Lamba Octave?

[JKowalski]

Grimy little frequency doubler would just about describe it.

I set up a quick n dirty version on the breadboard. It definitely gives out octaves for single notes, though obviously they sound a little strange. Quick decay, ring modulator type sounds with multiple notes. Bell like tones.

Overall, it's not really my type of sound so I don't think I will go through with building anything for it. But if you want to experiment with it, go ahead. Sorry to dissapoint those who were waiting for something extraordinary!

The core of it is simple and is shown in the simulator file I posted above a while back. You just need to dump a signal into the lambda diode + resistor with the right positive voltage offset to get a doubling effect (will vary with the base resistor and the transistors used) and retrieve the signal from the top of the resistor.

[Taylor]

Would you say it's any different from the usual rectifier octaves?

[jasperoosthoek]

Great result I am very impressed!

I know from experience how difficult tunnel diodes are to bias in the negative resistance region. How did you manage that? JFETs are of course a bit slower as tunnel diodes don't mind switching in the sub nanosecond time region. From what I've seen you cannot just connect tunnel diodes to any old voltage source. You have to to add parallel resistor so that the IV curve with the resistor doesn't have a negative region anymore. But that was a working TD setup right now I can build my own and see for myself.

My idea for an effect would not to double the sine wave but to first get a dip and then a bump again so you have double and tripple the frequency.

I explained that in another thread. http://www.diystompboxes.com/smfforum/index.php?topic=85989.0

My tunnel diodes should arrive this week and I can try to get some stable IV curves. They are Russian AИ101Дs. I'd have to ask Vladimir how to pronounce that.

[JKowalski]

Great result I am very impressed!

I know from experience how difficult tunnel diodes are to bias in the negative resistance region. How did you manage that? JFETs are of course a bit slower as tunnel diodes don't mind switching in the sub nanosecond time region. From what I've seen you cannot just connect tunnel diodes to any old voltage source. You have to to add parallel resistor so that the IV curve with the resistor doesn't have a negative region anymore. But that was a working TD setup right now I can build my own and see for myself.

The lambda diode and tunnel diode curves are very different. The voltages are much larger scale, enabling larger signal strengths, and more protection from over voltage. The current increases with voltage up to about 1-2 volts, then starts decreasing, reaches zero and stays there for a while until gate-source breakdown occurs, which takes quite a bit of voltage. This means that you can put your buffered signal with a DC offset across the lambda diode without any worries, drawing only around 2 mA peak. Biasing with a voltage source (vertical load line) keeps the lambda diode behaved. Adding the resistance (needed to convert the current into voltage) will tilt the load line slightly but not enough to cross the curve again. For the tunnel diode you'd need a load line that is fairly vertical and does not pass through two points at any time.

[jasperoosthoek]

Exactly that's what I also showed a bit in the other thread. This is a quick and dirty sketch of what I've come up for now:

The lowest differential resistance of the tunnel diodes that I ordered is 14 ohms. So the effective load resistance of the load line has to be smaller. Right now the diodes are loaded by ~10+2 ohms plus the backwards resistance of the tunnel diode in series. I hope the backwards resistance is low enough otherwise I have to put parallel resistors to each diode.
This is IMO the most effective way to keep it stable. They can handle 0.6 volts and that's where the (high current) diodes and 1.5 ohms resistor come in. I used a lm386 to supply the generous amounts of current needed to power the diode stack. (Forgot to draw a a bipolar at the output of the lm386 btw, it will be there when I build it).
The 2 ohms resistor severely lowers the signal strength but it is needed to keep the load line more vertical than the tunnel diode characteristics.

I'm going to test all this first with a scope, doing some IV curves. I need to get this stable first before I start breadboarding it.