Measuring Spark Energy ? 1

[Doodmeister]

Hi Guys,

I’m looking at comparing a few different ignition systems on a test rig and was wondering how I could go about measuring the spark energy output.
I’m a where that I could just increase the gap until the spark will not jump the gap any more and this would give me a visual idea of which is the more powerful but I’d like to get some trace data on an oscilloscope if possible.

Thanks for any input given.

 

[carnage1]

be verrrry carefull the primaries put out a lot of juice in there own right.
but if you have an attenuated probe that can handle 500v ish just hook it to the + on the coil
(you might want to use someone elses scope when there not looking)
and look for the largest area under the waveform.

 

[sdmays]

Are you talking about actually calculating the total number of joules? To calculate the energy under a curve, you integrate P(t) = v(t) * i(t) over one complete period(calculate the area the P(t) curve). For this you would not only need to measure v(t) with a 'scope, but you would also need to measure i(t) with a scope.

 

[btrueblood]

Measuring i(t) is fairly simple with an inductive probe (wrap one or two turns of fine-gage wire around the plug wire, and measure voltage across the ends of the fine gage wire). But 500v at the spark (+) sounds about 10-100x too low. I think spark gap voltages of 10,000 v to 20,000v sound closer to what you'll see in typical automotive cylinders.

 

[Warpspeed]

It is an interesting problem indeed........

The way I would do it would be to connect the secondary of the coil through a bridge rectifier into a fairly large value capacitor. This sounds rather odd, but I will try to explain.

Firstly the capacitor should be large enough to absorb the total coil secondary discharge energy without the voltage buildup on the capacitor becoming excessive. If you first discharge the capacitor to zero volts, then fire the coil, once, the amount of charge stored on the capacitor in millijoules can easily be worked out from the capacitance value and the dc voltage rise. If you zap it more than once, the capacitor voltage will rise in a series of equal energy steps, in other words like a charge pump.

Stored energy is a half CV squared. (Joules, Farads, Volts)

Now the diode bridge has a couple of functions. First it collects all the coil energy and all those nasty oscillations and stores them as a combined resultant dc charge, acting in a single polarity. Secondly, it prevents the capacitor from discharging back through the dc resistance of the coil.

The diodes only need to be relatively low voltage fast turn off types (1Kv) provided the capacitor final dc voltage remains safely well below 1Kv.

Please note I have never actually done this myself it is just an idea, but I think it will be about the only practical method of measuring the total coil energy where there is a spiky oscillatory waveform of non recurring symmetry. It is also a very simple test setup to specify, and it should be very repeatable no matter who does it.

Another problem you will face trying to do it any other way is the coil discharge is highly inductive, meaning voltages and currents are not in phase. Any clever means of calculating the energy from the frozen voltage waveform on a storage CRO sure is not going to be a simple problem.

I don't believe using a spark gap is going to help much either, only add more uncertainty. Hence dump all those millijoules into a capacitor, and just measure the resultant step dc voltage change with a very high impedance FET or vacuum tube voltmeter.

I have been designing switch-mode power supplies for many years, and an honest millijoule is an honest millijoule, but sometimes the waveforms and power factor can be quite misleading. Convert it to dc and the truth will be revealed.

 

[carnage1]

500v is the spike on the primary side not discharging into the plug (secondary side)

I really like the cap charge idea, but there is a possible drawback. the bouncyness after the main spike is something you don't want to measure, because it is not energy that is discharged into the spark. but it still might give repeatability and work well.

for a non CD system you could measure the max sustainable output current of the box, and the inductance of the coil. this will give you your max energy rating. (with a little math) then from there all you need is to calculate the rise time of the input waveform to see at what frequency your spark energy starts discharging before it peaks.

 

[ben91932]

Wouldnt it be simple to use an automotive oscilloscope?
Is voltage and time enough?
There are little spark check rigs available (lisle, kd etc) at any good parts store that haved an adjustable gap scale.
I have seen newer cars throw a spark 3+ inches... makes a loud *snap*. The highest voltage I've seen on a scope is just over 50kv.
Hope this helps,
Ben

 

[Warpspeed]

Ben, voltage and time will not give you a measurement of power. Current must come into it as well, but unfortunately current and voltage may be varying at totally different rates, and even in different directions during coil discharge. That is the nature of inductive circuits.

 

[carnage1]

are you looking for highest rpm operation, or for most spark energy?

 

[BrianR]

To measure the energy in an inductor you don't need voltage, you need current. E = 0.5*L*i^2

 

[Warpspeed]

Quite right Brian, but what if the inductance changes during discharge ? It can easily do that because the magnetising curve of the magnetic core will not linear, and not all of the stored flux can link the very heavily loaded secondary due to some leakage inductance. Quite a bit of energy can be clamped or otherwise absorbed in the primary circuit.

Your formula gives a pretty good rough initial guide to peak stored energy in the primary, but not all of it will be available during discharge across a loaded secondary.

A lot of energy can be dissipated charging up the various internal capacitances, between secondary turns and layers once primary current is interrupted. It is not a simple problem.