hardly ever have any logs at only 250 g/s. Serious. They are all over 300 g/s.

 

I can attach my scalar that I use, if that will help.

 

FWIW my pri fuel table is pretty solidly in the 13.2 range, and my AFR from my PLX WB shows a very nice and consistent 12:1. I can post a screen shot of that too if you want.

I talked w/ Mike about this long ago...

 

He said it was simply dealing with the resonant frequencies of your intake...

I never did notice that. But I will try and take some logs for you. FWIW I hardly ever have any logs at only 250 g/s. Serious. They are all over 300 g/s.

 

I can attach my scalar that I use, if that will help.

 

FWIW my pri fuel table is pretty solidly in the 13.2 range, and my AFR from my PLX WB shows a very nice and consistent 12:1. I can post a screen shot of that too if you want.

 

In the winter I got the same effect at 255 g/s, now I'm getting it at 235 g/s. Being at 300+ might affect the resonance, but we'll see.

 

Yes, it would be great to see a shot of your fuel table vs. the logs.

Here they are. FWIW I have recently added a bit more fuel up top for the summer.

 

http://i248.photobucket.com/albums/gg194/littlebluegt/Prifueltable.jpg

 

http://i248.photobucket.com/albums/gg194/littlebluegt/AFRs.jpg

It's interesting that your table varies quite a bit to get a steady AFR... but not in the way I think mine will need to vary. What max loads are you hitting?

It's interesting that your table varies quite a bit to get a steady AFR... but not in the way I think mine will need to vary. What max loads are you hitting?

 

It didn't vary at all for the run. it used to be all the same from 2.5 load and up, but now I am making it slightly richer from 6000 rpm and higher.

 

My loads in the summer are 3.6 to 2.7. Winter is 4.0 to 2.9.

 

If you are looking at my AFRs below 2.5 load, that is done richer to stop shift knock, and to ensure the alky system is turning on lightning quick when loads drop low and then come back up again (like after a shift).

 

My alky system is IDC based, both turn on point and flow rate.

OK, so it only varies by a couple tenths... Mine will need to dip by a full point or so at 5800.

 

Does your CAI look like this?

http://www.legacygt.com/forums/showpost.php?p=2155950&postcount=67

 

I'm wondering if there were two versions, or if maybe the CAI itself is not the source of my problem.

To me, it sounds like your MAF isn't scaled properly.

 

Log voltage instead of G/sec and look for weird dips and dives. Voltage will show you weird spots due to airflow variations. G/sec shows you weird calculation errors.

OK, so it only varies by a couple tenths... Mine will need to dip by a full point or so at 5800.

 

Does your CAI look like this?

http://www.legacygt.com/forums/showpost.php?p=2155950&postcount=67

 

I'm wondering if there were two versions, or if maybe the CAI itself is not the source of my problem.

 

 

Same intake, but mine is blue, goes with LBGT better.

 

Post up your scalar.

To me, it sounds like your MAF isn't scaled properly.

 

Log voltage instead of G/sec and look for weird dips and dives. Voltage will show you weird spots due to airflow variations. G/sec shows you weird calculation errors.

 

I don't think so... Look at the "AFR Error over MAF" scatter plot in the first post. (I see now that I left out the word "error" but bear with me.) The ECU is seeing a constant MAF reading (around 235 g/s) and trying to maintain a constant AFR (11.2) but the actual AFR varies from 10.5 to 12. The same thing was happing around 255 g/s in the winter, with the same MAF scaling. Here's the relevant fragment of the MAF scaling:

 

3.83      3.95      4.06      4.18      4.30      4.41      
189.59    209.77    231.70    255.49    273.15    292.02

 

There's no flat spot there which would allow create constant MAF readings for varying MAFv. The problem is that MAFv stays constant while actual flow (grams per second) rises and falls. The second scatter plot shows that the error (difference between actual flow and measured flow) is a function of RPM.

 

At least, that's how I interpret the data. I'll post a couple more log fragments that illustrate my point.

Actually, since I posted them at RomRaider already...

http://www.romraider.com/forum/topic4071-45.html

3.83      3.95      4.06      4.18      4.30      4.41      
189.59    209.77    231.70    255.49    273.15    292.02

 

Something is very wrong there. Are you sure you have the columns lining up properly? 4.41 v is not 292 g/s.

My bad. RomRaider copy-paste has them off by one, and I shifted the wrong direction trying to undo that, so now they're off by two. Here's a better version:

 

 

4.06      4.18      4.30      4.41      4.49      4.57    
189.59    209.77    231.70    255.49    273.15    292.02

I have an APS 70mm intake and went through a similar process when getting that dialed in. I made a lot of mistakes as I am still learning but eventually figured out to set the entire open loop fueling map to 11.0. Then I logged MAF volts and rpm and AFR with my LC-1 and that showed me where it was off. It was then relatively easy to dial in the MAF scaling and I've got it down to +/- .1 AFR. I would think the AEM should be easier since it is a smaller diamater. I think you are on the right track.

 

Have you checked if the ecu might be applying any long-term learning values to your AFR? Perhaps you could temporarily disable A/F Learning to rule that out? How does your LV look? Do you have it pretty close in closed loop? Have you cleaned your MAF sensor recently? I've seen a dirty MAF sensor throw readings off by up to 10%. Sorry if you've already done those things, just trying to help.

The problem is that MAFv stays constant while actual flow (grams per second) rises and falls.

 

that's what I was saying. If the MAFv stays the same but g/sec changes, then it's a calculation error in the ECU or a faulty sensor that affects the g/sec calculation. Check your intake air temp compensation table for anything funny.

I'm still running the stock MAF compensations. I did experiment with the load comp table, to try to counteract the problem, but the scatter plots in the 1st post were created with the stock settings, shown below.

have you logged IAT to see if it is funny?

 

 

The stock tables were set up for the stock components. Once you start changing out things, the stock calculations need to be changed as well.

have you logged IAT to see if it is funny?

 

 

The stock tables were set up for the stock components. Once you start changing out things, the stock calculations need to be changed as well.

 

Sorry to go OT, but do you think it should be necessary to tweak the IAT map for an APS 70mm as well?

 

I always thought that didn't need to be changed as long as you retain the same MAF sensor components.

 

If so, could you provide any advice or suggest any how-to's for how that's done? That might also help the OP as well.

Sorry to go OT, but do you think it should be necessary to tweak the IAT map for an APS 70mm as well?

 

I always thought that didn't need to be changed as long as you retain the same MAF sensor components.

 

If so, could you provide any advice or suggest any how-to's for how that's done? That might also help the OP as well.

 

I agree with this. It's a good question for this thread though.

 

I'd be shocked to see IAT fluctuating in accordance with RPM, but I'll log that on my way to the car parts place today. And I'll buy some MAF cleaner while I'm there.

Make sure you spray the cleaner up inside the housing. I used to think the IAT probe (the brown/orange probe that sticks out) was all you had to clean. The MAF wires are actually inside the housing. You probably already know that though.

Subaru programmed in IAT based off the OEM intake design and OEM intercooler. If you suck in air through a different spot than stock or have an IC that's different than stock, the Subaru OEM settings in the IAT table are null.

How does an ic void the IAT settings when it's after the MAF?

because the Subaru engineers figured out a calculation for how much heat the IC is able to remove under certain conditions. If you change the IC, you change the calculations.

 

The stock IAT table is such fine adjustment that in most cases it is negligible. But in some situations it may cause funny things to happen. I'm not saying the IAT is the OP's problem area, but it could be.