Analog signal conditioning

Chris Conlon synchris at ricochet.net
Sun Jun 14 19:35:13 GMT 1998


Ok, got a 2-parter here,

John Lagerquist wrote: (vision_a at srv.net)

> I am filtering all of my analog sensor signals (tps, o2, 
> temperature, etc.) through a LMF60 6th order active low pass 
> filter. I am looking for recommendations on where to set the 
> cutoff frequency.  

Well, I'll take some stabs at it, since noone else has:

Most temp sensors seem to have a time constant between 1-10 sec,
not exactly high BW. Check the specs for your particular sensor
and set the cutoff accordingly. I personally like the faster
sensors; on a supercharged engine with a so-so intercooler it can
make a real difference; I can get a 100F temp change in under 10
seconds.

MAP sensor: I'll bet you could get some good arguments going here.
My feeling is that the cutoff frequency should be related to RPM.
I prefer to sample MAP at high speed (1KHz or more, depending on
the natural frequency response of your MAP sensor), and average it
out over the last N complete firing cycles (2N rotations). You can
tell by now I'm aggressive ISO good throttle response. If you need
to pick a fixed frequency, you could probably pick something that
corresponds to a low RPM, but if you set the cutoff low enough to
avoid aliasing at low RPM, it may seriously hurt response at higher
RPMs.

The O2 sensor line is probably more open than most to induced noise,
but the way it's sampled is very noise-tolerant. If you're running
a typical system that just wants to see stoich crossings every so
often, my guess is 5-10Hz should do you ok.

If you're fighting a noise problem (as opposed to simply trying to
cut the sensor BW down to something appropriate) you might want to
work on cleaning up the supply, at least where MAP, TPS and temp
sensors are concerned.

Martin Hill wrote:

[With respect to converting from AFM to hot wire]
> You will also have to compensate for the automatic richening effect 
> from the overshoot of the flap, which doesn't happen with the hot 
> wire systems.

My *impression/guess* has been that AFM-type ECUs filter the flap signal
pretty heavily, to *eliminate* flap inertia, vibration effects, and
turning/braking effects from the signal. Can someone say for sure what
sort of filtering is applied to the AFM flap signal in any particular
ECU? Relying on flap overshoot to richen the mixture seems like an
ugly hack at best... but then again, what do I know?

   Chris C.




More information about the Diy_efi mailing list