gemellocattivo.com

Has anyone played with this?

I have had the pleasure of using this type of system at a former job.

There really is no substitute, the data is invaluable when calibrating spark and torque

Most useful to me:
-If the data is good (high sample rate >1sample/degree), and there isn't too much thermal error in the sensor (many sensors have thermal error which will always affect different parts of the combustion cycle), you can calculate a very accurate burn rate model. From that, you can calculate the ignition delay (crank degrees from spark to 5% or 10% burn), burn duration (crank degrees from 10% to 90% burn), and 50% mass fraction burned (CA50 angle). A CA50 of between 7 and 9 degrees is MBT.
-If the data isn't quite so good, you can look at peak pressure and LPP (location of peak pressure), an LPP of about 15 degrees is MBT
-In any case, you can directly measure IMEP and PMEP (integral of indicated loop/integral of pumping loop) and look for IMEP to stop increasing to find MBT
-With lean burn (for fuel economy at part throttle), looking at IMEP covariance is the 'standard' measure of combustion stability. Higher numbers means more noise in IMEP over a multiple-cycle period, which means the engine is less stable. Most companies set some threshold for CoV IMEP and use it to define the limits of valve timing, EGR, etc...
-CAS knock is really accurate, by high-pass filtering a time-domain sampled cylinder pressure signal and integrating it. This is the same processing method used for acoustic knock sensors but none of the other engine noises are picked up by the pressure sensor.

Aside from CAS data itself, a combustion analysis system is also really useful because it can sample signals in crank-angle space continuously. It's kinda neat to look at MAP vs crank angle, and other signals like that.


The downside is it's hella expensive. Optrand sells relatively cheap spark plug sensors, they are 1-2k each with modified plugs. Most CAS processors don't list prices, the only one I've seen prices listed for is the NI CAS system (formerly Drivven DCAT) which starts at $9500 for 4 channels at 100ksps each.

_________________
"Sometimes, the elegant implementation is a function. Not a method. Not a class. Not a framework. Just a function." ~ John Carmack

"Any sufficiently advanced technology is indistinguishable from magic" ~Arthur C. Clarke

That's the issue isn't it.......I think Kistler gets something like $4000 for one plug and it require another $4k-$5k or stuff to connect it to the DAQ system.

These guys have a system that I know is cheaper but I don't know the price:
http://www.tfxengine.com/

Kistler's sensors are all either piezoresistive or piezoelectric. PR type are static and absolute, so they are normally used for intake/exhaust runners, and PE type are much faster with less thermal variation but are dynamic, they are usually used in cylinder. They also require charge amplifiers to output an analog voltage which you can sample as pressure. Normally exhaust sensors are water cooled, and in cylinder sensors are expected to be cooled by the cylinder head.

TFX buys sensors from Optrand (you can tell from some of the pictures).

Optrand sensors are fiber optic, and have a ~1m fiber line to a conditioner box which has relatively normal electrical interfaces (power, ground, and differential voltage output). It can be connected directly to an ADC and read without a charge amplifier.

With any dynamic pressure sensor you will need another sensor to peg it to, since it doesn't have a zero offset you need an absolute sensor which is read during a certain part of the cycle to establish the zero offset for the next cycle. Usually intake or exhaust pressure is used for this.

Some CAS systems now use the existing 60-2 crank wheel so they are easier to use in a vehicle, but a lot of the lab-grade ones (which is most of them lol) need a high resolution (360 or 720 teeth/rev or more) crank encoder.

To get 1deg/sample up to 8k rpm you would need to sample at 48 khz for 1 cylinder. Most ADCs sample sequentially so the ADC would need to sample at ~600khz for a V12. Realistically you need about 100 khz per cylinder at the ADC. For a V12 this is still within the limit for a MPC5674F/MPC5676R by the way (which can do about 1mhz per ADC and has 4 ADCs).

The next issue with any daq system that can capture at 100khz+ per channel, is that's a ton of data to store.

_________________
"Sometimes, the elegant implementation is a function. Not a method. Not a class. Not a framework. Just a function." ~ John Carmack

"Any sufficiently advanced technology is indistinguishable from magic" ~Arthur C. Clarke