AW: Re: Germanium Diode Filters?

[Paolo Predonzani]

> Yes, it's surprising. Even Marshall (who have a certain reputation ...)
> chose LEDs for their tube-less preamp products.
> Anyway, I am still wondering why it is like this and not the other way
> round. Germaniums *do* have a "smoother" rounded threshold,
> don't they? So a rounded threshold can't be the whole story ...
> 
Signal filtering applied pre and post distortion affect the sound too.
If asymmetrical clipping in needed two diodes of different type (e.g.
LED + germanium) can be used in the feedback of the op-amp.
That is:
LED inverting_input ->|- output
germanium inverting_input -|<- output

Yes, smoothness is not the whole story. There has been a lot of discussion
on Hi-Fi magazines about tube distortion but I've never seen a good
explanation. Second and third harmonics issues are particulary confusing.
I think the nature of the components alone (vacuum or solid-state) cannot
determine the sound; the circuit schematic is much more important.
For example: does open-loop and closed-loop designs affect the sound?
Is it possible that harmonics are in different phase relationship with
the fundamental tone depending on the open/closed loop design? 
I know this can sound meaningless because our perception of sound is based
on spectral analysis and not on the instantaneous value of air pressure.
But our ears are non-linear systems! So the `shape' of the signal can be
important too and its importance increases as the input signal gets more
complex.

> BTW, the weirdest thing I have seen so far was a combination
> of si and ge diodes, to form a nonlinear voltage devider, like
> this (Spice synthax):
> 
> D1   input      output   silicon_type_diode
> D2   output   input      silicon_type_diode
> D3   output   0             germanium_type_diode
> D4   0             output   germanium_type_diode
> 
> May also have been the other way round (ge's in signal path,
> si's to gnd). Anyway, a quite complex transfer curve.
> Sound? Not bad, not outstanding either.

I don't know. The circuit should be tested. Probably they wanted a
non-monotonic attenuation. Something like:
^ gain
|
|  __
|_/  \
|     \________
+---------------------> input

The transfer funcion is surely monotonic.

Other circuits (called hyper-fuzz, I think) have non-monotonic transfer
funcion. I use one of them as a waveshaper.
 
-- 
+-------------------+----------------------------------+
| Paolo Predonzani  |  email: predo at dist.dist.unige.it |
+-------------------+----------------------------------+