MR2 Owners Club Forum banner

1 - 10 of 10 Posts

·
Registered
Joined
·
91 Posts
Discussion Starter #1
damon,

you know whats funny is I am involved on this interesting project at work where I am measuring the resistance of a plasma field to determine ion density (and reaction endpoint among other things).

I am learning a great deal on plasma physics and have been thinking of your ion sense project. Several things concern me regarding that project with the foremost being the power source to strike and sustain the plasma between the spark plug electrodes. You will most likely need a radio frequency generator dialed in to very low voltages (or else you will have a total breakdown and arc the sparkplug gap).

But then I was thinking further that another way of determining the pressure is to have a series of sparks beginning at bottom dead center and progressing through to TDC. Here you would have a pulse driven signal that starts at an arbitruary low bound voltage and quickly ramps the voltage up until the field between the spark plug gap deteriorates the gas dielectric (arcs). You would datalog this signal and it would increase as the cylinder compressed the air/fuel mixture.

The hard part for me would be coinciding the metrology with the signal generator. You would most likely need something to accept measurements in the thousand or tens of thousand hertz range. Maybe multiple dataQ boxes out of phase would do the trick......

Anyway man I am full of ideas here now. This would be an easy project compared to all the crap that goes on at work.
 

·
Registered
Joined
·
4,344 Posts
Wow! First off, where did you hear of this?

50 kHz ADC boxes are very commonplace nowadays and are quite cheap. Heck, your sound card is probably running at 44 kHz. The trick is circuit protection.

The series of sparks interests me, except that you would need a considerable power source to achieve this, right? That would be a tall order equipment-wise on our little 12 volt systems. Part of the whole idea of this project is to keep the equipment costs down.

Lastly, you don't need to sustain a plasma between the electrodes. The spark is of course the initial plama, and my circuit is protected from that by avalanche diodes. Then, 400 VDC is applied to the spark plug just after the ignition event (triggered off inductance from the wire). The resulting current from that non-arching voltage is directly related to the ion density, since your plug electrodes become electrically charged, creating an electric field. From this amperage data, cylinder pressure can be inferred using gausian regression curves (3, to be exact). Once you have the regression curves fit, you can very effectively filter the data for peak pressure point and muala!

Tell me more about your project. This is of interest to me!
 

·
Registered
Joined
·
91 Posts
Discussion Starter #3
my concern is the out-of-control variation you will experience due to spark plug deterioation, minute gapping differences, and most importantly the di-electric carbon coating your electrodes. From what I have learned in plasma kinetics is that you need radio-frequency oscillating power to sustain a plasma if there is any form of dielectric between the electrode and the gas. The question is if the 400v DC current will be applied in such a short time that your important information can be dicefered before the gas loses its conductivity.

The reason I would tend toward the field breakdown point as I mentioned above is you can normalize everything to a generic system correction factor. So you know how much effect you get from the individual spark plug, cylinder differences, or anything that you otherwise have no way of initializing.

Maybe system variation is so small it is relatively unimportant? The only way to tell is to proceed and if/when your gausian modelling fails you will know that the inherent variation is not so small in relation to the tested variable.

To generate the series of sparks you would only need a simple CDI ignition system. I used to have a link to a place that shows how to build one yourself and it was quite interesting. The only difference is you would need several (maybe 4) capacitors operating together. Or maybe replicate the circuitry used for aftermarket ignition systems how they have multiple spark capabilities.

The DIY CDI system went into a little detail on multiple spark but the research group veered away from this as they achieved better performance elsewhere for igniting a fuel mixture. But maybe it could be pursued in determining a breakdown threshold voltage.

I am a fanatic for data and statistic modelling. I feel much better when I have a curve I can model of a cylinder's pressure as opposed to a single ion source resistance point. But it sounds like you are well ahead of me in this arena so I would go with whay you have already.

I do know that with a "threshold voltage" method of correlating cylinder pressure the equipment would not be any more expensive then buying an aftermarket ignition system. Only how much more valuable would cylinder pressure be to you then anything else!

You could even do away with A/F monitoring, because the physics doesnt care how much fuel is there. All your engine cares about it extracting work from a high pressure gas wanting to expand. You could revolutionize engine tuning with a well functioning pressure monitor system. Seriously, if you knew crank angle (which you do) and cylinder pressure, factor in rod angle to calculate a lever arm variable (it varies as the rod angle chanegs), then plot a lever arm vs. force (force = pressure x surface area), integrate the curve, and maximixe that integral value by changing your engine's process knobs (fuel, timing, spark advance).

what really bothers me about all of this is how no one else has done it really. Or if they have it is kept confidential. Well, im sure everyone who expands to new frontiers asks this same question :D.
 

·
Registered
Joined
·
4,344 Posts
Wow, what a book. Okay:

Spark plug deterioration is a concern, but not one that can't be overcome by a new set of $2 copper plugs. Gapping differences don't play much of a role here, as I can compensate by changing the DC voltage. Dielectric coatings on the electrodes can also be overcome by a new set of plugs. For tuning purposes, this is quite adequate. As a long-term stand-alone controller, this would have to be addressed. That isn't part of the initial iteration of the product.

The gausian curve fitting has already worked at every operating point except idle vacuum, where the signal isn't strong enough to decipher a derivative that is worthwhile. Because I don't care much about what happens at idle, this is a non-issue to me. Light cruise conditions work fine.

Your system would involve sparks happening after the ignition, and then measuring the resonance and determining peak pressure point? That is interesting! I would be interested in learning more about this.

My system doesn't require a COP CDI ignition, although it is helpful. My system is activated by the ignition event and the combustion provides plenty of ions present to get a nice strong signal (except for at idle). Plus, detonation is VERY noticable with this type of system! Imagine, even trace knock being able to be noticed LONG before you can hear it!

By the way, I'm not looking for a threshold data point. I take the whole set of data from ignition to 30 degrees ATDC and fit 3 gaussian curves to it. I then subtract the first 2 and the remaining curve is the cylinder pressure. I then compare the peak of this plot to the engine triggers to get what I'm after.

Yes, different rod ratios will yield a different ideal peak pressure points. This means I model our system once and obtain a best crank angle to shoot for. I would then verify this on the dyno. I have read of measurements peaking different then they should on paper. I can only attribute this to viscous forces in the journals and cylinder sleeves, which can be significant, as you know.

There is one thing that I don't have licked yet. Cycle-to-cycle variations (differences in combustion qualities under identical circumstances) is one thing that I need to correct for. I am thinking I would take an average of 5 events, throwing out anything outside 2-sigma, and only THEN apply a correction to the igntion based on this average.
 

·
Registered
Joined
·
882 Posts
1 - 10 of 10 Posts
Top