interesting article on diodes

[pinkjimiphoton]

https://www.analog.com/en/analog-dialogue/raqs/raq-issue-108.html

[Rob Strand]

It's an old idea but a cool one.

Connect an LED to a multimeter set to volts.  Then hold the LED into a light source.  You will see a voltage.  If you have low current range on your DMM you can also measure the current.   (This test does not damage the LED in any way.)

There was a really cool puzzle in one of Bob Peases articles, maybe from the mid 90s.  It showed how you could create a negative voltage using a transistor.   

It goes like this,
- Take any NPN silicon transistor
- Ground the base
- Connect the emitter through 1k to +12 V 
   This causes the BE junction to breakdown like a zener.
   (At this current it will stuff the transistor so you should chuck it out after the test.)
- Leave the collector unconnected
- When you measure the voltage from ground to the collector you will get -0.3V  !!!

How can it be?   

At the time my answer for the puzzle was wrong. 

In a later issue Bob Pease explains when the BE junction breaks down it generates light inside of the transistor.  The light from the BE junction shines onto the CB junction and causes it to generate a voltage.  The BE junction is like a lighting LED and the CB junction is like the receiving LED in the above scenario.

One of the coolest puzzle of all.
Spice will not work because it doesn't know about the light shining inside of the transistor.

[pinkjimiphoton]

that is too cool, rob!!

[pinkjimiphoton]

hey, weird thought, but what about using leds as clippers on a standard led kinda clipper circuit, and using the light from the leds flashing to get a voltage flowing in an led somewhere else in the circuit?? like.... a germanium simulator or something crazy? add leakage to a silicon transistor fuzz maybe? if ya did it to a bazz fuss style clipping stage, would it make it soften the distortion character of the fuzz?

pipe dreams? lol

[Rob Strand]

that is too cool, rob!!

It's one of my favs.  It's a weird thing and there's no way to understand it without obscure knowledge.   Apparently it didn't take Bob Pease long to work it out.

hey, weird thought, but what about using leds as clippers on a standard led kinda clipper circuit, and using the light from the leds flashing to get a voltage flowing in an led somewhere else in the circuit??

The coupling is pretty weak but I guess it's unexplored territory.  My feeling is getting a germanium softness is unlikely.

[pinkjimiphoton]

i'm a sucker for anything converting electrons to photons

i wonder if a burnt junction is part of what makes some really leaky ge q's so they will pretty much work even if flipped 180 degrees.

but a slight bit of leakage... yeah, i guess it likely wouldn't "add" to the sound, other than maybe some noise. but one thing i've learned being a fool willing to release sacred and magickal unobtanial smoke is sometimes things wrong sound better or at least as valid as doing things right.

gotta be some cool way to exploit this for our nefarious purposes of making arcane and profane noise offerings to the fuzz gods

for some reason i'm thinking using it to control a transconductance amp. and i barely know what one of them is. i think i need less coffee and more sleep

[Rob Strand]

i wonder if a burnt junction is part of what makes some really leaky ge q's so they will pretty much work even if flipped 180 degrees.

It's fairly complicated stuff.   

This isn't a very good analogy but it might give you the idea:  If you imagine electrons being marbles and you roll the marbles down a hill but the hill has a bump at the bottom.  Some marbles will make it over the bump and some don't.   The bump for germanium is lower than silicon so more make it across.  Or the other way to look at it is the hill doesn't need to be as steep to get the same number across; the steepness of the hill is like voltage.   The material (Germanium vs Silicon) has a kind of barrier built in.

There's a trick where you put a small DC current through a silicon diode so the DC voltage across it is say 0.3V.   Then you only need to drive is a little hard to conduct to its 0.6V.  The difference is 0.3V which makes it look sort of like a germanium in the forward direction.  The problem with that set-up the silicon diode has a low impedance whereas the germanium diode is high impedance.

I've pondered of the germanium thing in the past and I keep going around in circles with the same old issues.

[PRR]

> some really leaky ge q's so they will pretty much work even if flipped 180 degrees.

If you don't have fantastic control of your process, there's little difference between E and C. Many were made by pulling a crystal while doping the vat rich/lean/rich/lean... E and C were two slices of the same dope. Eye of Newt, 1% Boron, water wash.... On a LOT of early Germanium the emitter breakdown was 30V (similar to Collector) and the collection efficiency was low, both ways.

And yes, this was touted as a "feature". You could buy Ge parts claimed "symmetrical". This was even useful for "choppers" to convert super-low DC to AC for easier amplification.

Now when you get finer control of the process, you see you have a trade-off between breakdown and collection efficiency. You can get a higher hFE if you accept a low emitter breakdown. Discrete parts tend to 7V as a happy zone (opposed to 30V-100V for collector breakdown). Super-Beta parts (usually embedded in a chip) accept way-low breakdown in return for absurd high hFE.