All but one of the Zeners I got from my friend's shop earlier this week turned out to be good so now I have a reasonable amount of data from which to draw conclusions.

The electrical setup needed for this is quite simple. I wired a precision 1.0 Ohm resistor in series with each diode and used a 3-1/2 digit Fluke voltmeter of 0.1% accuracy to measure the voltage drop across it to directly display the current to the nearest mA. This is more than adequate because the I-V curve of a Zener rises so steeply the accuracy of the current reading isn't the limiting factor. A second 4-1/2 digit Fluke voltmeter measured the voltage drop across the diode to the nearest millivolt. In this case the 0.1% accuracy means the absolute value could be off by as much as 14 mV, although relative values between Zeners are accurate to 1 mV.

The change of Zener voltage with temperature is a real issue so I'll start with the results on the three that came bolted to their original finned heat sinks.

For one of the three heat-sinked diodes I ran the current up to 1 Amp (~15 Watts), then up to 2 Amps (~30 Watts), then cycled back and forth a few times between there and 1 Amp. Each time I was at one of those current values I quickly recorded the Zener voltage. I determined from these measurements that even without air flow the heat sink was effective in keeping the Zener at constant temperature at these power levels, at least for the minute or so all the measurements took. That is, the Zener voltage I measured at 1 Amp after decreasing from 2 Amps was within a few millivolts of the value I measured when I initially had increased the current to 1 Amp when ramped up from 0. For this diode the Zener voltage at 1 Amp was 14.195+/-0.005 Volts. At 2 Amps it was 14.275+/-0.005 Volts. Extrapolating, at 4 Amps (60 Watts) it would be ~14.355 Volts, i.e. an increase of only ~1% from the 15 Watt value. Hence, to regulate a voltage at a fixed value one could do a lot worse than a Zener.

For the other two heat-sinked diodes I only measured the voltage after the initial ramp up to 1 Amp. For one of them it was 14.357 Volts and for the other it was 14.644 Volts (uncertainties in both are ~+/-0.005). Hence, the spread in values of these three was ~0.45 Volts.

For the other six diodes I didn't take the time to bolt them into a heat sink. Instead, I ran the current up to 1 Amp (15 Watts) and measured the Zener voltage as quickly as possible after I had stabilized the current (~5 seconds). Because of the internal heating caused by the current, the voltages continued to drift upwards but the range of values, if not the absolute values, are indicative of the spread of Zener voltages that would have been measured had they been in heat sinks. That spread was a full 1 Volt, or +/-3.5%.

This is a digression from my current interests so I didn't take the time attach thermocouples to the diodes, bolt them in heat sinks etc. However, I'm confident drawing conclusions from the results. My speculation is the measured spread of +/-3.5% I found for this limited number of diodes would increase to more like +/-5% with a larger sample size. This would be reasonable given the state of semiconductor production technology in the 1960s.

Assuming Lucas aimed for precisely the 14.84 Volts that Mark Z mentioned in his post, a +/-5% spread means at the low end some Zeners they supplied to the factories were 14.10 V and as a result undercharge batteries, while at the high end others were 15.58 V and boil their insides out. If someone got lucky and theirs was the perfect 14.84 Volts at fairly low speeds when it was only needed to dissipate 15 Watts, at high speeds when dissipating 60 W it would increase to 14.99 V, which by itself isn't too bad at all. However, because the heat sink can do only so much the temperature of the diode, and hence the Zener voltage, would increase further due to heating.

A typical 14 V Zener has a 0.08%/oC positive temperature coefficient while a battery needs a ~0.1% negative. For example, even absent internal heating from the current, a 14.84 Volt Zener that was optimum at 77 oF would increase to 15.02 V at an ambient of 105 oF whereas a battery needs 14.54 V at that temperature, which is nearly 0.5 V lower.

So, even a diode having a "perfect" Zener voltage at a given temperature isn't perfect when used as a voltage regulator on a motorcycle. That said, I haven't (yet) measured any of the modern replacement voltage regulators (Boyer, Mity Max, Podtronic, Tympanium, Wassell, others?) so I don't know if any of them do any better than a Zener for either the absolute voltage, or for the temperature compensation required for correctly charging batteries whether riding in snow or riding through the Sahara. However, any such measurements will have to wait until after I finish building the instrument mentioned in my post of December 20 and that's at least a few months away.

To digress for a paragraph, it turns out the temperature coefficient depends on the Zener voltage, and diodes smaller than ~4.5 V have coefficients with the opposite sign. This means four 3.71 V Zeners connected in series would have the perfect 14.84 V at room temperature, decreasing to 14.72 V at 105 oF, which is only 0.18 V too high. Similarly, cooled to 32 oF the Zener voltage would increase by 0.2 V whereas the ideal for a battery would be only 0.1 V higher than that. That is, four of these smaller Zeners in a good heat sink would regulate the charging voltage to within no worse than 0.2 V of ideal over the entire temperature range from freezing to above 105 oF.

To conclude this departure from magnetos, because of the apparent ~+/-5% variation in the ones originally supplied by Lucas the odds are the particular Zener that came in your bike is only doing a so-so job keeping the battery in good health. However, it only would take a few pieces of electronic gear to select a "perfect-ish" one that would do much better (within a Zener's intrinsic limitations due to the physics of semiconductors). Had I seriously thought I might ever want to "optimize" the one in my Trident I would have used this opportunity to select one from the batch from my friend's shop to do just that. Maybe someday I'll regret not having done so (and maybe I still will do it...). But, there are so many projects, and so little time.