Question about Germanium Diodes

[italianguy63]

I was digging at my surplus house today, and scored some very cool germanium diodes.  I just picked up whatever "looked cool" because I did not have a datasheet with me.

I got some 1N48, 1N52, 1N64A, and 1N58A's (these look WAY cool/BIG glass things).  This is all NOS 1950's-1960's stuff....

I also wrote down part numbers on others I saw:  1N46, 1N54, 1N67, and 1N68...

When I came back, I looked at the data sheet and was astonded that they all have very similar specs.  Most around .05A FWD, and differing voltages (25V to 115V, 80V typical).

Is this just because of the Germanium material?  i.e. it was unpredictable, and difficult to work with, so the specs/variances make them similar?!

And, I guess, are there particular numbers that are "desirable" so I can focus my effort?  I did not even wander to NTE#'s.

[midwayfair]

Datasheets are tough to understand for diodes. Most of the time they give the forward voltage under some specific characteristic, and things might be different between manufacturers. The best thing to do is just measure a couple of each and take that as a ballpark figure to compare against other things you've measured on your multimeter.

[Seljer]

In what you're going to use them for in a typical guitar pedal (i.e. you're not using them as detectors or mixers in radio frequency stuff or in a switching power supply) your going to have to judge its suitability by ear.

Other than forward voltage and series resistance, if theres a current-voltage characteristic given that'll tell you have smooth the edge is.

[italianguy63]

Sounds like:

1.  Sounds good
2.  Looks cool

Basic decision making factors...

[Seljer]

Sounds like:

1.  Sounds good
2.  Looks cool

Basic decision making factors...

It's distortion, we're not sending a man to the moon here 

[vigilante397]

Sounds like:

1.  Sounds good
2.  Looks cool

Basic decision making factors...

It's distortion, we're not sending a man to the moon here 

Very well put, I wholeheartedly agree.

[PRR]

Small Germanium diodes may differ in breakdown voltage (10V-120V), turn-off time, reverse leakage at room temp and at high temps, a few other obscure things.

For clipping diodes, none of these matter. Worst-case you might have 9V reverse; in all typical cases you have another diode the other way so it only ever sees 0.2V-3.4V reverse. Leakage should not matter at typical clipper impedance, is typically better than the spec on the sheet. Turn-off on all these parts is much faster than audio (it is VHF technology).

It is very likely that the same diodes were sold under many part-numbers. A "better" diode can replace a "lesser" spec: a 60V part in a 25V job. If the process was turning out a lot of 60V parts, they would try to sell them at the higher 60V price, but if orders were piling up for 25V parts, they might mark them for the low spec and let them go at the lower price. (Pentium 90 *was* a Pentium 100 but sold for less, marked "90" to preserve the price on the 100 speed parts.)

[teemuk]

When I came back, I looked at the data sheet and was astonded that they all have very similar specs.

Well...they are plain germanium diodes so what did you expect?

But if you look at it carefully, you might actually find that some of them have clear differences in characteristic curves, especially in curvature and "knee". Despite that, the average forward voltage may still end up being in the same region... if we don't care of few millivolts. But when the application is clipping then we probably do care...




e.g. 1N34A and AA112 are both germanium diodes with vF in the same ballbark but "knee" wise 1N34A is closer to plain silicon diode than AA112.  ...Except for forward voltage, of course, which is really no other issue than minding what voltage affects over the diode. e.g. If Si clips at 0.6V and Ge at 0.3V then to get Si clip at similar threshold just double the amplitude affecting over the Si diode.

After that you're basically left with just the knee characteristics.

How audible those details in characteristic curves REALLY are, and are those characteristics that could be compensated with plain series resistance, is probably open for debate. Personally, given proper "normalisation" of input amplitudes and compensation for diode current / series resistance I'm deaf enough to NOT bother to hunt down for esoteric diodes with dubious thermal stabbility.