Here's some real tuning data...answers and epiphanies inside

[legacy92ej22t]

Hmm, I'm a little confused by the numbers your throwing out here "other Phil" because what we were seeing is that the MAF didn't max out until around 12 psi, and that's on the TD05H-16G! Seems strange that yours is maxing out so quickly and on the VF11.

Any chance you could have a bad MAF or something?

Nobody decided to clue anyone else in?

Well I think there are like 2 or 3 threads going about it actually. I can try and dig them up but I know we talked about it a bit in Free5tyles LegaCU thread (actually I think georryan started the thread).

What did you use to monitor the MAF?

A laptop and Vikash's scan tool program. We were reading it straight off the ECU.

IIRC we also tested Phil car (BAC5.2) and got the same results with his 16G setup.

Do you have any numbers?

If I had a bad MAF, why would it give me a reading at all?

No numbers off the top of my head really. I just know that at about 12 psi on the 16G we were seeing the MAF hit the rail, 5v+, and that the ECU would throw the injectors into 100% IDC.

Your MAF can act funky and still function. I had a bad one once that didn't throw codes and still worked but it would cause hesitation at mid RPM's. It was weird.

Do you have a spare MAF you could try or a buddy with another EJ22T that you could test your results against? I'm not saying that your wrong or anything either, your data is just far different then what we've been seeing.

[THAWA]

110 with wrx tires and a turbo drivetrain is about 5650

[DeusExMachina]

My intake system is stock.

My MAF is functioning normally. I don't know why it would read high linearly like that.

Having a log taking data directly (at 60 Hz, also) from the sensor is more accurate than the scan tool. Did you turn his boost down and see if it was pegging the MAF? I'm betting near redline (and faster than you would be able to see it on the scan tool) he was pegging the MAF also. Even at lower boost than when I peg it.

Also, conditions were a cold night, which has a big effect.

I have a spare MAF, but not with me. The data really isn't that different, just that you seem to think me with the equipment is wrong compared to Phil's car with the scan tool and a different turbo. Apples and oranges.

[vrg3]

Dude, was it necessary to stick a 1319-pixel wide image in your post?

What do the terms "normal duty cycle," "total duty cycle," and "adjusted duty cycle" mean?

Yes, the stock MAF sensor has a very limited range.

And a 3" turbo-back exhaust and a Saab intercooler do provide for a lot more airflow than a stock exhaust and intake.

CFM does increase with RPM, but so does the quantity of fuel injected -- remember that the number of injection events is proportional to engine speed. The second graph seems to indicate that you managed to maintain an air/fuel ratio of about 11:1 up through the entire pull, doesn't it?

[DeusExMachina]

Eh, my monitor's big enough.

Normal duty cycle is what the ECU is running the injectors at, total duty cycle is what it is after eManage adjustment. I'm not sure where you see "adjusted duty cycle", looks like you made it up.

Yes, I did manage to get about 11:1. Thats with about 90% duty cycle, though.

With stock fuel maps (which it looks like I'm going to have to get, because people don't seem at all convinced), as soon as the MAF maxed out the AFR went up towards 13:1. This is not good.

[vrg3]

Eh, my monitor's big enough.

Mine's not. It makes it frustrating to read anything on the first page of the thread.

I'm not sure where you see "adjusted duty cycle", looks like you made it up.

You used the term at the end of the fourth paragraph. I'm guessing it means the same thing as total duty cycle?

With stock fuel maps (which it looks like I'm going to have to get, because people don't seem at all convinced)

There aren't stock fuel maps.

[DeusExMachina]

I'm sorry, I'll make it smaller if it makes you feel any better.

The adjusted duty cycle is the total duty cycle, sorry.

Stock fuel maps meaning, me not adjusting them with the eManage.

How aren't there stock fuel maps, though? There's a lookup table for "this MAF means this fuel", isn't there?

[vrg3]

Thanks, that'd be nice. Or you could just put thumbnails in. If you use Hardy's gallery it'll automatically generate the link for you.

There's a lookup table that says how to convert from a MAF sensor voltage to an airflow. Then the ECU does some calculations to figure out how much fuel to inject, but unless I'm mistaken it does it algorithmically rather than by looking it up in a table. In terms of fuel control our ECU's actually pretty primitive.

[DeusExMachina]

I cut out some of the crap in the middle, it'll be smaller. Also as a note, the vertical white line is the point in the graph the values on the right are from.

Well, I wouldn't think it converted the MAF voltage to an airflow value. It probably just takes the voltage and calculates fuel based on that. It might be a calculation but I'd bet it was a table. It doesn't add fuel based on anything else besides MAF voltage, does it? I assume that such and such MAF voltage equates to such and such IDC.

In all cases, 5.0V MAF = 80% IDC. So its getting that value from something.

[free5ty1e]

Well, I remember once I tried to verify the MAF's limits with my t-bird turbo with a friend and the scan tool and my little crap laptop that I found in the trash. But it seemed as though it couldn't update fast enough when accelerating at WOT. I remember Vikash saying that the ECU won't respond to scan tool data requests when its busy trying to keep the engine from grenading, which makes sense... so I have affixed a 5-pin connector to the MAF sensor lines that I can connect to with a meter and directly see the voltage while driving, and at some point I will try to pinpoint under what conditions the MAF maxes out.

I'm confused about the 80% IDC though. I thought if the MAF maxed out, the ECU threw the IDCs all the way up to full open 100%, thereby just relying on fuel pump flow to keep up with the demands of the engine. Then again, I've never monitored or logged this data so I tend to think there's something to be learned from those graphs. Just not sure what yet.

[vrg3]

I cut out some of the crap in the middle, it'll be smaller.

It still interferes with rendering the page.

A thumbnail would allow people to see the entire image in its original resolution and be much less intrusive at the same time, but still give people some idea what the image is before they load it up.

Well, I wouldn't think it converted the MAF voltage to an airflow value. It probably just takes the voltage and calculates fuel based on that.

It uses the table in the lower left of this figure:

http://www.graphics.cornell.edu/~v/pics/aa694.gif

It might be a calculation but I'd bet it was a table.

It seems to be a calculation based on my examination of the code and my admittedly limited analysis of its behavior.

It doesn't add fuel based on anything else besides MAF voltage, does it? I assume that such and such MAF voltage equates to such and such IDC.

At the very least it has to take engine speed into account! You can't inject twice as much fuel per rotation at 5000 RPM as you do at 2500 RPM just because there's twice as much air per unit time.

But also, it has various little enrichments and enleanments computed from stuff like coolant temperature, lambda, and throttle position derivative. And of course battery voltage affects injector opening time.

From what I can tell, it basically takes the MAF sensor voltage, converts it to an airflow, divides by engine speed, and uses that as its base fueling quantity. If the airflow per revolution exceeds about 100% volumetric efficiency (it's possible that it bases the actual threshold on barometric pressure; I haven't checked), it uses an "on-boost" enrichment in addition to all the other enrichments and enleanments.

In all cases, 5.0V MAF = 80% IDC. So its getting that value from something.

Can you elaborate on how the injector duty cycle is being measured?

[DeusExMachina]

The injectors open when they're grounded, correct? So I'm assuming the eManage takes the average amount of time the injectors are open over RPM, or something like that. Your guess on how its measured is as good as mine.

Would you really think that Subaru would make the injectors static when the MAF maxes out? Of course not, 80% is usually the "safe" value that everyone chooses to run injectors. Supposedly past 80% injectors can become unpredictable but every tuner I know runs it past there with no ill effects. Larger injectors are my primary concern at the moment.

[vrg3]

Yes, they open when the ECU side is grounded.

Injector behavior above about 80% (or 85% or 90% depending on who you ask) is unpredictable, yes, except when you're actually at 100% duty cycle. An injector running at 100% duty cycle is very predictable.

So is your piggyback extending the injector pulse on the output side, then? Like, it's wired between the ECU and the actual injectors?

[DeusExMachina]

Its not wired between them, its wired into them. Because to open them the ECU side is grounded, the eManage grounds them to open them.

[vrg3]

Oh, huh... So it just holds the ground longer than the ECU already holds it? That doesn't upset the ECU at all? Cool.

[DeusExMachina]

Right. I've gotten no complaints.

[georryan]

Wait, so it sounds like one side says that at max MAF voltage the injectors are at 80%, before we thought they'd be at 100%. When does it reach 100%? The voltage of the MAF is the same at the point it hits 80% to the point we think it hits 100%. Is the ecu also looking at AFR and just throwing fuel at the engine, then getting to the point of giving up and giving max fuel when it can't seem to keep up?

-Ryan

[DeusExMachina]

Well, unless the ECU is monitoring AFR and going static on the injectors when the AFR is dangerous, the injectors never reach 100% under the stock ECU's power.

Since I'm asked how I came across finding the injectors are 80% at max MAF, how did you find out the injectors are at 100%? If you say the scan tool, then Subaru definitely built in a fail-safe where 80% actual IDC is "100%", or the fastest the ECU will give them.

A lot of things will say 100% when its actually 80%, due to the unpredictability after 80%. The UTEC does it, my friend thought for awhile the eManage did it (which we proved wrong with this), and I think the WRX and RS ECUs do it according to DeltaDash and other OBD2 programs.

[-K-]

The ECU has no way to measure the AFR.

I took readings at Stock boost, AFR goes up to around 10.5 and stays there. Mods are 3" exhaust and it doesn't feel like 14 psi did.....

[vrg3]

The "100%" reading we got was from a little cheap aftermarket IDC meter that displays duty cycle with a bunch of LEDs.

[DeusExMachina]

Well, perhaps that cheap IDC meter was malfunctioning. Maybe whoever built it had in their design to stop people from going past 80 IDC. I most certainly exceeded my "maxed out" IDC, my AFRs changed after increasing from 80%.

-K-: So what's that stock O2 sensor do?

edit: Before anyone says anything, yes I know its a narrowband and doesn't do much, but at least the ECU will know if its lean.

[free5ty1e]

(in closed loop mode... which is everything but WOT where you really need the tuning)

OK, follow me on this... we now think the ECU never lets the IDC's approach 100% - short of us going to aftermarket engine management and larger injectors (which is the right way to do it but also expensive and out of the reach of the common Legacy enthusiast), how about a pressure switch that just grounds the injector control wires above a certain boost level...?

Sounds simple, sounds crude, but it might give us that bit of extra fuel spray that some of us want...

(Edit: If, indeed, our ECUs never output above 80% IDC...)

[legacy92ej22t]

Whaaa? I've seen my IDC go to 100% on the scan tool.

[DeusExMachina]

Whaaa? I've seen my IDC go to 100% on the scan tool.

Like I said before, its probably "100% of what the ECU wants to give it".

Look at my graphs...5.0V MAF my injectors are 80% requested from the ECU, and I'm giving them 10% more.

[legacy92ej22t]

Ahh, I see what you're saying. OK, I was confused there for a second. Hmm, that's very interesting.