PPP measurement (was Knock module for MS3)

piledriver · Post by **piledriver** » Mon May 13, 2013 1:33 am

Interesting looking bit... i'm still curious as to what it is. I seriously doubt it's a TC.

The thing is, as neat as it looks, OEMS have NOT switched en masse to ion sense, in fact, the MEMS pressure sensors such as those used in all the late model VW TDis (in the glow plugs) and piezo sensors integrated into custom spark plug wells (spotted several GM patents to that effect) seem to have serious advantages vs ion sense, which simply doesn't work ... at too much less than WOT, and is likely best suited to only knock detection---and still needs a metric assload of support electronics.

The direct (or indirect) pressure sensing bits don't have to cost more than ion sense, in fact, it would likely cost far less in quantity, and be far more reliable as is does it's deed at low voltages, and with reasonable (computer I/O) freindly signal voltages and ready to use signals. They also work from cranking speed up.

There's some dodgy looking Thai manufacturer flogging spark plug gasket-style combustion pressure sensors to OEMS, almost considered filling out the spam invite trying to score a sample... http://www.hellotrade.com/yamazen-thail ... ensor.html

I figured they were making them for Kistler or such but no joy.

subwoofer · Post by **subwoofer** » Mon May 13, 2013 4:27 am

The type of sensor works, apparently they were very much in use on MotoGP engines at the end of the 2-stroke era.

Keep us posted on your progress here, if they are cheap enough short life wouldn't matter much.

nuvolarossa · Post by **nuvolarossa** » Mon May 13, 2013 4:51 am

you're right, there is a lot of info about those on 2 strokes:

http://klemmvintage.com/deto.htm

a german manufacturer do this:
http://www.2d-datarecording.com/Downloa ... -DINA4.pdf

a lot more too, Honda HRC did one when was racing on 2 strokes, still doing it
scroll down in this page:
http://www.sp125racing.com/hrc-datalogg ... nters.html

some specs for that sensor at the end of this page:
http://www.nsr250.net/forums/viewpost.p ... 9cd25a329c

you may find more info searching for "det counter"

piledriver · Post by **piledriver** » Wed May 15, 2013 11:53 pm

Well, sacrificed my yahoo spam account to the spam invite on those under-the-plug sensors, we'll see if they respond.
Asked for datasheet and samples and/ or existing distributor info.

subwoofer · Post by **subwoofer** » Thu May 16, 2013 12:23 am

Good 'un! We'll just wait and see what happens.

I have been reading up a bit on under-the-plug sensors, seems knock detection will be as simple as a 4-20kHz BPF with an amplitude detector, so a first step could possibly just hooking one up to a regular acoustic knock detector board. But the really interesting bit would be to use it for tuning to determine PPP...

piledriver · Post by **piledriver** » Thu May 16, 2013 6:27 pm

I see no reason it couldn't do both with the proper code looking at the output.

The existing knock module is probably ideal with the preexisting center freq and time windowing ability. although you'd probably want ~full BW signal for the pressure trace.

Not sure if the MS3 CPU has the power to do the PPP analysis, perhaps the XSCALE can be of some assistance?
That would be holy grail-ish if it worked.

I'm still looking for a retail source of the sensors, the German site in a previous link didn't seem to have any info at all on that currently, nor could I find the second page of the PDF with the actual sensor specs.
(perhaps I need to hit waybackmachine)

nuvolarossa · Post by **nuvolarossa** » Thu May 16, 2013 11:56 pm

here you can find it, at the end of the page. at the cheap price of 277 gbp = 422 usd (only the sensor

)
http://www.sp125racing.com/hrc-datalogg ... nters.html

and this is the honda hrc sensor only, 450usd
http://www.roteg.nl/detonationsensorfor ... 45_803_815

what I read about the honda hrc is that usually in the full kit is coupled with a module that output digital signal, not analog. Maybe because on 2 strokes they use it as a det counter per km only.
But in the shop above there are cables to convert from the honda module to 0-5v dataloggers, but don't know how "detailed"/filtered the signal would be.

piledriver · Post by **piledriver** » Fri May 17, 2013 1:44 am

I'm betting that sensor can be had for a tiny fraction of that price, purchased direct.

Still poking around for datasheets to see if the signal will need a preamp or if the knock module can deal with it directly.

I suspect James' knock module is very similar to what they are using in their $$$$ custom boxes, if not more advanced.

piledriver · Post by **piledriver** » Fri May 17, 2013 2:46 am

From a Delphi witepaper on knock sensing at http://am.delphi.com/pdf/techpapers/2000-01-0932.pdf

"Nissan produced a system on a 3.0L V6 that used ring load washers under each plug to determine individual cylinder combustion knock"

Somewhere, there is a part#...

I'm starting to think it's part of the head, possibly not usually replaceable.

...

piledriver · Post by **piledriver** » Sun May 19, 2013 7:21 pm

The AVL//PCB version is pn# 140A14, here's the web page, no voltage output info, other than output is negative as pressure is increased, so will go positive with pressure in the typical application under a spark plug. (as combustion pressure increases, pushes plug out some, fighting plug tq)
http://www.pcb.com/Products.aspx?m=140A14#.UZmG59ciZdg

Nevermind, it does have an output scale defined in colombs/newtons, is directly convertible to microwatts, given internal resistance, to volts.
(assuming highZ input on amp circuit, easily calculated in any case)
10 pC/lb, 1000 lb range, 3K max, 400F max, 2% full scale accuracy.

Price="call" always gives me that uneasy feeling I'm about to find out I don't get the special discount.

subwoofer · Post by **subwoofer** » Mon May 20, 2013 12:37 pm

If I read the white paper correctly, absolute scaling is of little - if any - interest, so all we would need is to safely transfer the signal from the charge domain into the digital domain. Then the rest is relatively straight forward number crunching.

The "call for pricing" could mean one of a number of things, it could mean a large MOQ (bad!), it could mean prices drop like a rock with order size (indifferent), it could mean NDAs or other restrictions apply (solvable). No reason for panic - yet!

I love these white papers. They drown completely in the noise on the net, but they hold a completely different attitude to engine control than the (conservative to the point of being reactionary) tuning society. Not all the ideas are practical, though, but the one from Delphi looks doable - all we need now is to get hold of the dang sensors and some amplifiers.

piledriver · Post by **piledriver** » Mon May 20, 2013 7:50 pm

I'm thinking they might be of immediate use as "better" knock sensors with the existing knock module, possibly need a simple buffer amp to get the signal in the right range.

PPP would need high BW A/D and very probably a second processor, but AFAICT the actual algorithms are as old as dirt and public domain.

OTOH you might be able to pull it off with audio card inputs, even as post processing would be a huge win.

Still haven't heard a peep throm Thailand, not even in the form of extra SPAM.

subwoofer · Post by **subwoofer** » Tue May 21, 2013 12:50 am

piledriver wrote:I'm thinking they might be of immediate use as "better" knock sensors with the existing knock module, possibly need a simple buffer amp to get the signal in the right range.

We need something like this to buffer the sensor, charge mode sensors need special electronics for buffering. Still not a major problem though. Ideally, that buffer should be really close to the sensor, I read that triboelectric effects (charge generated by the wire strands rubbing against each other) are measurable for this kind of sensor.

http://pdfserv.maximintegrated.com/en/an/AN1127.pdf

piledriver wrote:PPP would need high BW A/D and very probably a second processor, but AFAICT the actual algorithms are as old as dirt and public domain.

Audio sample rates (48kSps) would be plenty for all the algorithms implied, but as you say a second processor will be necessary. We would have to design a method for feeding the calculated PPP back to the MS3 too, possibly as a correction signal centered around 2.5V, or better yet in some form of serial digital format (CAN for instance). I see short/long term trim tables for spark in the crystal ball.

piledriver · Post by **piledriver** » Thu Feb 13, 2014 1:05 am

One of my cats gave me a nice set of 0-100 lb range piezo pressure sensors today...
my now dead bathroom scale had 4 of these little spring bridge based sensors.

IMG_20140209_233234.jpg

I'm thinking I can build some simple trusses between my upper head studs (in free air) and put these in thr middle where I know the head flexes some above each chabmer, next to the plug.

Since we are only doing relative measurements I can see it possibly working.

piledriver · Post by **piledriver** » Fri Jun 27, 2014 1:03 am

Now that I have a nice 1000 lb sensor in hand (that does seem to work well OK driven manually, waiting for an INA125) my one remaining neuron fired and noted another possible way to skin the cat...

Force sensors are delicate (force range, heat) expensive, and may be too d*** slow to really be of much use.

What we need are perhaps motion sensors... Very small range, ~cannot be overstressed, could be optical fiber based and virtually impervious to temperature/vibration etc..

Logic: For PPP curve we really don't need calibrated jack, we just need self consistent results.
The head moves some for a given cylinder pressure, that's all a pressure sensor really sees.

subwoofer · Post by **subwoofer** » Fri Jun 27, 2014 1:59 am

Intriguing, but possibly flawed. I am pretty certain head motion due to cylinder pressure << head motion due to general vibration. But an old fashioned strain gauge (stick-on) may do the trick if placed correctly?

subwoofer · Post by **subwoofer** » Fri Aug 08, 2014 1:44 am

I wonder what the dimensions of these glow plugs are? Heads could possibly be modified to accept them, the glowing bit could even be cut/ground off without affecting the pressure measurement.
https://www.youtube.com/watch?v=HS-SMZ_d3Lo

Possibly interesting SAE paper: http://papers.sae.org/940381/

piledriver · Post by **piledriver** » Tue Aug 19, 2014 9:43 pm

Sawn off pressure sensing glow plugs look like the easy button for those who can drill random holes in their heads.
(for once it pays to be air cooled)
I have a great spot on my heads to install it, two, actually.

It may have to fight a GDi injector for one hole, but another likely OK spot exists in the intake bowl side of the chamber.
Bathtub chambers and aircooled heads, gotta love them.

Still trying to figure out if I'll have to set up a tiny water cooling setup to keep the injectors alive.
Do the glow plugs have head temp issues ?
(I actually probably know some of the guys that make the MEMS sensors...)

piledriver · Post by **piledriver** » Thu Sep 11, 2014 1:24 am

Looks like Daimler is now using a 0-5v pressure-only sensor and seperate ceramic glow plugs, anyone have a part# and vehicle app?

http://pressroom.sensata.com/phoenix.zh ... highlight=
http://www.sensata.com/sensors/automoti ... r-cpos.htm
http://www.sensata.com/download/cpos.pdf

Perhaps some of these would also be suitable?
http://www.sensata.com/sensors/automoti ... or-msg.htm

These would need extensions for my aircooled app, but that's a feature as it moves the "delicate" end away from the potentailly 350+F CHT well into the cooling airstream.

subwoofer · Post by **subwoofer** » Thu Sep 11, 2014 1:38 am

Wow! Now we are getting somewhere!

The CPOS sensors will fit the bill rather nicely, all that remains is to figure out a way to get hold of them and a way to install them in a head not originally equipped with them. But there are good machinists out there, I know of at least one who would love to figure out a way to fit them.

Megasquirt Support Forum (MSEXTRA)