[Diy_efi] About MAP sensing and it's jitter

Alexei Pavlov alexis.pavlov at st.com
Wed Jan 14 15:58:00 GMT 2004


tsokorai at xperts.cl wrote:
> 
> Right now in my test code, one sample is taken for every cylinder, as my
> timing marks on the crankshaft are 4, 90 deg. apart. That makes 8 signals
> every engine cycle (2 crankshaft rotations). Is it a good idea to still
> lowpass filter them?

The theory says that you must cut at half sample frequency. In your case
you should have cut at 180 deg freq, which implies a digital filter,as
RPM changes.
That's why it's simpler to sample at for instance 1Khz, cut at 500Hz
and then applie some simple digital filtering like crankrevolution
averaging.
Personnaly I sample at 300Hz, filter at 150Hz, then average to get one
sample per revolution, then apply simple 2nd order digital filter with
well choosen (1/3, 1/3, 2/3) coeffs. Of course at very high 12000RPM
I will get less then one sample per revolution, but with a 8Mhz 8bit MCU
I can not do better.
Then the problem is that we often need the future and not current
value of the measured value. That's another story.


> It's output now looks almost good to me, considering that my valvetrain has
> some wear.
> BTW, I converted the datalog images on my site from PNG to JPEG as it could be
> the problem that prevented you from seeing them.
> Can you please explain the part of the "more inter-cycle oscillations when WOT
> or at high RPM" ? Wouldn't the higher air speed flatten the oscilations at
> high RPM?

Likely not. I had measured on two oposite engines:
210 deg long 0 overlap
300 deg long 80 overlap

At 6000rpm the oscillations are higher than at middle RPM.
May be because residual cylinder pressure increasing (lower exhaust
air flow), the exhaust gases enter more into intake before being
pushed back into cylinder by the intake air.


> 
> On Wednesday 14 January 2004 11:15, Alexei Pavlov wrote:
> > Hi,
> >
> > I did not succeed to see you signal images. Even if you got better
> > signals afterwards I'd like to see the following:
> > The question is: do you want to see real pressure or do you
> > want to use average cycle pressure to manage the engine ?
> > Probably the second. In this case you need a correct sampling
> > of intake pressure oscillations. You must sample at a frequency
> > you've choosen and low-pass filter higher frequencies.
> > Then the CPU will compute the average pressure of 1 engine cycle.
> >
> > If you use higher frequency sampling you'll always get unstable
> > signal, as cylinders are all different and intake system has its
> > proper frequencies.
> >
> > From my experience with rather high overlap and very low overlap engines
> > I'd say you'll get more inter-cycle oscillations when WOT or at high
> > RPM.
> 
> --
> Tomas J. Sokorai Sch.
> Socio director, Gerente Proyectos Especiales
> Xperts Ltda. F: 63 244591, Valdivia, Chile
> _______________________________________________
> diy_efi mailing list
> diy_efi at diy-efi.org
> http://lists.diy-efi.org/mailman/listinfo/diy_efi
_______________________________________________
diy_efi mailing list
diy_efi at diy-efi.org
http://lists.diy-efi.org/mailman/listinfo/diy_efi



More information about the Diy_efi mailing list