I got tired of trying different restrictor and hose schemes to get rid of the MAP sensor "noise" (which is actually the manifold pulses getting through). If I used a small restrictor, the response was slow enough to cause throttle tip-in hesitation. Without it, it was introducing a ~2% error in my MAP value. I couldn't find a good compromise.

I discovered this when I was trying to tune my VVT by examining AFR's. I noticed run-to-run inconsistencies.

So I built and tested an electronic filter. (For the geeks, it's a 2-pole Bessel). 2-pole filters have a better combo of noise rejection and rise time (or delay time).

See MAP results, before and after. Left is before, right is after. The red trace is the MAP sensor voltage output. Green trace is the AEM MAP signal after the software filtering. (Blue below is RPM) As you can see, the spikiness gets filtered, but the MAP signal sometimes rides the peaks of the MAP voltage (when <4000 RPM), sometimes it rides the troughs (when >4000 RPM). It's some artifact of the smoothing method, I guess. However it's not consistent.

See schematic attachment. The diodes are to bypass the filter circuit when the MAP changes very rapidly. Without it there was a subtle tip-in hesitation.

See attached MAP rise time when I jab the throttle. It's between 60-80 ms.

If the input impedance of the ECU is < 1 Mohm, the total of 3k resistance of my circuit will introduce an error in the voltage vs MAP curve. In the case of the AEM, it has 100 kohm input resistance, and with the Motorola 2.5 bar MAP sensor I had to adjust the parameters in the software (was offset=10, scalar = 2.6, it's now offset = 2.5, scalar=2.58).

 

Nice. A wee bit of overkill I'm sure, but there's nothing wrong with that.

Here's a simple filter for lazy people.

 

I'd like to point out that the simple filter above has a time constant of only 220 us and a -3dB cutoff of 280 Hz (8400 RPM on a 4-cyl). This is too fast. To get a rise time of 25 ms (which I empirically tested is the slowest that will not affect fueling), it needs a time constant of 10 ms, so the 0.22 uF capacitor needs to be increased to 10 uF.

Remember the purpose of the filter it to remove out of band noise (including the pressure pulses in the intake mani).

At 4000 RPM on a 4-cyl, the 2-pole filter (mine) attenuates the pressure pulses 2x as much as a 1-pole filter (Ben's schematic) using a 10 uF capacitor.

 

Cool stuff. I've always noticed that the MAP signal has more noise than almost anything else we track. I switched from MAP-based accel enrichment to TPS-based because of this. I was actually getting MAP-based accel enrichments while just cruising. That makes it pretty difficult to tune fuel.

Need to try these filters. Thanks guys.

 

Here is the effect of the same filter (without the diodes) on the TPS signal in the AEM. The time constant I used was 2x10 ms (2 pole 10 ms each).

I totally hacked up my AEM's PCB layout/grounding and fixed newbie engineer boo-boos.

The effect is much smoother throttle enrichment, most noticeably on very light throttle application at low RPM. The decel fuel cutoff also doesn't jerk in and out anymore on trailing throttle turns with very light throttle application.