Q: BF96× mos tetrodes for instrument input

[tv]

mos tetrodes - has anybody tried that? low cost with low noise figure.

reason: input with g1 could be designed exactly like jfet or tube stage (68k + 1mohm) while biasing could be done via g2, independently of g1, maybe even with some additional dc servo circuit if directly coupled with next amplifyign stage - and to minimise component differences if used for multiple inputs. What puzzles me is g2 is often used for agc - wouldn't biasing alter the gain as well if used as grounded source. Wouldn't matter as much if used as folower I guess.

probably could work with single +15V rail or with +/-8V if additionally regulated/dropped from +/- 15v.

Am I right?
(If coupled to next stage via cap, I could simply use bs170 instead and simplify a lot, could use +/- 15v for the fet)

 

[PRR]

I didn't know they still made silicon tetrodes.

> low noise figure

At audio?

Traditional MOSFETs have 1/f noise starting as low as 10KHz, getting atrocious at audio bass.

> g2 is often used for agc

Yeah, and I bet you find it mostly makes sense for tuned loads. AGC happens because current drops. That hardly matters in RF/IF tanks, bothers audio transformers, and R-C coupled audio stages starve to death before you get much gain reduction.

I did recently look at a tetrode-like audio AGC idea, and decided the bias-shift was just a little too much to handle.

> wouldn't biasing alter the gain as well

Biasing always alters gain. Maybe a lot, maybe very little; depends on details and how much of the naked device's potential gain you are willing to give up.

 

[tv]

yeah I'm aware these are rf devices, but having g2 biasing option could let me use dc coupled input - no caps, just like usual tube inputs, with single supply. PLus they have onboard zeners already. Main reason was to use g2 with some sort of servo to be able to couple 2 stages (fet and opamp) without capacitors and make a sort of mu-stage input so it would have a little *bigger* sound.

if I was using ordinary bs170 I'd need to provide a small VCC- supply or calculate circuit so it would work with +/- 15V - if I want to avoid capacitor on input. but I'd need to add zeners anyway. Ciis is 25-40 pf for bs170 so it's in the same ballpark as usual tube instrument inputs.

using jfet would be simpler, but I'd like to have mosfets to drive them a bit if I want.

using ca3130 (mosfet opamp) would be simplest, but it's noise is like tl072 and it needs +/- 7V or so (more chips/regulators again). My budget is low.

besides - I have some bs170 already ... but those bf961 look nice too.


hmmm ... since there's already some weird ideas here - what if I'd make a white follower input stage with bs170 - did anybody use such circuit with fets yet - how does that sound on an instrument? I mean not for hi-fi etc?

and yet another: would a mu-stage work well as a gainstage in a solidstate pultec version? opinions?

 

[bcarso]

Borbely has shown some White follower JFET circuits---there's an article in Audio Electronics a while back called something like "FETs: The New Frontier."

The main reason for tetrode MOSFETs is to reduce Miller C in one compact stage, allow higher bandwidth, and obviate unilateralization. But although the reduction in surface and interface state densities due to improved MOS processing has led to astonishing reductions in 1/f noise, vide National Semi's CMOS opamps, JFETs, unless they are really lousy ones, are gong to beat MOS for audio noise. So stack some JFETs if you must---especially since discrete tetrode MOSFETs are likely made the same way they were in the '70's, and have horrendous noise. Maybe somebody is making some better ones, driven by applications where low frequency noise is important, like for low phase noise oscillators, but I believe JFET is still the way to go.

The use for gain control is problematic---for audio the control bandwidth is too close to the signal bandwidth so, without at least a matched balanced topology you will get nasty control feedthrough. At RF this is not an issue.

 

[tv]

I was eyeing bf961 which I can get *affordable*. Agc aspect - I don't want no agc on input, but it would be convenient to use g2 for biasing and offset cancelation etc. - if possible that is.

Using a jfet config looks like least hassle. I've seen the litle bs170 mosfet in some input stages too. maybe I better put them later in chain to drive them a bit if needed. I'd like to make a *tiny* inductor-less pultec some day. with true-zero cpu load.

btw. mu follower is essentialy a (hidden?) simulated inductor/gyrator? that's why it is said it sounds bigger than a ccs loaded gain stage (srpp???)? I read some tube circuit explanations lately but never occured to me before that a mu stage is a gyrator ... it isn't said so anywhere (everything concentrated on pushpull conditions) but it's essentially the same thing as this:
http://www.epanorama.net/documents/telecom/gyrator.html

live and learn ...

yeah the wcf is in fig. 16A here:
http://www.borbelyaudio.com/adobe/ae699bor.pdf

I wonder how would wcf with bs170 mos perform. Anybody know of a schemo with mosfet? I need some inspiration for biasing the lower mosfet - I guess getting bias voltage with three Si diode drops would work well?