Detonation Monitor

[Bernd Felsche]

Jorgen Karlsson tapped away at the keyboard with:
> This was funny, I just made a diagram showing the speed of sound at
> different temperatures today. I wanted it to get a better overview for
> header construction.

> A quick look at it shows that at 1000oC the speed of sound is
> 707m/s. More then twice the speed at room temperature.

At what pressure? Also, the gas temperature at knock is nowhere near
constant. 1000 degrees is somewhat high - the knock oscillation
occurs mostly in the unburnt end-gas (primarily because it's at a
lower pressure, ahead of the flame-front) that's probably a few
hundred degrees "cooler".

> But to an other subject, I don't think that the helmholtz
> frequency has anything to do with the detonation frequency. I
> don't even have a clue how to apply the helmholtz equations in
> this case.

Nor do I. Helmholtz defines attenuation (frequency response) of
flowing gases.

> I have always heard that it is the diameter of the bore that is
> the only variable. By doning a quick calculation on my engine that
> has a 79.5mm bore and assuming around 1000oC combustion temp I
> came up with 8949Hz resonance.  I think that only using 1X the
> diameter is the right way to do it, but I would probably try to
> monitor from half of that frequency and up.

It's the primary mode of oscillation - being the lowest-energy one
possible. Typical knock-detection is by listening to frequencies in
the range of 6kHz to 10kHz. Higher frequencies do exist but their
magnitude is several dB down on the main one. An infinite number of
frequencies is possible; the main ones are determined by the longer
dimensions of the combustion chamber. Near TDC, the bore is the
longest dimension.

Knock detection in the 1980's tended to employ simple analogue
signal filters for that bandwidth and an integration of the output
for a knock-amplitude. When one looks at that amplitude in the
"knock window" - the time in which knock is likely (usually within a
few degrees ATDC), it gives an indication of knock. The HIP9010
builds on that knowledge; it takes into account the "background"
noise level outside of the knock window for a more reliable
indication so that anti-knock measures aren't employed
unnecessarily.

There's been some recent research on using specialised washers under
head bolts to determine cylinder pressures during combustion; less
invasive than a pressure tapping, but subjected to other stresses as
the head flexes under load.

> After consulting with one of the guys at volvos engine lab I have
> decided to glue an electret microphone to a bolt that I then
> attach to the block. By listening to the output I should be able
> to hear detonation very early.

That's about all it takes. You can use a typical Bosch knock sensor
to the same effect. It's just a piezo element bonded to a metal
structure that's bolted to the block; typically at an anti-node
location of the block.

-- 
Bernd Felsche - Innovative Reckoning, Perth, Western Australia
----------------------------------------------------------------------------
To unsubscribe from diy_efi, send "unsubscribe diy_efi" (without the quotes)
in the body of a message (not the subject) to majordomo at lists.diy-efi.org