Sorry, I thought I had replied to this, but I don't see it:
Quote:
Originally Posted by Old Mechanic
which produces a high swirl action and better distribution of the fuel and air mix.
|
FWIW, the high swirl is more related to how it effects the flame front. There are always little vortex from filling the chamber, but they subside fairly quickly, so at low load/rpm, most engines have a fairly laminar flame front. When you are running very lean mixtures, this becomes a problem of time, since they reach the 50% burn point very slowly.
What a vortex does is it actually spirals the flame front, giving it greater surface area. They was some incredible images of this in a trade journal just a few years ago, but I can't find it at the moment.
We've known this helps at higher RPM's for a long time, since it flattens out that part of the P/V 'box', but that design brings the benefits down to low RPM/load situations so that they can be run much leaner.
Quote:
Originally Posted by Old Mechanic
Combined with a special Oxygen sensor (very expensive) it can run at mixtures as high as 22-24 to 1 at low revs and low throttle positions.
|
They call this their "Linear" O2 sensor. It is actually a fairly impressive feat of materials engineering, but inherently expensive. It's interesting to us, because of our 'Digital Direct' measurement method. We end up solving the same problem, but by accident.
Typically, a O2 measurement is made by pumping or withdrawing O2 to stoich. The current is then reported as a measurement, and is inherently non-linear. Klaus was frustrated because of the speed. It is akin to a PID motor controller. You are always really looking at a moving average and are only accurate under constant values that saturate the PID loop. In an engine, AFR is really always changing.
In looking at the Nernst equation, Klaus realized that the sensors should follow a certain behavior if, instead of driving to stoich, measurements are pulsed between two reference points. Right away, it was clear that he got the speed he was looking for (you can see valve action or even individual misfires), but it also turns out that the response doesn't have to be fitted to a curve, it is effectively linear. This matches the original math, but surprised him (and me, when I first tested it for myself) nonetheless.
The whole objective behind Emisense is to allow techniques like Honda's, or various lean burn's etc, to be executed with cheaper, more robust, sensors.
-jjf