[AudioCats]

KGSS has small signal diodes between stage 1 and 2. The standard part is 1n914, a silicon diode.

1n914:
100v peak reverse
Vr=75v,
If=300mA,
Vf max=1v,
Ir(@20v)=25nA

According to Kevin Gilmore's KGSS article, the normal current through the diode is <2mA.

HeadWize KGSS article


I have tried some Germanium diodes.

1n270
100v peak reverse
no Vr figure in the spec sheet
If=40mA,
Vf max=1v
Ir(@50v)=100uA

1n34a
65v peak reverse
Vr=20v
Io=50mA
Vf max =1v
Ir (@50v)=500uA

Impressions:

1n914:
-- more "live", more "reverbs",
-- guitar sounds more metalic, more glare.
-- female voice feels more "sweet and pretty", more "young and gitty"
-- has a more edgy/rushing feel.

1n34a
-- has a "calmer" feel to it, "slower flow".
-- female voice more "mature" (somehow sounds 10~15 years older than the 1n914 voice)
-- kind of like the mic is closer to the singers mouth. Maybe more mids?
-- more "phat"

1n270 sits in between 914 and 34a, but it is not a happy middle ground.... it has a kind of "thick and chewy" feel to it.

I prefer the sound with 1n34a installed, but I can not decide whether to keep it in there. It has high leakage current. 1n270 spec is closer to that of 1n914.
But in normal operation the current flows forward, the leakage spec might not matter?
.

 

[kevin gilmore]

The diodes are to prevent negative voltages from damaging the input fets when the power gets turned off.
there are other ways to do the same thing including the shorting the diodes and adding reverse biased diodes
to ground.

 

[AudioCats]

Thank you Sir.

So in this scenario the LV+ in the 2nd stage is already gone, but the HV- still has some juice left, so it tries to draw current from ground, through the JFETS? Am I understanding this right?

 

[kevin gilmore]

that is what can happen. depends on cap sizes.

 

[AudioCats]

Thanks. The voltage blocked by these diodes is also depending on the cap sizes? if so, which spec is more importatnt, the leakage current or the peak reverse voltage?

1n34a has leakage current spec of 500uA at 50v, that is 0.5mA, it is really not blocking much. Does this mean I might have done damage to the JFETs by trying them?

 

[AudioCats]

Trace of voltage across the diode. I tried powering up/down several times, sometimes the spike happened on powering up as well. The highest I saw was around 6v, I guess the JFET's are still ok.

 

[AudioCats]

Using a variac to turn the amp on and off causes no voltage spike. This allows further experiments.

 

[AudioCats]

*** "sound" with a piece of wire
--- powering up/down using the variac to avoid possibility of voltage spikes ----

the sound with wire is more close to 1n34a than the 1n914. Really "mature" sounding, Norah Jones sounds like she was at least 35, if not 45, in the Come Away With Me album (the album came out in 2003, she was in her early 20's).
It feels that the mic is very close to the singers mouth now, gone is the smoothness in voice, it is now more raw with imperfections. Norah Jones sounds more like a country singer (or maybe that is how her voice really sounds like). There is almost nothing pretty or sweet in any of the female vocals, with the exception being some of Katie Melua's early songs.
All female songs sound (emotionally) darker.
very little effect on male vocal, though.

 

[AudioCats]

*** a piece of wire, with reverse biased diode (like KG has suggested in post #2)
-- The sound is basically the same as above (just a piece of wire, no diode).
-- At times it felt (maybe) a little less open but I am not sure.

I much prefer the sound of wires. Micro details improves quite a bit. I will not go back to the stock configuration.