Kinda. I need to write some fancy code to have it process the incoming signal a lot more than it does presently. This thing is becoming a real PITA, and honestly I am a tad tired of messing with it. Yes, you lose a few ponies in a few spots on the power-band, but the car drives so nice that I just don't care. I have put a little over 10k miles on the thing with no complaints. In fact, the power loss issue only happens if you have the gas 100% opened below ~3200RPM. If you ease on the throttle you don't even notice.
Here was one of the WOT data logs I took. The first one here is the raw voltage values.
The signals that matter in this case are the two MDot ones (mass flow rate). The magenta one is the raw sensor output, and the dark purple one is the post-filtering one that the converter actually processes. This is then used with the IAT value to generate a density correlation and then compute a final VDot value (volume flow rate).
All may seem well here...the crazy input signal is cut nicely down to a fairly consistent average voltage value. This is the entire problem, however. If the sensor's output was linear, I would be a millionaire by now lol. It is not though, and I am obviously not a millionaire. To put it simply, the sensor is a lot more sensitive to low flow rates than it is to high flow rates. As intake pulses pass by the element, the low pressure waves/rarefaction pulses cause a steep drop in voltage whereas the high pressure/compression pulses will cause less of an effect as they increase (confusing, I know). SO, by averaging the VOLTAGE rather than the actual mass flow value corresponding to the high & low peak values, I am actually taking in a below average voltage for the calculations, and thus getting lean conditions. To help folks understand better, below is a diagram of what I am talking about (I'd rather have a pic worth 1000 works than type another 1000).
Here's the deal. As cylinders suck in air, the create a pressure wave over time (short as it may be). This generates a pulse output from the sensor as it passes the element with a high and a low peak. The 2 black lines on the drawing bracket in this flow pulse, and you can see that they are extended to show the corresponding voltages. So, the little filter I put in averages the voltage pulse (green dashed line). This average voltage is taken into the converter, and when you trace it back to the flow value it represents, it is too low. What NEEDS to be used is the blue dashed line which corresponds to the actual averaged flow rate.
There really is no cost-effective analog trickery to pre-process the input signal to give me the proper voltage. I need to remove the pre-filter and have the converter handle the input directly. I already tried once, but that was before I built the new fast-logger, and when it misbehaved I had no idea why. I know how to tackle the problem, and theoretically my new software should have, but I need to go out and log the thing to see why it runs horribly lol.
Anyway, here is a plot of the logged data with things put into units you actually care about.
You can see for yourself how the big resonance spots directly correlate to the lean peaks. I KNOW why it is doing it, and I just need to find time to work more on it.
Step 1: Build MAF Converter
Step 2: ?
Step 3: PROFIT!!!
