I'm datalogging with the stock ECU, so it is using the stock MAF "calibration curve." My MAF hasn't been modified at all. And the numbers I am getting make sense.

 

On the EVO, the MAF response curve gets pretty nonlinear under much-higher-than-stock loads. Fine for tuning but, like Ron said, not accurate for actual measurment.

shiv

 

There is definitely a point where the MAF starts losing counts, which is not fine for tuning. Of course, I'm nowhere near that point.

I'm not really using airflow for absolute measurement, other than for comparison to my previous runs. It is very cool to see the airflow curve and how it relates to RPM, especially once you start bolting on mods like cams or intake manifolds.

Like I said, though, you would be surprised at how closely airflow in grams/rev with a multiplication factor tracks boost level. Obviously, it would have to, since the ECU relies on it for load.

 

I'm no fluid engineer but for a fixed density of air isn't MAF more or less defined by boost pressure?

(boost pressure + atmospheric pressure) X (cross-sectional area of inlet) / (density of air at that temperature)

(neglecting flow issues which is i guess what causes the nonlinearities when pressure is high enough)

 

Yup, you got it. Without a cal curve of the sensor there's no way to discern actual airflow in those high-flow situations-- its deceptive because people look at datalogs of 'airflow' in the nonlinear region of the sensor's range, then try to plot those values on a compressor map. Or use them to estimate horsepower. It's like trying to use a narrowband O2 sensor to dial in fueling under openloop WOT conditions.

 

It isn't like I am datalogging the MAF Hz directly. I am datalogging the airflow in lbs/min, which has been massaged into an airflow number by the ECU. The air temp and atmospheric pressure are factored in, and then the resulting value is corrected using an internal MAF compensation table.

The ECU does NOT count on the MAF being linear, because it simply is not a linear device. The ECU does, however, know the correction factors needed in order to make it linear.

If you really wanted, I could post an excel spreadsheet detailing the difference between the 1G and 2G MAF compensation tables. You can get a 1G ECU working well with a 2G DSM MAF if you swap the 2G compensation table into the 1G ECU EPROM, despite the MAFs being VERY different!

Once again, what I am datalogging is the corrected output of the ECU, not the input to the ECU.

 

This is the basis of Speed Density systems. Like Honda uses.

 

Still, the question that remains is how to verify whether the coefficients applied by the ecu to linearize the MAS output in fact remain appropriate when dealing with frequencies well outside the intended operating range. Can you be sure they are, and if so, how?

 

I can ask someone who knows.

 

So does anyone know, at what amount of airflow our MAF readings/correction factors begin to get inaccurate (in terms of boost psi at stock turbo cfm)? 23? 24? 25? Much higher?