A further blog in my series about transistors [I think I've given myself a lifetime job, mining this rich seam]. This
one looks at the reverse breakdown of the base-emitter junction and one application of the transistor if it's used in
this way [actually, I only know of one application, but no doubt D_Hersey will be along in a while to quote half a
dozen from the Electropedia that sits inside his head].
On a datasheet the maximum figure for this junction is often given as 5V or 6V (for the kind of silicon, small-signal,
bipolar junction transistor that I'm using in my experiments).
Obviously, the manufacturer gives themselves some leeway, so the actual breakdown voltage is going to be several volts
greater than that. Unlike the case of the collector-emitter breakdown blog, where I used a switching circuit with an
inductor, here I'm just going to treat the junction as though it were a zener diode and put a resistor in series to
limit the current. These test circuits are getting simpler and simpler.
Here's the circuit:
The simulation just reported that the voltage at the transistor-resistor node was 12V, so it doesn't model the
breakdown (at least, not for the 2N3904 part that I selected from the library).
Before I try it, I'm going to do something that won't necessarily make much sense now, but you'll understand why I've
done it a bit further down; I'm going to measure the beta of the part [I haven't done beta yet - talk about
unstructured blogging - but it's the DC current gain of the transistor, the ratio of collector current to base current
measured at a particular collector current, and on a datasheet it's often referred to as hFE - I haven't done hybrid
models either yet, but if I keep throwing in these little morsels it'll all seem quite natural and familiar when I
eventually do]. Here is what my cheap meter with a transistor test facility says for the gain:
Here are the results of the reverse breakdown:
and here's what it looks like on an oscilloscope:
So I've got myself an 8.7V zener.
Now I've had the transistor running like that for 10 minutes, I'm going to take it out of circuit and measure the beta
again. What we can see is that the breakdown has done some damage to the device and affected the gain.
I have read in books that there's a certain amount of recovery of the gain afterwards, but I'm not seeing that with
this device; several hours later it is still reading 158.
We haven't had any Shakespeare for a while, so a quick quotation
"Be not afeard; the isle is full of noises, ..."
William Shakespeare. Tempest [III, 2]
The one application that I know of for the reverse base-emitter breakdown is as a noise diode. The avalanche effect is
quite noisy and chaotic, and if it's ac coupled and amplified it makes a reasonable white noise source, so I'm going
to try that here. Apparently, transistors vary widely in the amplitude of the noise they generate in this way, so if
you were doing this for real you'd probably buy a batch and pick out the 'good' ones. I just want to look at it and
see it there, so I'll keep working with the device I've been experimenting with - but bear in mind it might not be
Here's the noise, as seen by my scope. Unfortunately, that's the best the scope can do.
Before you go away with the idea that all of that comes from the emitter-base breakdown, here's what the scope shows
when the input is set to ground.
So that I could look at the noise in a bit more detail, I assembled a simple gain stage from a pair of TL081s on a
breadboard - I gave each a gain of 10, so overall this is one hundred times the noise generated by the transistor. I
also modified the simple circuit of the transistor and resistor by substituting a simple current source for the
resistor, since in the original form it wasn't too happy driving the amplifier.
Here's a single trace of the noise (if you're attentive, you'll see that mixed in with the noise is something that
looks a bit more regular and that's because the circuit is picking up some interference).
And here's the accumulated trace.
It looks a bit biassed one way (the view is ac coupled), so the noise isn't symmetrical, which probably reflects the
physical basis of its generation. I could probably have done this a bit better, but it gives an idea of how it coud be
used as a noise source with a bit of work.