[Fred Winterburn]

Want to know the difference? Read my account of the experiment from a few years ago. Erase from your mind what you may have read before. Furthermore, a CDI does not need high voltage to outperform an inductive system, and the effective spark duration can be made longer by various methods. Fred


CD Ignition and its Superiority in Overcoming Shunt Resistances and the Fast Voltage Rise Time Myth
by Fred Winterburn, December 6 2018
I have learnt something that I should have known already with the amount of experimentation I have
already done, and not surprising considering this explanation has never been published before to my knowledge.
I had assumed for quite some time that the ability of CDI to overcome a shunt resistance (carbon fouling of spark
plugs or wet spark plug insulators for example) was due to sheer power, but this is not the major factor. Nor is it a
fast voltage rise time as is commonly written. It is easily proven that voltage rise time is more a function of the
coil characteristics and is similar no matter whether CDI is driving the coil or the coil is being used conventionally.
Voltage rise time certainly must matter and so must available power, but those two factors are not major
contributors to the reason why CDI is vastly superior in overcoming a shunt resistance. Those two factors are
important for any type of ignition system, but are eclipsed by the main reason, which is power factor.
When comparing Kettering systems using their intended coils to most CDIs that use low inductance coils,
one will notice on an oscilloscope, that CDI has a much quicker voltage rise, in the order of ten times faster
especially if they are small engine CDIs that need very low inductance coils due to the small capacity discharge
capacitors. Another reason it was, and still is commonly thought that voltage rise is the main contributor to
overcoming a shunt resistance, is when Kettering(points and condenser) and CDI are compared when using the
same coil. One will observe the voltage rate of rise is about 1.5 times faster with CDI. The condenser in the points
system changes the natural frequency, so the no-load sine wave frequency is slowed accordingly.
That same Kettering type coil switched by a transistor (pertronix for example), such that the condenser is no
longer in the circuit, will result in a voltage rate of rise that is close whether the coil is used inductively or driven
by a CDI. Furthermore, despite the voltage rise being quicker without the condenser, the condenser actually helps
with overcoming fouling by a couple of percent over Pertronix when all other factors are considered, such as the
1V, semiconductor voltage drop of the transistor switch. From those two observations, it can be deduced that it is
not the rate of voltage rise that makes CDI many times better at overcoming a shunt resistance. Also, a modern
e-core type coil meant for inductive systems that is more efficient than a canister style coil, will still be quite poor
at overcoming a shunt resistance. With one GM e-core type coil I tested, the ability to overcome fouling was even
worse than a standard canister type coil at 4 amps and that was with 8 amps through it. (approximately 4 times
the energy storage) So it is not voltage alone, or energy storage, or voltage rate of rise that is the main
contributor to overcoming a shunt resistance.
And thus a myth was born that it is a fast rising voltage that allows CDI to overcome a shunt resistance
much better than an inductive ignition. It turns out that CDI being vastly superior in this regard has more to do
with finesse rather than brute force.
The experiment that opened my eyes to this occurred on Saturday the 7th of October, 2018. I was re-testing
a CDI before I sent it out to a customer. A sudden moist warm front had come through the area and the inside of
my shed where the test equipment is located became dripping wet with condensation. Things were so damp I had
to wipe the meter faces on the instruments to read them. As part of the test I have the switch in STD mode to
prove the switch works and that the standby Kettering system will work if required. With the CDI switch in STD,
and the spark gap on the test machine set at 25kV, there was arcing to ground down the ceramic insulator post
for one of the spherical electrodes of the spark gap. Only 50% of the time was the spark crossing the actual spark
gap, the rest of the time it was shorting down the wet and dusty ceramic insulator. If I opened the spark gap to
30kV, all of the ignition energy shorted down the ceramic insulator with no spark at all across the electrodes. (the
ceramic insulator is about the same size and length as a spark plug insulator)
I decided to see how much worse or better it would be in CD mode. It was better beyond my expectations. I
fully expected with the increased power of the CDI that it would arc down both the ceramic insulator and the
spark gap, IE two sparks in parallel, and that it would be able to do so close to the upper voltage limit of
approximately 30kV for the CDI (with a standard canister coil). However that is not what happened at all despite
the higher wattage available with CDI. Instead, there was absolutely no shorting down the insulator post at any
spark gap. All of the sparks were across the spark gap where they should be. Even more interesting, is that with
the spark gap opened beyond 30kV and beyond the available voltage of the CDI, there was still no shorting down
the ceramic insulator !
My present working theory is that it has to do with power factor. A spark gap behaves like a leaky
capacitor before breakdown occurs, IE it is a capacitive load. Current(amps) leads voltage by something less than
90 degrees because of resistance in the load circuit, however the load is mostly capacitive. That type of load
wants a leading power factor, IE leading VARS (volt-amps-reactive). The dirty insulator is also slightly capacitive,
and likely also slightly inductive, but mostly resistive until flash-over when the resistance drops suddenly. A unity
power factor where current and voltage are in phase with one another is perfect for supplying resistive loads (no
VARS either lagging or leading). In Kettering mode, current lags voltage by well over 45 degrees (exact value
unknown and will vary with coils) as the field collapses. A lagging power factor results that is excellent for
supplying inductive loads (lagging VARS), but not efficient for capacitive loads that want leading VARS (IE, the
spark gap!). A large capacitor discharging through the primary winding of a coil provides a leading current or
power factor, of some angle greater than unity, but not a full 90 degrees leading. It certainly provides enough of a
leading power factor (leading VARS) to make the spark gap the preferred load over a resistive load providing its
resistance is high enough.
Another way of looking at it that might be simpler, is that a capacitive load such as the spark gap, prefers a
fast rising current rather than a fast rising voltage and with current leading voltage with CDI, the spark gap must
break down first. With current leading voltage, the voltage across the shunt resistance doesn't build up until it is
already too late. In the interim, the spark gap breaks down and becomes the low impedance load so 99% of the current
flow is through the spark gap every time.
In addition, an inductive system is somewhat self regulating as far as voltage goes. If a spark doesn't form,
the voltage keeps on rising until such time that the spark gap does break down, or more likely, that insulation
breaks down. Usually with an inductive system the voltage will rise far enough for current(amperes) to flow
somewhere, be it through fouling, or a carbon track inside the distributor cap created through previous misfires.
To make matters worse, an inductive system unless extremely powerful will have a higher voltage overshoot than
CDI. That is, the peak voltage at the coil secondary will be far higher than the actual break down voltage at the
spark gap even when there is a spark. The greater voltage overshoot with inductive ignitions is also due to the
current lagging the voltage. The voltage must rise to a higher value before there is enough current available to
pre-ionize the gap and provide enough current to actually break down the gap. It takes voltage and current to
break down a spark gap (IE power). A weak inductive ignition will show a higher voltage overshoot on the scope
compared to more powerful inductive ignition, because the voltage must rise even higher to provide the required
current. A CDI on the other hand, can be engineered with the voltage controlled at the source, IE the CDI power
supply, and as such the voltage cannot rise higher than the turns ratio of the coil dictates. So, if the voltage is
controlled to a reasonable value, insulation is protected despite the higher power of the CDI that can breach
insulating materials more easily. This is why it is imperative that CDI be voltage controlled at the source.
The experiment was inadvertent, but quite useful as it showed as nearly as possible how wet spark plugs
would behave in a car engine with both ignition types. Plug fouling inside the engine at the electrodes, or wet
electrodes would behave similarly with CDI preferring to fire through the spark gap rather than a shunt resistance.
The caveat to this is that not all CDIs will be this forgiving. It is only an advantage if the CDI voltage is controlled
to a reasonable voltage, which is not the case for most that have been in production. In my case it explains why I
never needed to change distributor caps, or wires on any of my old cars using similarly designed CDIs (Hyland
and others designed by my late father, Lloyd Winterburn). Fred Winterburn