Which diodes are best for octave pedal?

[Austin Translation]

What solid state diodes would be the most sensitive for use in an rectifier type octave pedal? Also it seems like a 12ax7 or other small signal tube configured as a diode might work in this application, anybody know if it would? Thanks, Austin

[R.G.]

Clear ones with green stripes.



The real answer is "any diode that has a turn-on voltage that's much, much smaller than the signal level it's being fed.". You can get that with either a big signal by amplifying it, or a smaller diode. But diodes only come in discrete forward drops. The smallest practical ones are germanium; the 1N34 and 1N270 are often used. After that are Shottky (about 0.4-0.5V) then silicon (0.5 - 0.7) and then a whole range of LEDs which run from 1.3V to as much as 4V before they'll break over. Cat's whisker diodes can be made to conduct at very nearly 0V, but it's impractical to use them because they have to be hand fitted and changed often.

So if you limit yourself to not amplifying, the answer is "germanium ones".

The only other option is to do some electronic work. If you make the signal maybe 10X the size of the diode breakover, the diode breakover voltage becomes almost unimportant. This is the technique used in the transformer-based Octavia, which uses an amplifier to run a step up trans former to get a lot bigger signal. Other approaches are to use opamps to construct an "active diode" rectifier which rectifies at millivolts, essentially a perfect diode as far as low conduction threshold. Some pedals use this approach. Another approach is to put some minimal amount of DC through the diode to make it just on the edge of conduction so a smaller signal will push it over. The Fender Blender, Fox Tone Machine and Green Ringer use this approach.

Vacuum triodes do start conducting at nearly zero volts, but their conduction transition is large, slow, and sloppy, and they may not make good signal rectifying diodes, especially the 12AX7.

If I knew what you were trying to do, I could focus more accurately on the application.

[Austin Translation]

Clear ones with green stripes.



The real answer is "any diode that has a turn-on voltage that's much, much smaller than the signal level it's being fed.". You can get that with either a big signal by amplifying it, or a smaller diode. But diodes only come in discrete forward drops. The smallest practical ones are germanium; the 1N34 and 1N270 are often used. After that are Shottky (about 0.4-0.5V) then silicon (0.5 - 0.7) and then a whole range of LEDs which run from 1.3V to as much as 4V before they'll break over. Cat's whisker diodes can be made to conduct at very nearly 0V, but it's impractical to use them because they have to be hand fitted and changed often.

So if you limit yourself to not amplifying, the answer is "germanium ones".

The only other option is to do some electronic work. If you make the signal maybe 10X the size of the diode breakover, the diode breakover voltage becomes almost unimportant. This is the technique used in the transformer-based Octavia, which uses an amplifier to run a step up trans former to get a lot bigger signal. Other approaches are to use opamps to construct an "active diode" rectifier which rectifies at millivolts, essentially a perfect diode as far as low conduction threshold. Some pedals use this approach. Another approach is to put some minimal amount of DC through the diode to make it just on the edge of conduction so a smaller signal will push it over. The Fender Blender, Fox Tone Machine and Green Ringer use this approach.

Vacuum triodes do start conducting at nearly zero volts, but their conduction transition is large, slow, and sloppy, and they may not make good signal rectifying diodes, especially the 12AX7.

If I knew what you were trying to do, I could focus more accurately on the application.

Thank you very much for the info! I am trying to understand better how AM radio, amplifiers and power supplies work as a hobby, I thought it might be fun to build an octave pedal for my guitar to experiment with like a little frequency doubler.

Would a subminiature tube signal diode work in this application?
here is one list of sub miniatures: http://pw1.netcom.com/~wa2ise/radios/penciltubes.html

here is a link to a diystompbox where somebody has already had this same idea: http://www.diystompboxes.com/smfforum/index.php?action=printpage;topic=46719.0

I wonder how it turned out? ~Austin

[Mark Hammer]

Though not particularly critical, it helps to have diodes that are reasonably well-matched such that the two half cycles that are added together provide equal proportions.  Matching the overall amplitude of the two half-cycles can provide a more robust octave, and mismatched halves tend to produce an octave that seems to be standing in the back of the crowd, rather than in front.

That said, the diodes are:
a) not the most important factor in determining level-matching, but merely one of several.
b) frequently matched well enough if coing from the same package or order.  If you're like me and you have a parts drawer with an assortment, you may find discrepancies between two of the same type, like Si type with forward voltages of 504 and 620mv, or Ge type with forward voltages of 237 and 323mv.  These fall within spec but have the sort of differences that would likely ask for adjustments via series resistors.  Matching within 10mv is probably more than sufficient.

I should add that, because series diodes impose some crossover distortion by demanding that the signal be above their forward voltage in order to pass, it can be helpful to use either the lowest forward voltage you can find (i.e., Schottky) or use one of the voltage-biasing methods RG alludes to.  That is, of course, unless you find the crossover distortion and gating aspect to be a desirable component of the sound produced.  Your choice.

[amptramp]

Consider using an ideal rectifier circuit of this type shown here:

http://sound.westhost.com/appnotes/an001.htm

Diodes inside the feedback loop have a forward voltage that is reduced by the feedback gain of the op amp.  But if you are making an octave pedal, consider using a squaring circuit based on the MC1496 analog multiplier.  It avoids the excess harmonics that you get from fullwave rectification.

[Austin Translation]

thanks for the replies

I should mention to avoid confusion that I am not attempting to construct the best sounding octave pedal. This is like a self appointed homework assignment in an effort to solidify a basic understanding of some fundamentals.

So if the best solid state diodes will be germanium type that have matching forward voltage, what part number is made to the highest tolerances? I found a thread here that indicated 1n60 had a close tolerance- http://www.diystompboxes.com/smfforum/index.php?topic=49337.0

Is there any reason not to use subminiature tube type diodes in a project like this? Like noise?

I would like to someday build some opamps from "scratch" for use in effects pedals, maybe something like this:


from http://www.national.com/rap/images/CCC.gif
This could be easily adapted for subminiature twin triodes I imagine. It should work right?

another tube opamp link http://www.diyaudio.com/forums/tubes-valves/65811-tube-op-amps.html

I understand there is no good reason to use tubes for things like this because they probably won't sound any better than silicon in this application and it will be more hassle and more cost, maybe I am nostalgic or something but it just fits my definition of cool.  

[R.G.]

I should mention to avoid confusion that I am not attempting to construct the best sounding octave pedal. This is like a self appointed homework assignment in an effort to solidify a basic understanding of some fundamentals.

Self appointed homework is very good indeed.

So if the best solid state diodes will be germanium type

You really do have to watch that word "best". The word is completely free of any meaning without also saying how you score "best". There have been societies which thought the best women were the ones who weighed the most. Our society has other ideals in this instance. What's the best automobile? Some people say a Ferrari. Others say a Honda Civic. Yet others say neither - the best car is a truck.

So you do have to define what you mean by "best". If it's "smallest clipping threshold", yep, that's going to be germanium. Germanium also has the biggest ratio of rounded "knee" of conduction to either fully off or fully on, so it's the sloppiest in terms of how cleanly it goes from non-conduction to conduction. Shottky has a low threshold and a quite-sharp transition to conduction. Silicon comes in many flavors of conduction, knee, etc.

that have matching forward voltage, what part number is made to the highest tolerances? I found a thread here that seemed to indicate 1n60 had a close tolerance-

The best matched will be hand-matched. All devices will have a manufacturing variation. And you can't, in general trust that old-stock germanium will be at its original datasheet tolerances. What germanium semiconductors there are left in the world may well have been picked over for one use or another. For self- appointed homework, count on hand matching. Even if the first two you test are perfect, the side effect of setting up the test is that YOU'RE now better than you were.

Is there any reason not to use subminiature tube type diodes in a project like this? Like noise?

In general, no.

I would like to someday build an opamp from scratch for use in effects pedals, something like this:
...
I understand there is no good reason to use tubes for things like this because they probably won't sound any better than silicon in this application and it will be more hassle and more cost, maybe I am nostalgic or something but it just fits my definition of cool. 

Nostalgia, or just wanting to is reason enough. Whenever I do stuff that doesn't make sense to other people I say it's being done for my education.  I count on the finished work being the garbage left over from the learning.

[Austin Translation]

Point noted about use of the word "best", thank you for the gentle reminder

There is an electronic parts store not far away that may have something current production that will work, but as they don't offer any advice at all, require a part number to cross reference to there brand.

It just occurred to me that the spring reverb driver in an amp might be a super easy way to add octave effect, remove coil insert rectifier.. Could this work?