Page 1 of 5

Ion Sensing ideas and circuit designs

Posted: Wed Oct 29, 2008 7:10 pm
by davebmw
OK just like the Knock sensing thread this is a sponge for all of your good ideas and prospective designs that could well be used on FreeEMS in the future. I have been working on both knock sensing and ion sensing for the last 2 months and have a couple of ideas to play with.

What is Knock sensing? Well in its simplest description it is a way of monitoring the pressure within the combustion chamber in realtime.

Why the hell do we need that?
if you are asking that then you are on the wrong forum! but i'll humour you anyway :p. Detonation is lethal for an internal combustion engine it blows head gaskets to shreds, it can burn holes straight through pistons, holes in the block bend conrods and wreak all sorts of havoc of your engine. it is much more serious in forced induction engines as the "charge" in the cylinder is much more volatile. in NA engines detonation is much less common but it does happen on a lesser scale in the form of pinking/pinging or knock this is still serious and most engines won't put up with much of it before blowing a head gasket, or worse!

So in answer to the question, it would be really handy when tuning your engine and trying to get the last couple of foot pounds out of it to be able to spot trouble before it happens and correct the tune accordingly.

The theory?
The gap between the spark plug is biased to a few hundred volts DC, the current flowing across the gap through the mixture in the combustion chamber is measured. as the pressure increases the current flow increases and is measured by a sensitive front end.
A sharp spike in the pressure after the initial ignition of the mixture indicates an abnormal burn and is likely a knock event occurring. the job of analyzing the waveform is probably too complex to do with discrete logic and op amps. it is very likely that any successful project would require a dedicated micro controller with the ability to take in to account RPM, load, atmospheric humidity and Manifold Air Temp in fact most of the factors required by the XDP in FreeEMS possibly including including timing information.

So come on guys lets get some ideas on e-paper and get some cutting edge feedback signals back to the FreeEMS of the future.

Re: Ion Sensing ideas and circuit designs

Posted: Wed Oct 29, 2008 7:21 pm
by davebmw
An interesting paper on ion sensing:

http://www.fs.isy.liu.se/~larer/Projects/main.html

Re: Ion Sensing ideas and circuit designs

Posted: Wed Oct 29, 2008 11:03 pm
by Fred
We could share all the core parameters out over the CAN bus to it to use if required.

Re: Ion Sensing ideas and circuit designs

Posted: Wed Oct 29, 2008 11:31 pm
by davebmw
Another cool project and some biasing ideas:

http://vvnet.fi/ville/ion/DIY-Ion-Sensing-2.pdf

Re: Ion Sensing ideas and circuit designs

Posted: Sat Nov 01, 2008 8:15 pm
by karlo
Reading out the ion current isn't rocket science, but it does require a voltage in the region of 300V (I've seen 180V as well). This has been a showstopper for me so far.

I've seen some designs which eliminate the 300V power supply by charging a capacitor to 300V with the 30kV (or so) spark and then using that to drive the ion current. A very elegant and simple solution, but I haven't had a chance to test it yet.

What I would like to do is analyze the ion current signal with a laptop and for example using that as realtime feedback to autotune in MegaTune. Maybe the algorithm could be made fast enough to port to a DSP or similar.

Re: Ion Sensing ideas and circuit designs

Posted: Sat Nov 01, 2008 8:57 pm
by Fred
karlo wrote:What I would like to do is analyze the ion current signal with a laptop and for example using that as realtime feedback to autotune in MegaTune. Maybe the algorithm could be made fast enough to port to a DSP or similar.
Yep, quite possibly, but not an hcs12!! :-)

Fred.

Re: Ion Sensing ideas and circuit designs

Posted: Sat Nov 01, 2008 10:58 pm
by davebmw
karlo wrote:Reading out the ion current isn't rocket science, but it does require a voltage in the region of 300V (I've seen 180V as well). This has been a showstopper for me so far.

I've seen some designs which eliminate the 300V power supply by charging a capacitor to 300V with the 30kV (or so) spark and then using that to drive the ion current. A very elegant and simple solution, but I haven't had a chance to test it yet.

What I would like to do is analyze the ion current signal with a laptop and for example using that as realtime feedback to autotune in MegaTune. Maybe the algorithm could be made fast enough to port to a DSP or similar.
OK for me this idea while novel has some serious drawbacks.
Firstly you are loading the ignition coil with an extra burden and it will effect the spark.
2nd the storage of enough charge to enable a current flow of up to 6mA for the duration of the power stroke also has potential problems e.g. how are you going to regulate the voltage of the DC bias while under varying current flow?
3rd the amount of available energy varies greatly across the range as the width of the pulse driving the coils changes.

I don't mean to be negative but for all intents and purposes we should try to minimise the number of variables in the equation so that when we get a signal back from the ionic current flow we can say with confidence this is right.
IMO we need to:
1, concentrate on designing a regulated 300V PSU to give us a solid reliable biasing voltage.
2, Make the system as passive to the primary function of the ignition system as possible.
3, Make a sensitive front end capable of measuring down to a couple of tens of microamps.
Find a DSP/DAQ capable of measuring accurately and quickly enough to produce useful data, maybe an audio ADC could be useful here. it could be a case of translating the current into a voltage and monitoring using laptop line in for now to get some idea of what we are looking at. :)

Re: Ion Sensing ideas and circuit designs

Posted: Sun Nov 02, 2008 2:14 am
by jharvey
Hmmm, could you do a min revers spark just before normal spark, and if it fires this revers spark, you know it's to early? Hardware for this could be as little as a small transformer, mosfet, and latch. If it the latch is fired, you know it either detonated, or is close to detonating. I don't think that would require much on the software side.

Perhaps that would be problematic with the dwell before the spark?

Re: Ion Sensing ideas and circuit designs

Posted: Sun Nov 02, 2008 12:40 pm
by davebmw
I think the point of ion sensing is to sense knock conditions before they happen rather than trying to blow the head gasket! ;)

Re: Ion Sensing ideas and circuit designs

Posted: Sun Nov 02, 2008 6:20 pm
by jharvey
My understanding of ion sensing is that it's done to prevent knock and detonation. I believe it's not typically used on a per cycle detection, but more as a general indication. So if you measure a little bit of current flowing in this cycle, you aren't knocking or detonating quite yet, so change the timing for the next cycle, such that you don't cause damage. I think the ionization happens mostly while compressing, but perhaps you can get readings while exhausting, ect.

I was thinking it might be easier to measure little reverse pulses, rather then having a circuit that will produce a semi continuous reverse current.