Plasma Ignition

Geek: your pm-box is full!

Excellent paper or book. Everything what is needed here is there:
Rudolf R. Maly and Rudiger Herweg, "Spark Ignition and Combustion in Four-Stroke
Gasoline Engines", C. Arcoumanis, T. Kamimoto (eds.), Flow and Combustion in Reciprocating Engines, DOI: 10.1007/978-3-540-68901-0 1, Springer-Verlag Berlin Heidelberg 2008
http://www.amazon.com/Flow-Combustion-Automotive-Engines-Arcoumanis/dp/3540641424/ref=sr_1_3?ie=UTF8&s=books&qid=1256654023&sr=1-3#reader_3540641424
http://www.springerlink.com/content/l556j16086u10757/?p=3b58edb94720481cb4338086621cd6bb&pi=0

There are 3 phases in spark ignition: the breakdown phase (few ns, several hundred of A), the arc-phase (few us, several A) and the glow-phase (few ms, hundred of mA).

"We notice a very short (ns) first phase - the breakdown phase - during which the
spark current rises to a first current maximum of several hundred amperes. Its peak
value is determined by the ignition voltage U0 and the impedance of the near-gap
circuit:

    IB = U0/Z p ≈10 kV/50 ohm= 200A

Within a few ns the gap voltage drops to very low values of around 100V. This
phase is uniquely determined by the capacitive (Cp = 5 − 15 pF) and inductive
components (L p + LG ≈ 5 nH) of spark plug and spark, respectively. This phase
is followed by a second one - the arc phase - lasting only for about 1 μs. During
this time the capacity of the high voltage cable (CL = 40 − 100 pF) and the coil
capacity (Cc ≈ 50 pF) discharge through the radio interference damping resistor
(Rr = 1 − 10 kohm) in series with the cable impedance. Hence typical values for the
second current peak are:

IB = U0/(Rr + ZL ) ≈10 kV/2kohm = 5A

Finally, a third phase follows - the glow phase - delivering the lion part of the
originally stored electrical energy into the spark discharge. In a TCI this phase lasts
for several (typically 2) milli-seconds. The peak glow current is mainly controlled
by the coil impedance (Zc ≈200 kohm) and decays approximately linearly with time
(the logarithmic scales are deceiving on a first glance):

IG = U0/(ZC + ZL + Rr ) ≈U0/ZC ≈10 kV/200 kohm = 5mA

If the glow current exceeds about 100mA the discharge may transit into an arc.
This occurs for low impedance coils or in Capacitive Discharge Ignition systems -
CDI - , where the electrical energy is stored in a primary capacitor and the coil is
replaced by a pulse transformer. In these cases rapid voltage oscillations may occur
due to rapid transitions between arc and glow phases."

"Energy balance for Breakdown, Arc and Glow discharge plasmas under idealized
conditions using thin electrodes:
                               Breakdown     Arc          Glow
Radiation losses                 <1%      ≈5%      <1%
Heat conduction                 ≈5%      ≈45%     ≈70%
Total energy losses            ≈6%     ≈50%      ≈70%
Total spark plasmaenergy ≈94%     ≈50%      ≈30%"

What nology-wire tries to do is to increase breakdown-phase time and energy. What this plasma-ignition (or better arc-ignition) does is to increase arc-phase time and energy. Long duration glow phase is effective to ignite lean mixtures. Increased arc-phase energy is corrosive to plugs (6000 K temp arc melts material away from electrodes). At glow-phase temp much lower and that increases heat conduction (losses) to electrodes.  


Edited by shmot - 28 October 2009 at 6:37am

Originally posted by geek Just looking at the data, let's start by clearing up some basic digital scope use.  Sorry if you already understand this.  On the screen, we can see your running the scope set to 1V/div.  You have a divider that takes this to 77.4V / division.     Let's say with this setting, the scope can read up to 10 Volts.  Let's say the scopes ADC is 8-bits.  This would allow you to resolve to about 40mV.   With your attenuator, this brings us to a resolution of 3 Volts.   Now, if you look at the data you can see the ADC steps.   Also notice that as the ADC moves from one count to the next there there is all sorts of noise.   Even when things are steady state, the data is not clean.   The ADC will have so many bits of quantization noise.  What you would be looking for is some sort of ENOB (effective number of bits) for the part (which is not really right as well for what your looking at but gives you some idea.).  There will be other sources of noise too.  Let say we have a total of 3 counts of noise.  This is a total of 9 Volts in front of the attenuator.     One way to reduce this noise (assuming the signal is stable) is to collect more data. My point is that while you can draw some general conclusions, don't expect the exact numbers you read with the cursors to be right.

Yes. Those numbers are what scope did measure. I did have 1 kohm, 10 kohm, 2x33 kohm resistors in series and I did measure voltage over 1 kohm resistor. That why it is 1 V at scope is 77.4 V at real. Max scope input voltage is 20 V (peak-to-peak I guess). In that other scope (when it arrives) there are operation amplifiers in AD-inputs. It uses program controlled DA-transformer to scale OA input range. So, it is always possible to scale it to use max number of bits. Its input range is -100 mV to 1000 mV and I will have to use resistors build extra box for rough scaling.   


Edited by shmot - 28 October 2009 at 7:08am

Originally posted by geek I tend to believe your post on the 6 vs 7 dyno tests.   Not that I have EVER seen any ignition dyno tests, I just find it had to believe there are going to be any gains ( lower E/Ts) for a non-lean burn gasoline or methanol engine.   I have yet to see one study using an engine dyno to prove otherwise.  I don't believe any papers or books have been written without errors.  Like the internet, people build off of other peoples work and people make mistakes.   Nature of the beast I guess. Say your scope is 8-bits and we wanted to look at those few hundred mA or even Amps of current after the breakdown.   Let's say the breakdown current (    IB = U0/Z p ≈10 kV/50 ohm= 200A ) is close.  Note, it's a lot higher than I measure with my little grill starter.   Lets add 10% of head room to make sure we are not near saturation.  That gets you to .86A / count.   Staying with the 3 counts of noise, that's 2.5 Amps of noise.  Most of the DSOs I have used have at least  20mV of noise.   The new higher end Agilent scopes I have used are much better.    It's been a few years since I have looked at their products, so I am not sure what they have for real high speed stuff with deep buffers.      If you try and zoom in to the area your looking at by changing the scopes front end attenuators, your not going to see what you think and only fooling yourself (and may take out your frontend).   Best thing yo do is setup a test to prove what your system is capable of, before making any measurements that your going to hang your hat on.   I would like to run the test just to see but there is no way with the equipment I have that I would get anything meaningful out of it.    I think what the Nology wires are going to buy you is breakdown.   Your still playing with pretty low pressures.   This weekend I'll bring home some nitrogen to test with.   At least you will have an idea of what the two wire sets and coils I have fail at.   I'll flush my box today.

Are you going to use nitrogen in your test?

I ususally want to first understand things theoretically and if it not busted at theoretically then do measurements. At some point we have to do some dynotesting (with that methanol-cbr or other). We could test this arc-ignition, nology-wires and magnecor-wires.  

Originally posted by geek Say your scope is 8-bits and we wanted to look at those few hundred mA or even Amps of current after the breakdown.   Let's say the breakdown current (    IB = U0/Z p ≈10 kV/50 ohm= 200A ) is close.  Note, it's a lot higher than I measure with my little grill starter.

Did you use resistor or non-resistor plugs? Resistance is usually 5000 ohm in resistor plug. If impedance is also close to 5 kohm then IB = U0/Z p ≈10 kV/5000 ohm= 2A. Probably easiest and chaepest way to improve ignition system effieciency is to use non-resistor plugs and resistor wires if stock plugs are resistor plugs.

There is always some capacitance and inductance in coils, plugs and wires. It would be nice if "pro"-ignition component manufacturers would tell these values. Nology does give understanding that only their wires have capacitance. Magnecor only talks of resistance but didn't mentioned capacitance, inductance or impedance of their wires.  


Edited by shmot - 29 October 2009 at 4:26am

Originally posted by geek Again, do you understand how to read a Smith chart?    If so, we can dive into this further, if not and you would like to learn a little bit about  S-parameters, the internet would be a great source.     I would start by reading about how TDR and VNA work.    You could even build a simple TDR cheap.

No, at least I don't remember it anymore. I do some research.

Originally posted by geek There are always complex components to any electrical device.   So your statement is correct.   But, for a manufacture to post this information would be very difficult and not very useful.   I think the average racer is concerned that their ignition is robust, what features it  has and cost.      If I buy a resistor, they will not normally publish the S-parameters for it.  Some parts like capacitors will some times have their resonance frequency listed.   All depends.       You understand there is a complex part to both the inductors and capacitors  (Xl, Xc) and these are a function of frequency.    However it goes so much deeper than this.      Normally when we deisgn high speed equipment we will measure the parameters we are interested in.   Even if you do a high speed circuit board layout, the board house will add test traces to the boards to measure them.    As you work at higher and higher frequencies, the more details become important.    Again, do you understand how to read a Smith chart?    If so, we can dive into this further, if not and you would like to learn a little bit about  S-parameters, the internet would be a great source.     I would start by reading about how TDR and VNA work.    You could even build a simple TDR cheap.

I see. It is needed to fit appropriate capacitor or inductor to measurement circuit so that complex part of the inductance is zero at scope frequency? How to get that "known source"?


Edited by shmot - 30 October 2009 at 3:00pm

Originally posted by geek I am not sure I understand what your asking.     Any part, even a wire, will change with frequency.   Equipment, like a VNA works by sweeping the frequency over some range and looks at the outgoing signal and what comes back.      So, say you take a wire.  Nothing special about the wire.  It has two ends.  Now you connect one side of the wire to a battery positive terminal.  What happens?     You may say nothing, or you put 12 volts on the wire.     Again, it's not quite so simple.  As this wave travels down the wire, it reaches the end.  The speed it travels depends on various things.  Once at the end, it reflects.    The VNA is looking at this reflection.    Knowing what we are sending out and what is coming back, we can figure out things, like impeadance.    Let me give another example with the wire and battery.   Let's say that you take the same battery and wire but this time have one side of the wire attached to the negative side of the battery first.  Then you touch the other side to the positive.    Sure you may burn your hands, blow the battery up, etc.   But, before all this happens,  at the time you first touch the positive terminal, the wave again starts to travel down the wire heading for the negative side.  Once it reaches the end, it still reflects.     The amplitude of the returned wave will be the same no matter if the wire was shorted or open on the other end.   What will change is the phase.     If you could get a VERY large spool of some coax wire, you could prove this to yourself with your scope.  The problem is that your scope is not very fast, so to get this much delay would require more wire than you want to drag into the shop.    People can use this same technique to find a break or short  in a spool of wire without ever taking it off the spool.      I'm sure you have a DVM.  It has an ohms setting.  This is the DC value of the part. The impeadance, being frequency dependant will have a shape to it and can be graphed.    This is where a Smith chart can be used.  It's nothing more than a way to graph  the impeadance.

DWM=digital voltmeter, VNA=vector network analyzer. Yes I have old Fluke and I do understand what is phase difference, but that was long time ago. This is first time when I do really need it. I did work with automation 15 years ago. Now I work with operations research and computer simulations.  

I did some testing with 200 psi and Nology non-resistor plug with 0.024" cap. There was similar breakdown glow on plug insulator between wire boot and plug ground.

Test system:


Glow:


I did wanted to know where that comes from and test it without boot:


It seems that glow starts from halfway of the insulator to the ground. Not from the tip. Weird. Maybe plug insulation will breakdown. I need to do same test for other plugs.

Now that you have a way to run some higher pressures, after a week or so of playing around, let me know what the plugs look like.   Mine all oxidized pretty bad when I was using air.    RH in the house has been around 65%.

Scottie we need warp drive now! I am giving her all I got Captain!