I had looked at the google groups reference to the zener discussion ages ago, and that got me curious. Pulled some 20+ year old 2M2219A transistors by S?? and did some noodling around using my 7CT1N curve tracer plugin. When one floats the base and biases the emitter positive, one gets a breakdown near 6V that is "nasty" with a setting ot 1uA/div (1mA/div divided by 1000). Tweaking the DC voltage so it is just on, one gets a relaxation oscillator settable to about one burst per second. The highest peaks seen are about 130uA and the smallest peak is about
50uA; i did not change anything on the 7CT1N, i changed the effective sensitivity by changing the driven plugin from the nominal 100mV/div to a much less sensitive value of 2V/div. The impedance characteristics the 2N2219A sees changes as one changes settings on the curve tracer, and ruins the nice adjustability of the relaxation oscillator.
****** Using a 22Meg resistor in series with this "zener" from an adjustable supply, and connecting (the collector) to a 7A26 plugin input (DC coupled), and slowly increasing the drive voltage, one eventually sees a large spike, slowly settling down to a DC level that appears to be (Vsupply-Vz)*(1Meg/23Meg) in value. I did not wait to see if there would be a second "trigger" event; i had to reduce the supply to repeat the process.
Thought i should mention it, as this is in the (tens of) microamp region, and not in the milliamp region. BTW, the E-B zener is nice and stable...