Appendix II: Replacement Condensers for Post-WWII Magnetos
Note: A version of the material in Appendix I and II was part of a series of two articles I wrote for the Fall and Winter 2011 issues of 'The Antique Motorcycle,' the journal of the Antique Motorcycle Club of America. Tests to Find a Suitable Replacement Condenser
The full range of parameters that affect a magneto's condenser are current, voltage, temperature, and frequency (i.e. for a condenser, it's not just the magnitude of the voltage across it, but how fast it is applied). Some of the less-common instruments that have allowed me to make my studies of condensers include a variable frequency impedance bridge for determining AC losses to within 1%, a resistance bridge with maximum range 50,000x higher than a standard "megger" insulation tester, and a 200 Watt pulse generator for high current pulses of rise time 0.01 us and repetition rate up to 1 MHz. I also have a four-channel 400 MHz oscilloscope, a 40 kV probe for directly measuring the time dependence of the output of the coil up to 75 MHz, a 2 kV probe for directly measuring the voltage across the condenser up to 100 MHz, and specialized current probes for directly measuring the flow in/out of the points, condenser, and high tension leads up to 60 MHz. Stated differently, this oscilloscope and probes have allowed me to simultaneously measure all electrical parameters of operating magnetos to within less than 0.02 us of the onset of ignition. Even on an engine at 6000 rpm, in 0.02 us the contact breaker points have opened by just 36 uinch (0.92 um), which is only twice the wavelength of light.
As for how I conducted my studies, after using manufacturers' specifications to select the most promising replacement condensers (described in the next section), I tested them when attached to an armature and contact breaker points, as well as with other appropriate laboratory tests made directly on the condensers. Using a modified distributor tester and two commercial magneto testers I was able to simulate actual field conditions in my laboratory, including with a condenser and coil in an oven at temperatures up to 150 oF (65 oC). Where appropriate, I then used those measurements to accurately simulate certain parameters under accelerated and/or overstressed conditions. For example, I used a 200 Watt high frequency pulse generator to subject the replacement condensers to current pulses 60x higher than the ones I had measured using a magneto tester, doing so at the equivalent of >100,000 rpm so that I was able to simulate "150,000 miles" of operation in minutes instead of months.
Some of my tests were a combination of simulations and "field conditions." For example, one long-term, test involved submerging condensers in 181 oF (83 oC) beakers of 30W Castrol and hot water for a number of months, taking them out periodically to measure using a low voltage capacitance meter (which measures them at only a few volts; much less than the several hundred volts they are subjected to during operation in a magneto) and a General Radio 500 V Teraohmmeter (which measures their resistance at an appropriately high voltage, but at DC rather than the ~100 Hz repetition rate of a magneto). Although neither of these electrical measurements tested performance under operating conditions, I designed this aggressive "environmental simulation" to test their ability to survive heat and solvents -- it would have been no good if a possible replacement condenser had the necessary electrical performance, but if it degraded in the presence of oil vapor or humidity.
The water test was particularly harsh since, even if a magneto fills with water, it will quickly dry once it is operating again, so seldom will the condenser of a functional motorcycle be in contact with liquid water for more than a short period. While the ones in oil had no detectable change in their capacitance or resistance, beginning at 1680 hours— the equivalent of them having spent 42,000 miles at an average speed of 25 mph submersed in hot oil and water — the resistances of the ones in water started dropping with time, from over 2 TOhm when new to ~100 MOhm at 3624 hours. However, 100 MOhm is still much higher than required to function in a magneto. Further, the ESR at room temperature of even the most degraded of them was still 3x lower than the lowest ESR I have measured of a Lucas
condenser removed from a functioning magneto. Plus, the ESR remained unchanged at elevated temperatures, while by 120 oF (49 oC) that of the still-functional Lucas
had increased a further 3x, to a value 10x worse than that of the most degraded replacement condenser.
Since my tests had established that prolonged immersion in hot water degrades the electrical properties of the condensers, albeit very slowly, at the 3364 hour point ("90,600 miles") I removed the water. After a further 120 hours ("3000 miles") in air at 181 oF (83 oC), the resistance of even the most degraded condenser had recovered to above 1 GOhm, and its ESR had improved by 58%. After an additional 336 hours ("11,400 miles" total since removing the water) the resistance was over 1 TOhm and the ESR more than 90% of its as-new value. The fact these properties were easily reversible indicates the measured degradation was due to the electrical conductivity of tap water that had slowly permeated the protective coating, rather than a permanent water-induced chemical breakdown of the dielectric. This means that in the actual operating environment of a magneto these replacement condensers would operate significantly longer than the at-least "90,600 miles" they survived immersed in hot water.
Periodically during my long-term "environmental" test of their resistance I also made a full set of measurements on these replacement condensers. After "90,600 miles" in hot oil and water, these replacement condensers then survived the equivalent of an additional 150,000 miles subjected to current pulses 60x higher in power than are generated by a magneto. I also connected the condensers and an armature coil to one of my magneto testers and measured all of their electrical properties at 150 oF (65 oC), comparing these measurements with ones I had done on them when they were new to see if I could detect any degradation. Judged from the lack of any apparent increased sparking at the contact breaker points, and unchanged oscilloscope patterns, all of these replacement condensers performed as well in a magneto after "90,600 miles" in hot oil and water as they had when fresh out of the box. Continuing on with hot oil only, my final complete set of measurements was at 7080 hours ("177,000 miles"), with the condensers again passing all the tests.
One typical accelerated lifetime test of electrical components is based on the observation that most chemical reactions approximately double in rate for every 10 oC increase in temperature. This "doubling rule" makes possible another kind of lifetime estimate. Assuming it applies to the chemical processes at work breaking down the dielectric material of the replacement condensers, surviving 7080 hours at 181 oF (83 oC) predicts they would survive at least 51 years parked in a storage shed at 73 oF (23 oC). What Replacement Condenser Should Be Used?
The replacement condensers I recommend are a pair of Panasonic 0.082 uF polypropylene film capacitors (part no. ECQ-P4823JU). When soldered in parallel they produce a 0.16 uF condenser that fits into the available space in the end caps of Bosch, Lucas
and BTH single and twin rotating armature magnetos. I hasten to add that it is possible to damage them by applying too much heat during soldering so, if you are not careful, they can be inadvertently made to fail before you even start. This capacitor has the published specifications that caused me to select it for my tests, the demonstrated electrical performance to survive the high voltage, high current pulses generated by a magneto, and the ability to survive the hostile environmental conditions of heat, oil, and humidity.
Although I believe the tests I've conducted are as thorough and comprehensive as they need to be, and although none of the capacitors failed, to extract a statistically meaningful minimum expected lifetime would require subjecting a much larger number of them to these tests. However, based on my measurements on a limited number of units, my conservative estimate is there is a very high probability these Panasonic capacitors will function without failure in a magneto for at least 140,000 miles or 40 years.
Unfortunately, these Panasonic capacitors are now out of production. However, certain other polypropylene film/foil capacitors made by other manufacturers are likely to function just as well as the ones from Panasonic, although I cannot recommend them until I have an opportunity to test them.
Importantly, no electrical measurements are even needed to know that any capacitor that easily fits into the cavity of an armature, with half the space left over, definitely is not up to the electrical rigors it will face. A half-century of developments in chemistry has resulted in significantly improved capacitor lifetimes, but no amount of development can overcome fundamental laws of physics. Surviving high current pulses requires relatively thick electrodes. Surviving high voltages requires relatively thick dielectric layers. From the known electrical properties of materials, what this means is any suitable condenser for this application necessarily must be quite substantial in size. Physics, not coincidence, is why the soldered Panasonic pair is remarkably similar in total length, width, and thickness to the magneto condensers it replaces.