Search RomRaider





  • View Page
  • Page History/Versions
  • Print Page

Nsfw Snippets

Occasionally I write things on forums that I may want to refer to again. Occasionally I put them here too.

1. Speed Density and MAF

The ECU has to calculate how much fuel to spray for every cycle. To figure that out, it has to know how much air the engine is inhaling.

The stock ECU uses a mass air flow sensor (MAF sensor) to figure out the air flow rate. It divides the flow rate by RPM to know how much air is going into each cylinder with each cycle. From that, it can figure out how much fuel to spray to match that air with the desired air-fuel ratio (AFR).

“Speed density” is an alternative to the MAF sensor. The ECU looks at RPM (speed), and manifold air temperature and pressure (density). It uses those tow things to look up the predicted air flow rate in a table (called a volumetric efficiency table). Basically, the VE table says “at this speed, and this density, the engine will pull in this much air.” And from that, the engine can figure how much fuel to spray to get the desired AFR.

2. ATP’s EWG exhaust housings

I think that one of the main benefits of an EWG is to vent unnecessary exhaust gas before it has to go through the restriction of the exhaust housing. Therefore, I am not fond of ATP’s EWG setup. I got my 3076 with the IWG option, and will probably add an EWG to my UP later.

The exhaust housing inlet is 1.75″ and it probably constricts a bit further before it reaches the spot where ATP puts the EWG. By putting the EWG on a 2″ up-pipe, I’ll only be forcing about half of the peak exhaust flow through the 1.75″ restriction. Putting the EWG on the exhaust housing means that at peak flow, 100% of the exhaust still has to go through that restriction. It just doesn’t make sense to me.

The IWG with the supplied actuator works well up to at least 23psi peak boost. Probably beyond, but my clutch started slipping when I turned it up further so I can’t say for sure yet.

3. Reality versus the Open Loop Fueling Table

The ECU looks at how much air is coming in (that’s what the MAF sensor is for) and it looks at what AFR you’ve “asked for” in the OL fueling table, and it makes the injectors spray the right amount of fuel to achieve that AFR.

The following things will cause the actual AFR to deviate from the desired AFR:

  • your MAF scaling might be inaccurate
  • your injector settings might be inaccurate
  • your injectors might not flow as advertised at all RPM and Load combinations
  • you might have a leak somewhere
  • your might have a FMIC that “absorbs” air while boost is building, causing a rich dip
  • probably other things I’m forgetting
  • probably other things I don’t know yet

It’s common for tuners to fill that table with the values that they want to see on the wideband, and then while logging, tweak cells in the table until the wideband itself shows the values that they want to see. In the end, the values in the table may not match the values shown by the wideband. But if the values in the wideband are what you want, that’s more important. But, the closer the table is to reality, the easier it is to understand the tune.

It’s surprisingly hard to get the values in the table to actually match the values in the exhaust. You’d think that getting the injector scaling and MAF scaling correct would be sufficient, but unfortunately it’s not that simple. Kind of a pain in the ass, actually.

For months, I used a spreadsheet to generate a ‘desired AFR’ column with values of “max(OL_TABLE, 11.2)” so that I could richen up the OL table in places where my injectors weren’t behaving like they were supposed to it. Then I’d compare the wideband readings with that synthesized column rather than comparing them to the actual OL table values. I got the right AFRs in the end, but there were patches where the fuel table didn’t match reality. Switching to ID 1000 injectors helped immensely but there’s still some funny stuff going on that I haven’t figured out yet.

4. What to look for in your AF Learning values (Cobb calls them long term fuel trims)

The closer to zero, the better. If they’re under 5% they’re close enough, for all but the last one. The highest fuel trim is a little more interesting - the 4th or 5th depending on the ECU, the range may be 40+ or 50+. That one impacts open loop fueling.

Imagine you’ve just flashed a new tune and you go log some pulls to verify that your OL AFRs are where they should be. And they are what they should be. Then, over the next week or two, that last fuel trim drops to −5%. Now you’re 5% leaner at WOT, which could be bad.

Or suppose you haven’t flashed in a while, but you do some logging and find that your AFRs are where they should be. Have a look at your top-end LTFT, because that’s how far off they’re going to be right after your next reflash.

If you can keep it close to zero, then none of this matters much.

Some people like to raise the MAF range of that last trim to 100+. You’ll never be in closed loop at 100+, so it will always stay at zero. On one hand, you ECU will no longer be able to compensate for varying amounts of oxygen in your fuel. On the other hand, it takes a long time for that trim to settle in, so the ECU isn’t very good at compensating for it anyway. I dunno how much it matters, really. I left mine at the stock value of 40+ and just got the MAF scaling dialed in well.

5. Turbo model numbers, names, sizing, etc

The turbo designations are mostly just model numbers.

Mitsubishi makes the ‘G’ series, for example 16G, 18G, 20G. The “big 16G” is actually almost as big as the 20G. No, it doesn’t make much sense.

Garret turbos are mostly numberd like GTXXYYR where XX refers to the size of the exhaust housing (roughly) and YY is the size of the compressor wheel exducer (the larger section) So, GT3076R = GT30 hot side, 76mm compressor. GT3071 = GT30 with 71mm compressor. But the GT3071 and GT3076 hot sides are not the same. No, it doesn’t make much sense. But GT35s are bigger than GT30s, and GT28s are smaller. There is no GT29, or GT31-though-GT34. But there’s a GT40 and GT42. Kinda random.

IHI turbo numbers just get bigger every time they come up with a new one. My 05 LGT came with a VF40, later ones had a VF42 or VF46 (I forgot which) and there’s a more recent WRX that comes with a VF52. The number has nothing at all to do with the size.

Precision turbos are (if I remember right… if not, someone will correct me…) numbered for the size of the compressor inducer (the smaller part of the wheel) and the turbine exducer (the larger part). The PTE 6262 is around the same size as the GT3076. There’s a PTE 6762 that’s similar to a GT3582, or something like that. Ironically, the only company whose model numbers actually make sense is the one that I am least familiar with.

And then there’s Borg-Warner. S200SX, 65–78, WTF. I don’t know where their numbers come from.

The closest you can get to a single number that describes a turbo is the flow capacity of the compressor, which is expressed in pounds of air per minute, or sometimes cubic feet per minute, or cubic meters per minutes.

18G = 40 lb/min 20G = 44 lb/min GT3071 = 47 lb/min (but really tends to perform about like a 20G) GT3076 = 52 lb/min GT3582 = 65 lb/min

Flow capacity is kind of a fuzzy number, there’s no real standard for it. They’ll all flow more than advertised but they get less and less efficient, blowing hotter and hotter air. Hotter air = less dense and more knock prone = you get less in the combustion chambers and you have to retard timing, so you end up losing power at some point.

It’s all kinda vague, but if you understand the basic concepts, you understand enough to make sense of things. Or if you want to dig deeper, you can…

6. The stock front O2 sensor is a wideband.

If you move the front O2 sensor to the downpipe, it’s very clear that it’s a wideband. It doesn’t agree perfectly with my PLX/Bosch wideband, but with minor tweaks to the sensor scaling that could be fixed (and there’s no guarantee that my aftermarket sensor is perfect either).

However, the stock sensor is subject to a minimum AFR issue reading. Some people have logged as low as 11.04 or something, others only down to 11.25, but there’s definitely a ‘floor’ there.

If you’re willing to tune for 11.5 AFRs, you could probably get away with using the stock sensor alone, as long as you put it in the downpipe. However at the time of this writing I don’t know if anyone has actually tried that.

7. Big turbos for daily drivers?

DD means different things to different people. I think it mostly comes down to how much shifting you’re willing to do to stay within the powerband.

For some people, “DD” means “I want power right away with no lag or downshifting.” These people should go with a 20G or probably smaller.

I have an ATP 3076. If I want to have fun, I have to be revving 4000 or higher. (Above 6500, it fades a little, but not a lot.) Tons of people around here say that means the turbo is too big to be any good for a DD. But, I think that downshifting is a reasonable price to pay for extra power. I didn’t get a bigger turbo to spice up my commute, I got a bigger turbo for those times when I can really use it.

With ATP’s 35R you’re probably looking at 4700ish for best results. The 4700–7000 range with that turbo will probably be more fun than the 4000–7000 range with the 30R version, but it will probably take you more time/practice to get accustomed to keeping the revs up in that slightly higher and slightly narrower range.

I can’t pick the turbo for you, but hopefully this makes it a little easier to decide.

8. Dynos and drivetrain losses

Bear with me for a short note about how dynos work, and then I’ll explain the relevance… Most dynos don’t measure power and/or torque directly, they measure something else and calculate power and torque from that. For example you can measure power by measuring the time that it takes for a car to spin the rollers from X RPM to Y RPM. If you know how much the rollers weigh, then you can calculate power. If the dyno measures RPM along the way, then it can also calculate torque. For other dynos, time is not a factor - they apply a controlled load to the motor and ramp up the speed over an arbitrary time interval.

If the time interval is the fundamental thing being measured, then rotational inertia will have a big impact on measured power. If you put on heavier wheels, it will take power to spin them up, so it will take longer for the rollers to reach their top speed, and that will show up as a lower number on the dyno chart.

If the time interval is not the fundamental thing being measured, then the impact of rotation inertia will vary - if the speed is ramped up slowly of a long time interval, rotational inertia will have almost no effect on the measurements. If the speed is ramped up over a shorter period of time, it will have a larger effect. So it depends.

Incidentally, at least one of the dyno manufacturers (Maha?) can set the dyno so to measure the time it take for the car+dyno to coast to a stop after a pull. Then they have the software compare that to the time that it takes for a dyno to spin down without a car on it, and they use the difference to estimate drivetrain loss.


PTFB = Part throttle, full boost. It’s a standard side-effect of using a manual boost controller (MBC).

MBCs are convenient for testing your tune at various boost levels, because they give you super-simple adjustment of your boost level - just twist the knob. But they do have the drawback of giving you full boost when you’re at less than full throttle. The MBC doesn’t know where your throttle is at, it only knows if you’ve reached full boost or not. In daily driving this means that when you want mild acceleration, you’ll find yourself lifting off the throttle as boost builds. With my 3076-sized turbo, it’s no big deal, but people with smaller turbos sometimes find it annoying since it happens more frequently.

In some cars, PTFB can cause the car to lean out, because the throttle position is used to estimate air flow. In modern Subarus (the ones you can tune with RomRaider), PTFB does not affect AFR at all. Modern Subarus use a “mass air flow” sensor (MAF sensor) for fueling calculations. The sensor tells the ECU how much air is flowing into the motor, so the ECU injects the appropriate amount of fuel to achieve the target AFR. It’s that simple. The throttle position is not a factor in these calculations, so using an MBC won’t affect your AFR at all. If your O2 sensor isn’t reporting the AFR that you want, you have a tuning problem, not an MBC problem or a PTFB problem.

Timing is based on airflow, not throttle position, so you’ll have the appropriate timing whether you’re at full throttle or part throttle. If your timing is off, it will be off with or without the MBC - either way, you just need to fix the tune.

10. Knock Correction Ranges

I don’t know for sure why the stock tunes disable some forms of knock control at high loads and high RPM. However I think I can guess with some confidence…

Most people seem to think that it’s because the knock sensor isn’t reliable at high loads or high RPM. That may be true. However lots of people, myself included, have found that it works fine. Perhaps the OEM tuners know that, statistically speaking, there’s some percentage of cars where it’s not going to work fine. If so, perhaps they disabled it at high load and high RPM to avoid having that percentage of car owners complain about power loss due (due to timing pulled in response to false knock).

Subaru is confident that the OEM knock control system would work well enough - with the stock tune - even when some aspects of knock correction are enabled only within certain RPM and load limits. (We know they are confident because they provide a warranty!) Basically, they decided that they don’t need knock information from higher RPM/load ranges to determine that you’re running on crap fuel.

Perhaps most importantly, I suspect that they OEM tuners tuned the higher RPM and load regions extra-conservatively for the above reasons. Since they knew those regions were tuned conservatively, it’s no big deal to disable some of the knock control mechanisms in those regions.

However, aftermarket tuners get more power by leaning out AFRs and increasing timing to near the knock limit. We get power by giving up some of the OEM safety margin, so I think it makes sense to expand the knock control ranges. Especially if you find (as most of us do) that the knock sensor remains reliable across the entire RPM and load ranges that the motor ever sees.

11. Stock location versus Rotated

There’s so much bullshit floating around the forums about this subject, it’s ridiculous. A lot has changed in the last couple years, and not everyone seems to be paying attention.

ATP 3076 versus rotated:


You get the same power with less boost with a rotated setup, which is kind of interesting. I’m not sure why EFI chose not to run the same boost with the rotated setup. You might guess that it would make even more power. But, they’re smart people so I’ll trust their judgment.

12. MAF Tuning

At some point I’ll do a proper how-to on this. But for now…

In my opinion, the best way to do MAF and injector scaling is to force the car to run in open loop all the time, and the datalog MAF, RPM, WBO2, and “primary open loop fueling.” Log 15 minute of driving around (no boost), open the log in Excel, create a “fueling error” column (WB02 / POLFueling) and make a scatter plot with error over MAF. That scatter plot will basically show you what changes need to be made to the MAF scaling.

Also make a scatter plot with error over RPM, and keep an eye out for RPM-dependent quirks, especially when you’re logging pulls.


RomRaider Forums

Join our forums, the best place to find help and answers!


RomRaider is developed and supported by volunteers working on their own time. To support their efforts please consider making a donation.

Page last modified on May 28, 2011, at 01:47 PM
Powered by PmWiki