Current/New ION information

[Gar Willis]

On Wed, 19 May 1999 08:15:17 +1000, Peter Gargano
<peter at ntserver.techedge.com.au> wrote:

>   http://www.fs.isy.liu.se/~larer/Projects/main.html

>> "measuring the ION current, and in particular the "shape" of the ION current,
>>   based on pattern recognition of the ionization current"

Hi Peter the ION-hearted. B)

Just a few possibly sage observations (I'll leave you to decide :) from
a ole fossil who's been gazing in rapt wonder at this area for awhile:

You have to constrast and put the presented level of sophistication of
these latest Swede papers in perspective, tho, methinks. If you consider
where their PRACTICAL implementation came from (the original Saab patent
tells the story; yes, the one with the schematic). And it began as a
means of detecting misfire and providing cam decommutation, to eliminate
the need for a cam sensor. If you look at the initial deployment from
'90-'92, it did NOT provide for detonation detection. Next step: if you
look at the follow-on system that replaced it in their Turbo cars, the
so-called Trionic system that was intro'd in '93, and has been their
mainstay ever since (looking carefully at their documentation is all
that's really necessary), you'll find that while the pre-ignition,
misfire, and cam decommutation ARE done within the Trionic module, the
knock detection is NOT.

Hmmm, plot thickens, eh? They pick off the raw ionization signal, handle
the wee bairn with extreme care, and present it to a fast A/D port on
the ECU. That's sorta mutha naitcha's way of saying it's not so easy to
even detect detonation without some 'computational horsepower'. Doable,
and doable well, judging from the success and lack of further major
revisions in the Trionic system, but NOT so easy as to make it something
to do in straight mixed-signal within the Trionic module.

OK, so where am I going with this? Well, there's a pretty big leap from
just being able to DETECT detonation (let's put it this way, to perform
pattern recognition sufficient to discriminate between normal and
detonating combustion), to being able to determine an order of magnitude
more recognition and separate out many MORE and MUCH more subtle
variations in the ionization "pattern", so that you can MEASURE things
like cylinder pressure, and charge AFR, for example.

Now, it's in the very nature of the beast, that you see in these papers
quite alot of forward thinking and perhaps speculative thinking/results.
But let me just point out one interesting thing to note about the
Saab/Swede's publications and even their patents (or the lack of them)
concerning their currently deployed system. AFAIK they have never
actually shown/said how their *current* robust/practical/deployed system
actually determines detonation. All we know is that the ECU ingests the
ionization signal; that's it. We don't REALLY know if they do "pattern
recognition" or even "spectral analysis" or even simply some sort of
mundane filtering. What we DO know is that unlike the conditions in the
research papers, their ECUs seem to do this discrimination bloody
accurately AND bloody fast.

That's why I said I thot you need to keep the level of sophistication
being described in these papers, especially these multivariate pattern
recognition schemes, in perspective. And also consider how much compute
HP was available to the Saab ECUs in '93. Just some practical matters to
consider.

>I'd recommend all interested ION project people look at this article as it
>also summarises much of the research that has been carried out to date.
>It also provides a common information database from which we can all work from.

There's actually even MORE stuff going on than the picture he paints.
The Swedes are going in some interesting directions, but so are the
Japanese, and you get a glimpse of that from the patent sifting AND some
of the JP papers he DOESN'T site. B)  I'll give ya two hints; if you
look at the patents on the subject, you'll see LOTS more Japanese work
in this area, and in diff. directions, than the Sweded, and much more
work than is represented in his paltry references to JP papers. Second
hint: scan the titles and authors of the SAE papers on this subject by
doing your own search of the SAE documents, NOT via his bibliography

In contrast to the Swede papers, OTOH, what you'll find in the JP papers
is some GRADUALLY increasing sophistication to detect detonation, for
example. And those DO NOT depend on the giant leap to "sophisticated
pattern recognition" that the Swede's seem to imply.

So finally, what's my point? Well, those papers and the bibliography you
suggest above we use as a "common info database", might be fine if we
wanted to step into the moving stream of Combustion Ionization research,
but for us hobbyist-experimenters, methinks (at least this is true for
me; I don't speak for everybody, obviously) that our interest is largely
sorta tryna catch up with some of the cool albeit heretofore hidden
things in engine technology, that some of the more advanced mfg's have
been able to use and benefit from for some time.

That's MY interest in ION, and it parallels pretty nicely the history of
EGOR as well. Not too many of us are interested in keeping right up with
the latest in O2 sensor physical chemistry so as to be able to be a real
"practitioner in the art", but I DO know that many of us are interested
in CATCHING up at least enough to be able to USE/play-with this better
O2 sensing technology, which after all HAS been in use for yearsNyears,
but hidden by a cloak of concealment for that long.

Sure, we wanna know what's goin on with the latestNgreatest, but
nobody's approaching EGOR by tryna catch up with the latest research;
we're just tryna figure out how all these other bloaks have been doing
for years!!! B)

So the same obtains with ION, methinks. Sure, it IS neat and useful to
read and glimpse the latest in ionization measurement research, but I
warrant most of us are more interested in just GETTING to USE something
that works as well as the stuff Saab has had deployed since '93!!

And that brings me to my last BIG hint: it's in the JP papers, IMNSHO,
that you will find the incremental advances and techniques that will
help you to better understand how you could IMPLEMENT a robust
detonation detection method. NO, you probly couldn't use the same
approach to determine all these advanced parametric extractions that the
Swede papers are discussing. But hey, there ya go! Take yer pick. B)

Now, obviously since I've chosen to be a partial "concealment artist"
meself, by being a vendor of electronics in this area (which I do outta
necessity, not by preference), I can't tell ya any MORE at this point,
cuz then I'd hafta shoot ya. Heh. Suffice it to say that IF you're
interested in the more practical/mundane issues of how to make ION work
for detonation detection, I recommend a more careful, balanced look at
ALL the papers, and in particular that means the Japanese work. End of
diatribe.

>So, we're talking about engine optimisation, not just knock/pre-ignition
>detection!

Ah yep, pretty exciting, eh? The moment I read the first papers on this
stuff, I too fell in LUV. That's why I thot it was completely
appropriate to anthropomorphize this tecnology as a "beautiful lady
named ION". Especially for an electrical type, the very idea of using
the spark plug as a major sensory window into the combustion chamber,
AND be able to tell so MUCH from interpreting the ionization picture is
appealing enough to have an aura of beautific sublimnity. B)

For the present, I'm sure many of us would be content to hear ION reveal
her secrets about the mystery of detonation, but it IS neat to think
that this one magical song may be just the beginning of her wispering
much more poetry in our wondering ears.

And to think, all this was lurking on the other side of the bloody
spark-plugs!! Sheesh. B)

Gar