Inaccurate DFI digital tach readout

[Tony Cooper]

Robert Wilkinson wrote:

> When I am driving with the car in gear (standard shift), and the car
> is in a steady state condition, the RPM displayed on my DFI will wander
> all over the place.  The RPM will be close, but it wanders off as much as
> + or - 75 rpm.  This occurs at all rpm's.  Is this normal with the DFI or
> other computers which can digitaly read the rpm signal??  I have 3
> different tachometers hooked up to the car, and none of them wander.
> FWIW, I noticed the other day on TV that a Nascar using telemetry was
> reporting rpm at about + or - 2 rpm.  Also, I recently replaced my pickup
> coil and 7 terminal module (gm hei), and the readings were not affected.
> And I still have a miss that occurs during light cruising conditions which
> I cannot get rid of by adjusting timing, and it may or may not be related
> to the DFI problem.
>
> Robert Wilkinson
> http://ccwf.cc.utexas.edu/~rw

Can I assume that you have built the DFI system?
If so then it could be 1 of 2 things.

1. What you are seeing is an inherant problem with digital RPM calculation. It
sounds like you are calulating the RPM every pulse. ie, if you have a 30 tooth
wheel for timing, you are calculating the RPM 30 times a rotation. If this is
correct, then you are seeing the accelleration/deceleration curve of the engine
as it rotates. What you need to do is use somthing like a running average, and
average the RPM value over 'some' pulses. This will smooth out your +-75 RPM
jumps. Only thing is, the averaging must be adaptive - if the engine is
currently accellerating/decellerating, then the averageing must reduce its count
- how can I explain this with ASCII. - Ok, try this
If the engine is steady state, then use the average of say 10 pulses, if the
engine is not steady state, use 3 - I am sure you get the idea (Sorry bouyt
this, but its early in the morning for me - more caffeine).

2. Your measurement clock has insufficient resolution. I assume that you are
measuring the time between engine pulses. If the clock is not high enough
resolution, then you will see large jumps in RPM measurment while the engine is
steady. Unfortunatly, this problem can cause misfires as well, cause there is a
good chance that you are using the same clock for ignition timing/fuel injection
as well. However, with some jiggery pokeery, this can be 'got' round. If you
don't have enough resolution, try measuring the time over multiple pulses, and
dividing result by number of pulses. I have some math somewhere that will tell
you the limits of youor current system (Can't put me hand on it directly at the
mo).

Of course, this is all rubbish if you didn't build the DFI, and if this is the
case - send this mail to the bin bucket.

Tony - serevrly caffeine deprived.

More coffee vicar????
No thank you, the last cup made me fart......

--
Sent By Tony Cooper.
email: tony.cooper at virgin.net
Allow at least 10 working minutes for reply. ;)