Knock Sensor Solution: TPIC8101

[Matt Cramer]

The conditioned output of the pressure sensor could easily double as a knock sensor input.
A knock waveform will show the same amplitude/frequency characteristics.

A pressure sensor can be better than a knock sensor at sensing knock, with the right hardware. In cylinder pressure probes are actually what OEMs use to calibrate their knock sensors.

[subwoofer]

I still have no response from the plug washer pressure sensor vendor from the other thread, I'm not buying 10000, so no info for you, apparently.
Sadly I did not lie.

Which sadly only leaves us with two options: either find a way to stick a diesel glow plug into each cylinder, or wait for the OEMs to reach a standard for spark plugs with pressure sensors built in. I have heard rumours that the OEMs are actually working on incorporating cylinder pressure measurements into their engine management. Closing the loop on ignition timing will provide both better fuel economy and fuel tolerance at the same time, both very welcome as the petrol engines are downsized.

The conditioned output of the pressure sensor could easily double as a knock sensor input.
A knock waveform will show the same amplitude/frequency characteristics.

Actually, the knock signal is the high-pass filtered version of the pressure signal, drowned in noise... I have seen cylinder pressure and knock signal comparisons, and the knock has to be severe to be visible on the knock signal.

But I guess we are getting off topic here...

[piledriver]

You can downsize motors quite a bit when ~2HP/cubic inch is the new normal...

Looking forwards to buying some NGK irridiums with 0-5v MEMS pressure sensors integrated at my local FLAPS.

[subwoofer]

Indeed.

In Europe diesels are now outputting 90+ BHP/liter (1.4HP/CI), by analogy that should put 200 BHP/liter (3HP/CI) within reach for petrol engines if you compare NA outputs.

[piledriver]

Yes, but in Europe, a production car that gets 80MPG highway hasn't been a myth for well over a decade.

VW phased out that model at the end of its std 10 year model run, I'm sure there is a new Polo 1.4L Tdi that does as well or better.
The old one was about the size of my Mk1 Cabby. The engine and trans will probably bolt in.
OTOH I'm looking for a cheap FSI head and intake etc...

But way off topic...

[subwoofer]

(Moderate at will)

The latest Polo gives 59MPG(US) highway for the 1.2l 110BHP TSI engine (petrol), and 76 MPG for the 1.4 90HP TDI (diesel). I believe the highway figures, the mixed cycle numbers are VERY difficult to come close to in practice - we are seeing the same optimizing tricks that were in effect for crash tests for quite some time (a five-star US market car could get two stars in Europe and vice versa).

[elaw]

What, you're saying measuring MPG on a steep downhill with a tailwind isn't realistic?

[fixmann]

Hi
what about ionsense? http://delphi.com/pdf/techpapers/2013-01-0354.pdf



Fixmann

[racingmini_mtl]

Hi
what about ionsense? http://delphi.com/pdf/techpapers/2013-01-0354.pdf



Fixmann

There's already a ton of posts on this subject. Use the search function.

Jean

[subwoofer]

Actually, I don't think I have seen that paper before. At first looks it may seem that Delphi isn't too far from a commercial product.

[racingmini_mtl]

For some reason, I doubt we'll soon see that neural net based setup any time soon as a commercial product. But I might be wrong. In any case, that's not something that can be reproduced easily with off the shelf components.

Jean

[fixmann]

hi

yes i know there is alot of info about the subject on the web, but it was just a hint for those who havent heard about it.

Saab 9-5 1998 and up has a CDi cassette wich they use on their Trionic system.
Those engines run fully sequential, but dont have a cam sensor, at startup they run wastespark 1-4 and 2-3.
And after the engine starts to fire the ECU use combustion information from outputs on the CDI cassette to determ wich cylinder fires, then goes to sequ.
Just Google "saab trionic ionsense" and "T5suite"

some moore info:
http://www.ieeecss.org/CSM/library/1998 ... eTuner.pdf

Well just my input...
Fixmann

[subwoofer]

But I might be wrong. In any case, that's not something that can be reproduced easily with off the shelf components.

I read through it more thoroughly this time, I don't think the hardware itself would prove too expensive. A SHARC 32-bit floating point DSP is <20$, 8 channel ADCs in the MSPS range are <10$ for a 12 bit version. Not too DIY friendly if you have to use BGAs, but not undoable either.

The hard part would be the coils, apparently they made an ASIC for coil and ion current sensing at some stage, but they don't say if that ASIC is in production and used in any products or if it was made for lab use. My gut feeling is that it actually exists in some coils, possibly the HD ones, since you would not sink the cost of an ASIC for a lab project.

[racingmini_mtl]

I only very quickly glance at the document but what about the neural net part?

Jean

[subwoofer]

Neural nets are software, and AFAIK not black magic. Training them is the problem, but I don't think it's an insurmountable one.

[piledriver]

...but ion sense only works at high loads...

[subwoofer]

Which puts us back to pressure sensors...