Let’s Understand The High Tension Ignition
Final version Appeared In Gas Engine Magazine Feb/March 2021
Recently a friend’s 1952 Clinton engine wouldn’t start, Figure 1. After some analysis we narrowed the cause to spark. The component values may vary but all high-tension ignitions are schematically identical. That made it easy to quickly determine that the condenser was bad. At the time, in his parts drawer was an old automotive condenser, likely for a 1949 Chevy. Could it be used to get the Clinton running again?
Figure 1. 1952 Clinton.
Figure 2. The high-tension ignition.
Before answering that question, it is very helpful, if not necessary, to understand the high-tension ignition system and why the condenser is there. Figure 2 is the schematic of all high-tension ignitions. In these systems a changing magnetic field strength causes a current to flow in the primary coil through the closed points to ground. Changing is the key word in that sentence. The amount of voltage that drives the current in the primary coil is directly related to the rate of magnetic field strength change that the coil sees. A weak magnet will create a good spark if the system manages to change the field strength that the coil sees very rapidly. On the other hand, a very strong magnet will fail to deliver a good spark if the magnetic field strength changes slowly. It is easier, of course, to get a big change in the magnetic field strength quickly if you are starting with a strong magnetic field. As already said, all high-tension ignitions use the same electrical schematic. The capacitor value may change and the number of turns in each coil may vary but the base schematic never changes. Where they differ is the method used to create the changing magnetic field. Clinton and Maytag engines have a circular shaped magnet or magnets attached inside the flywheel Figure 3. As the flywheel turns, the coil sees no magnetic field then the N pole approaches then suddenly a switch to the S pole then back to no field.
Figure 3. Magnet (brown) inside a Maytag flywheel.
The popular Wico EK suddenly adds two large air gaps to the magnetic loop as it trips. Air is the worst possible medium that a magnet can see, adding those air gaps causes the magnetic field strength to drop like a rock. The Wico Series A, and many tractor magnetos, such as the Fairbanks Morse RV-2B, have a compact magnet that spins sitting adjacent to the primary coil. Those are three common methods of creating the changing magnetic field strength. Each creates a very different looking and seemingly different behaving magneto but from there on, all of them are identical. The Wico EK is very rectangular and boxy looking, tractor magnetos are shaped like a loaf of banana bread, Maytags don’t appear to have a magneto but underneath they all use the electrical schematic of Figure 2. It is rather simple schematic, consisting of the condenser, a set of points, a magnetic core used to route the magnetic field through and two coils on that core. The left coil is called the primary or sometimes the low voltage coil while the other one is called the secondary or high voltage coil. In operation the changing magnetic field is going to cause the primary coil to store up energy in the form of high current and low voltage. The system will then transfer that energy to the secondary coil at a much higher voltage but much lower current.
Before we try to understand why the condenser is there, we need to understand what it does and how it does it. A mechanical analog will help us do that. We will start with a sealed (no air can get in) full 50-gallon water drum. Both the left and right ends have an inlet/outlet pipe with a shut off valve. Attached to the right inlet pipe is an energy source, a pump. What makes this tank interesting is that it has a large stretchable membrane in the center that completely separates the left half from the right, Figure 4. In the end we want this tank to store some energy with the capability of doing some useful work for us.
Figure 4. Water tank with a rubber membrane.
With the left valve closed, we open the right valve and attempt to pump 3 gallon in. Nothing happens. In order to push water into the right side the membrane would need to stretch leftward to make room, but it can’t because the left side is full. Now let’s open the left valve and pump 3 gallons in the right side. The membrane stretches leftward and pushes exactly 3 gallon out the left side. We have put 3 gallon in but the drum is still 50 gallons. If we now close the right valve, we will read a pressure on the pressure gauge as the membrane tries to return to its neural position, Figure 5.
Figure 5. Water added to the right side.
If we open the right and left valves and pump another 3 gallons in, 3 gallons will leave the left side as the diaphragm stretches more. Closing the right valve, we would now read a higher pressure. We might note that at this point the ground water on the left has an excess of 6 gallons while the ground water on the right is reduced by the same amount. If we set up the same experiment with a drum of the same length but with a smaller diameter, the extra 6 gallons will have stretched the membrane much more and we would read a much higher pressure.
Figure 6. Small diameter tank with 6 gallons added.
With the right valve closed, and 6 extra gallons on the right side under pressure, the system is storing energy which came from the pump. We could use that energy to do something useful. We could close either valve, disconnect the tank at the shutoff valves and wheel it across town to the local mill and release it slowly to turn a waterwheel to grind flour.
Now we can apply those principles to the condenser. The symbol we use to represent a condenser is very similar to its actual construction: two electrically conducting plates separated by a nonconducting layer. Also similar to the two halves of our water tank that are separated by a membrane. In the days of our old engines a common implementation was two sheets of aluminum foil separated by a sheet of waxed paper then rolled up like those jelly cake rolls you see in the grocery store.
Figure 7. Unrolled condenser.
In Figure 7, I have partially unrolled a typical early 1900’s condenser. On the left you see a layer of waxed material and under that the upper plate. Mid picture I have pulled up the upper plate to reveal the lower plate. The brass plate is the contact to one of the aluminum sheets.
Keeping in mind the water tank, in Figure 8 the condenser is in its initial state (tank is full no water pumped in), both plates at the atomic level have exactly the same number of positive and negative charges, they are electrically neutral. I have shown 4 on each plate but the actual number of those positive and negative charges is huge: it would be a 1 followed by zeros across the page.
Figure 8. A neutral uncharged condenser.
In Figure 9, we have a 0.2µF (gallon) condenser with switches on each contact as well as a voltage source (pump) and a voltmeter (pressure gauge). The left switch, goes to ground, and is open (nonconducting), if we close (conducting) the right switch and apply a voltage, nothing happens. This is represented by the tank with the left valve closed. This is the second most common failure mode of condensers, one of the leads becomes disconnected from its brass plate. We call that an open. Now we close both the right and left switch and apply a voltage. Depending on the condenser size and voltage used, a given number of positive charges will flow onto the right plate. In Figure 9, I show the excess positive charges on the right plate. I have shown six excess charges, in reality, for even a very small voltage the number of charges is gigantic. With this excess charge an electric field, similar to the rubber membrane, develops between the plates which causes an interesting thing to happen: the exact same number of positive charges leave the left plate.
Figure 9. Charged condenser.
So, for every positive charge that is accumulated on the right plate a positive charge departed the left plate, the condenser as a whole is still electrically neutral. It still has the same total number of positive charges; however, some are not where they would like to be. Like the water tank, on a larger scale, the whole world is still neutral as the battery took those extra charges that are on the right plate from ground on the right while those missing on the left plate can be found at ground on the left. The device has not condensed charge, it is still neutral, therefore for the rest of this paper we will call them by their correct name, capacitor. With the capacitor charged, if we open the right switch the voltmeter will read a voltage. If we increase the stored charge by pushing more current in, the capacitor voltage will increase. In a similar manner, if we have the same number of charges on a smaller capacitor the voltage will be higher. In either case, a very important point is that the capacitor is storing energy that we could do something with. For example, we could attach a small LED bulb to the capacitor and close the right and left switch, we would see a flash as the capacitor discharged, dumping its stored energy into the bulb. In a more notable example, I received some capacitors from Lighting Magneto for evaluation. The last test was a 600V leakage test, after which the units were placed on the bench. After more than 15 minutes, during cleanup, the first capacitor was accidently picked up by its leads – wow, it had 600 volts worth of stored energy still onboard. Like the water tank, for a charge to leave one side of a capacitor another charge must enter the other side. Had I picked that capacitor up by one lead only, the experience would have been a lot less painful.
It is very difficult to count those excess positive charges but easy to measure current in Amps. Earlier I said there would be a gigantic number of charges, to be more specific, 1 Amp is 6240000000000000000 positive charges per second. If we force 1 Amp of current into the capacitor, that many charges per second will be accumulating on the capacitor plates and the voltage will be rising at a rate dictated by the capacitor value. Each excess charge in Figure 9 then, really represents more than a hundred trillion charges.
The most common failure mode, and one you have likely experienced, is for some form of conductive path to develop through or around the waxed film separating the plates. Those positive charges we have stored on the right plate will migrate to the left side. The capacitor is leaky, similar to our water tank having a small hole in the membrane. I learned that the Lightning Magneto capacitors don’t leak. Modern capacitors are still parallel plates but the dielectric (separation material) and the lead connections are greatly improved making today’s versions much smaller and much more reliable.
In order to put the high-tension system all together, a short discussion of how coils, inductors, behave is needed, see Low-Tension Ignition , Gas Engine Magazine Oct Nov 2020. They are a bit like big flywheels, it takes time and a lot of effort to get a flywheel spinning 500RPM. Once spinning it is storing a huge amount of energy and will not stop spinning and come to rest until it has dissipated all that stored energy. If coasting to a stop, most of that energy is converted to heat in the bearings, cylinder wall, etc. With a lot of effort, it can be made to stop more quickly but it can not be stopped instantly. In a like manner, it takes effort and time to get current flowing in a coil, once the current is flowing the coil is storing energy in a magnetic field that builds up around it. If the force, voltage, that caused the current to flow is removed the coil current will continue flowing until that stored energy is dissipated. Energy is always conserved; it can’t just disappear. In our case it is converted to heat as current flows thru the coil’s internal and external resistance. The current can be stopped more quickly with effort, resistance or an air gap, but it can not be stopped instantly because the stored energy can not be dissipated instantly. Think of a coil as giving current momentum. It takes effort and time to get it rolling but once rolling it is very difficult to stop. Stopping it more quickly, for example, in the case of a low-tension rotary magneto means tripping and opening the igniter. The current can not stop instantly, it will only stop after all the energy is dissipated so it jumps across the igniter points dissipating a lot of energy very quickly.
Now we can look at how the high-tension ignition works. In Figure 2, when the points are closed, we have a changing magnetic field creating a small voltage and high current flowing out of the primary winding through the points to ground. The capacitor is uncharged, sitting at zero volts because the closed points have it shorted out. Primary coils will typically have a resistance of less than 1 Ohm, often near 0.25 Ohm, so a very small voltage will create lots of current. We know then, the coil is storing lots of energy. The stored energy is ½ L I2 where L is the coil’s inductance (number of turns, core size, core material, etc.). Because energy goes as current squared, doubling the current creates four times more energy. The primary coil’s stored energy will become the totality of the spark energy so it needs to be maximized by maximizing the current. Everything in the sparkplug spark originates as energy stored, current, in the primary coil.
Figure 10. Primary side of a high-tension ignition without capacitor.
Figure 10 is a stripped-down version of Figure 2. For discussion the secondary coil is left off as well as the capacitor. The circuit is identical to a low-tension rotary magneto and igniter. Initially the points (igniter) are closed, the magnetic field is changing, current is flowing in the coil and energy is being stored in the primary coil. The path of the current is in red. When the igniter trips (points open), the system has asked the primary coil current to go to zero instantly. From above, this is not going to happen. Current is going to continue to flow until the coil has dissipated all its stored energy. The current has momentum. If it’s going to continue for an instant, it must jump, arc, across the points. That arc will dissipate a huge amount of energy very quickly. If that happens the energy that was going to be transferred to the sparkplug just went up in smoke, in the form of a spark. Wasting the collected energy on the primary side as a spark before it can be transferred to the sparkplug doesn’t seem like a good idea.
Figure 11. Primary side with capacitor.
Figure 11 adds the capacitor back in. Again, with the points closed, current is flowing and energy is being stored in the primary coil. With the points closed, the current follows the path shown previously in Figure 10. When the points open, the coil starts dumping its stored energy by shoving current (positive charge) onto the capacitor, the red path of Figure 11. Rather than going up in smoke as a spark across the points, the primary coil energy has now moved over and is sitting on the capacitor (voltage and positive charge). The situation now is very similar to the water tank with six gallon pushed in. The capacitor is not really happy being loaded up with all this energy. Using the voltage it has developed, like the water tank pressure, it shoves everything back to the coil. The coil in turn shoves it back to the capacitor. The system oscillates. Figure 12A shows the voltage on the capacitor and coil as the points open. The energy, in the form of current and voltage, is first heading toward the capacitor then turns around and heads back toward the coil then back toward the capacitor again. The process continues, with each pass less in magnitude, until all the energy is dissipated.
Figure 12A. Wico EK voltage on the coil with a small 0.06µF capacitor. Vertical scale 100V per box, horizontal 50µs per box.
Figure 12B. Wico EK voltage on the coil with a large 0.78µF capacitor.
Now we can begin to answer the question of whether or not the 1949 Chevy capacitor can be used to get the Clinton running. As the value of the capacitor is lowered, similar to reducing the water tank diameter, both the voltage and frequency of the oscillation will increase. Conversely, if the capacitor size is increased the voltage and frequency of the oscillation go down. Nothing else changes. Figure 12A and12B compare the voltage and frequency of a Wico EK magneto with 0.06µF and a 0.78µF capacitor. The larger capacitor, Figure 12B, clearly has less amplitude and is a lower frequency.
Now we are ready for the final step, transferring this oscillating energy over to the sparkplug. The primary coil and the secondary (high voltage) coil are wound on the same magnetically conductive core and become a simple transformer. As the name implies, a transformer transforms voltage and current. A typical Wico EK magneto has 196 turns of 22-gauge wire in each of the primary coils and 8400 turns of 40-gauge wire in each high voltage secondary coil for a ratio of 43 to 1. The voltage available at the sparkplug will be 43 times larger than the voltage on the primary coil and capacitor but the current will be reduced by 43. So, the transformer moves the primary coil’s energy over to the sparkplug at a much higher voltage but lower current. But here is an important point, the transformer is like the magnetic field, it only works when things are changing. Transformer action won’t work if the primary coil has a steady DC voltage. The primary coil voltage needs to be changing or oscillating. It now appears pretty convenient to have added the capacitor to the primary coil causing that side to oscillate, enabling transformer action.
Figure 13 demonstrates the transformer action when a Wico EK is tripped. In order to keep the high voltage in the range of the oscilloscope the coil and capacitor voltage were purposely lowered by increasing the capacitor to 33µF. The yellow trace is the capacitor/primary coil voltage displayed at 20V per vertical box and peaks at 60V. The blue trace is the open circuit (no sparkplug) output voltage of the high voltage coil displayed at 500V per vertical box, peaking at 2640V. The voltage multiplication is 44, very close to the turns ratio of 43. The error here is probably in reading the scope voltages or my counting the 8400 secondary turns.
Figure 13. Demonstration of transformer action.
A quick summary of the purpose of each component in Figure 2 might be helpful. The primary coil’s job is to store up as much energy as possible during the time the magnetic field is changing and the points are closed. Its second job is to transfer that energy to the secondary coil by transformer action when the points open. The point’s function is to short the primary coil so it can maximize its current and thus maximize its energy collection. The points also precisely time when the spark occurs at the sparkplug. The secondary coil’s job is, by transformer action, to accept energy from the primary coil at such a high voltage that a spark at the sparkplug occurs dissipating all the energy.
And finally, the purpose of the capacitor. First it keeps the energy built up and needed in the primary coil from being destroyed as a spark across the points. A useful advantage, although not its purpose, is that it also reduces points pitting. Second, it causes the primary side to oscillate thus enabling transformer action to move and transform the energy to the sparkplug. Changing the value of the capacitor only changes two things: the voltage of the oscillation and the oscillation frequency. Neither change significantly so it’s likely the 1949 Chevy capacitor will work just fine in my friend’s Clinton engine.
After getting the Clinton back up and running it seemed that an experiment involving capacitor value was in order. A real test might be to up size and down size the standard 0.2µF capacitor 2X, but 3X or 4X would really make a point. In addition, a 1949 Chevy going 5000 miles prior to changing the points and condenser would have mechanically bounced down gravel roads and open and closed the points more than 40 million times. That capacitor might be a fair test also. In the end, the original in each magneto (generally about 0.22µF), a 0.06µF, a 0.78µF and a 1949 Chevy capacitor were compared. The magnetos tested were a Maytag on a testbed, Figure 14, a Wico EK on a testbed, Figure 15, a running Clinton seen in Figure 1, a Wico EK running on an Hercules, Figure 16, a Wico Series A running on a 3 horsepower John Deere EP Northern, Figure 17 and a Fairbanks Morse RV-2B. This resulted in data on all three types of changing magnetic fields mentioned earlier. There are several questions to be answered: would some or all of those capacitor combinations start and run an engine smoothly and do so without excessive arcing across the points which would lead to pitting. There are several parameters specified on capacitors but unless you dig deeply most will be specified by only two parameters, capacitance value and operating voltage, so determining the maximum voltage on each capacitor would be helpful also. Model T buzz coils are electrically very noisy with high frequency noise spikes so another capacitor specification, maximum rate of voltage change, must be watched for them. Although buzz coils use the same schematic as Figure 2, they are not addressed in this paper. Yes, buzz coils and all pre-electronic ignition automobiles use the Figure 2 schematic. Rather than using a changing magnetic field to create the current in the primary coil they use a battery.
Figure 14. Maytag testbed.
Figure 15. Wico EK testbed.
Figure 16. Wico EK running on a Hercules.
Figure 17. Wico Series A on a John Deere EP Northern.
Before looking at the results, let’s review what should be expected. First, the energy collected and stored in the primary circuit is unaffected by capacitor size. The primary coil and changing magnetic field determined the collected energy, therefore the amount of energy available to be transferred to the sparkplug is unaffected. The size of the capacitor will affect the frequency and amplitude of the primary circuit oscillation. Frequency has very little effect but voltage needs to be looked at. If too high, the max capacitor voltage rating could be a problem, if too low the transformer turns ratio might produce an inadequate voltage for the sparkplug. Arcing of the points is a horse race as air arcs at about 70V per 0.001 inch. As the points open, they begin to move apart at a rate determined by the cam profile and the engine RPM. As this happens the capacitor begins to take on charge (current) from the primary coil, its voltage starting at zero begins to rise upward at a rate determined by its size and the amount of current coming from the primary coil. The rate that it rises, in volts per second, is I/C. For example, forcing 1Amp into a 0.2µF capacitor will cause the capacitor voltage to rise at a rate of 5 million volts per second or 5 volts per microsecond. If the points get to 0.0005” before the capacitor gets to 35V, 7 µs in our example, or the points get to 0.001” before the capacitor gets to 70V all is OK. If the capacitor voltage goes up too fast a spark will occur across the points. That spark is difficult to see with the naked eye but fortunately very easy to observe on an oscilloscope trace as an instantaneous drop in capacitor voltage.
Now the results. The Maytag testbed, the Wico EK testbed and the Fairbanks Morse all showed good consistent spark for all capacitors at all speeds (Maytag and FM). The Clinton, and the John Deere EP with a Wico Series A all started easily and ran well on all capacitors. The Hercules started and ran nicely on the 0.06µF, 0.22µF and 49 Chevy capacitors but was difficult to start and ran poorly on the 0.78µF capacitor. It did start easily and run smoothly on a 0.68µF capacitor however (3X the standard). The condition of the magneto magnet and its points gap are unknown. It was also later learned that the Hercules was very sensitive to trip mechanism set up. The capacitor oscillation voltage and frequency went up, in the four cases measured, with decreasing capacitor size as expected. For the 0.06µF capacitor, which is the worst case, the maximum voltages were as follows: for the Maytag 75V, for the Wico EK testbed 400V, for the Wico Series A 250V. The Maytag however had a noise spike, coming from the sparkplug, of 125V. The noise spike was the same for all capacitor values, no other magnetos showed a significant noise spike.
The Maytag, Wico EK testbed, Wico Series A and the Fairbanks BV all showed the points arcing under some conditions (running or starting) for all capacitors except the 0.78µF. Physical location made it impossible to get a scope on the Clinton and Hercules. It was later determined that above about 0.47µF all points arcing ceased. Surprisingly it did not appear that the 0.06µF capacitor caused significantly more points arcing than the base capacitor or the 49 Chevy. Points arcing and sparkplug sparks are similar to lightning in that what appears to your eye as a single flash is actually made up of a multitude of back to back short duration arcs. In Figure 18, the blue trace is the capacitor voltage at 50V per vertical box. Immediately after the mag trips the capacitor voltage starts moving, then twice it abruptly goes back to zero before getting off to its nice sinusoidal decay as energy moves to the sparkplug. Those two abrupt returns to zero are each a small spark across the points. The yellow trace is the sparkplug voltage at 2kV (2000V) per box. The plug arc is made up of short bursts of very short bidirectional arcs. In all of the magnetos studied here, at most one or two of these short duration points arcs were observed when the points opened. Those one or two arcs apparently give the points time to separate far enough apart to save even the 0.06µF capacitor from further arcing and wasting of energy before it could be transferred to the plug.
Figure 18. Sparkplug voltage in yellow (2kV/box), capacitor voltage in blue (50V/box), after an EK mag trip (0.22µF, 0.020 plug gap).
Although primary coil current is unaffected by capacitor size, for reference, low tension rotary magnetos tend to generate about 1A in their coil while in these high-tension experiments the Wico EK generated about 2A, the Maytag generated about 1.5A at running speeds while the Fairbanks Morse generated about 2.4A. Figure 19 is a condensed summary of data collected.
| DC | Maytag | Wico EK | FM RV-2B | Series A | Clinton | Hercules | ||
| Testbed | Testbed | Testbed | JD EP | Engine | Engine | |||
| Base C | Max V | 40V | 90V | |||||
| Frequency | 33kHz | 13.3kHz | ||||||
| 0.062µF | Max V | 75V | 400V | 150V | 250V | |||
| Frequency | 58kHz | 20kHz | 35.7kHz | 25kHz | ||||
| 0.78µF | Max V | 30V | 225V | 60V | 150V | |||
| Frequency | 12.5kHz | 5kHz | 7.1kHz | |||||
| 49 Chevy | Max V | 40V | 300V | 90V | 200V | |||
| Frequency | 40kHz | 9.1kHz | 10kHz | 16.6kHz | ||||
| Good Spark | Yes | Yes | Yes | Yes | Yes | Yes | ||
| All Conditions | ||||||||
| Run & Start | Yes | Yes | 0.78 poor | |||||
| 0.68 fine | ||||||||
| Max Primary I | 1.5A | 2.0A | 2.4A | |||||
| Arc Points | Yes | Yes | At detent | Yes | ||||
| Turns Ratio | 43.00 | |||||||
Figure 19. Condensed summary
There is a lot of information and even more misinformation on various websites instructing us on the proper capacitor to put in a given high-tension magneto. My conclusion, however, is that the high-tension ignition system is very tolerant of capacitor size. Although this discussion was nontechnical, a rigorous mathematical analysis of Figure 2 also puts very little restriction on the capacitor value. The capacitor size sets the oscillation voltage and should not exceed the capacitor rating but should be high enough to spark the plug. For reasonable deviations around the standard 0.2µF capacitor, engines will run and start fine. The energy collected and delivered to the sparkplug is independent of capacitor size. The maximum voltage on the capacitor, again for reasonable deviations, is below 300V making the often quoted 600V specification for engine ignition capacitors very adequate. Unless the capacitor size is doubled to about 0.47µF all four magnetos tested with a scope showed minor points arcing under some condition.