John Curl's Blowtorch preamplifier

[PMA]

Good question, Rudi.

The equivalent noise resistance is 10 ohm at 25 deg Celsius.

 

[Nelson Pass]

Out of curiousity, have you ever measured the noise
performance of one of your hunky power mosfets ?
At a reasonable current and (for the moment) disregarding
the input capacitance.

I have used IRF610 and 9610 types as input devices for power
amps and as the only devices in some line stage circuits, and
while nobody complained, they clearly are way noisy compared to
good JFETs or Bipolars.

If noise is a serious consideration, I don't think I would even look
there.

 

[john curl]

Actually, I stopped at 0.4nV/rt.Hz long ago, because it is low enough and I did THAT in 1973. I, too, can do 0.1 nV/rt. Hz with paralleled bipolars or fets, but it is too expensive for anything practical.
Glad to know that you have gotten off your duff, and done something really quiet for a change.
1 Nanovolt/rt. Hz is good enough for some things, but not really good enough for MC or ribbon microphone input.
On second thought, we could parallel 100 of your AD797's and get 0.1 nV/rt.Hz Is that how you do it? Remember, audio frequencies are important too, not just RF.

 

[john curl]

Well, KBK, my input good enough for you?

 

[scott wurcer]

Actually, I stopped at 0.4nV/rt.Hz long ago, because it is low enough and I did THAT in 1973. I, too, can do 0.1 nV/rt. Hz with paralleled bipolars or fets, but it is too expensive for anything practical.
Glad to know that you have gotten off your duff, and done something really quiet for a change.
1 Nanovolt/rt. Hz is good enough for some things, but not really good enough for MC or ribbon microphone input.
On second thought, we could parallel 100 of your AD797's and get 0.1 nV/rt.Hz Is that how you do it? Remember, audio frequencies are important too, not just RF.

You are funny John and my skin has grown thick with age. A couple of MIT guys wanted us to do a transformerless ribbon mic amp. I think we used those obsolete RHOM devices to build a prototype but our transistors did not like to operate near saturation so the integrated version did not work. Yes, you can parallel devices to get lower noise this is NOT profound.

 

[john curl]

And you don't parallel devices? You noticed that problem working near saturation? You should try the Hitachi transistors that are just as quiet and twice the beta. Impossible, near saturation. Do you know what the problem is? My patented design is quieter and lower open loop distortion than any that you use, but it doesn't like Rohm or Hitachi devices. Please tell me why, as you work in the industry.

 

[scott wurcer]

And you don't parallel devices? You noticed that problem working near saturation? You should try the Hitachi transistors that are just as quiet and twice the beta. Impossible, near saturation. Do you know what the problem is? My patented design is quieter and lower open loop distortion than any that you use, but it doesn't like Rohm or Hitachi devices. Please tell me why, as you work in the industry.

It's known as quasi-saturation the Mextram models cover it, but you don't use SPICE.

 

[john curl]

What is the physical reason for this?

 

[Bob Cordell]

And you don't parallel devices? You noticed that problem working near saturation? You should try the Hitachi transistors that are just as quiet and twice the beta. Impossible, near saturation. Do you know what the problem is? My patented design is quieter and lower open loop distortion than any that you use, but it doesn't like Rohm or Hitachi devices. Please tell me why, as you work in the industry.

Hi John,

What patented design? Patent number, please?

Thanks,
Bob

 

[gerhard]

We all are misinformed, except one. Guess who.

_THE_ Guess Who ???

 

[Bob Cordell]

What is the physical reason for this?

Hi John,

I believe that quasi saturation has to do with a witches brew of collector resistance, Early effect, and beta degradation at high current. It causes the transition from the saturation region to the normal region to be much more gradual, rather than fairly sharp.

Bob

 

[scott wurcer]

What is the physical reason for this?

You have to pick apart the components of the collector series resistance. The device saturates internally not at the terminals. There is also the problem of charge injected into the base region, it behaves badly. It gets complicated.

 

[gerhard]

[
@John: Your Siemens EL34's are most probably Philips
(Siemens never had a fab, to my knowledge).
- Klaus [/B]

Since you are in Berlin, you really should know better:

http://www.siemens-stadt.de/siemroe0.htm

3 years ago, I still saw guideposts there in Siemensstadt.
I think they still make microwave TWTs

regards, Gerhard

 

[Cal Weldon]

gerhard,

Habaneros?

 

[Steve Eddy]

What patented design? Patent number, please?

4,035,737

An externally hosted image should be here but it was not working when we last tested it.

se

 

[scott wurcer]

Hi John,

What patented design? Patent number, please?

Thanks,
Bob

US4035737 But Schmook's multi-tanh input stage has zero distortion for small inputs and is extendible to complimentary operation.

 

[john curl]

Ho, de, do!

 

[Terry Demol]

Actually, I stopped at 0.4nV/rt.Hz long ago, because it is low enough and I did THAT in 1973. I, too, can do 0.1 nV/rt. Hz with paralleled bipolars or fets, but it is too expensive for anything practical.
Glad to know that you have gotten off your duff, and done something really quiet for a change.
1 Nanovolt/rt. Hz is good enough for some things, but not really good enough for MC or ribbon microphone input.
On second thought, we could parallel 100 of your AD797's and get 0.1 nV/rt.Hz Is that how you do it? Remember, audio frequencies are important too, not just RF.

You are funny John and my skin has grown thick with age. A couple of MIT guys wanted us to do a transformerless ribbon mic amp. I think we used those obsolete RHOM devices to build a prototype but our transistors did not like to operate near saturation so the integrated version did not work. Yes, you can parallel devices to get lower noise this is NOT profound.

Interfet NJ3600L probably not a bad place to start (<0.3nV). Not too
many of these would get to 0.1nV, but the IP C will be horrendous.

Shouldn't be an issue for the source Z of a ribbon mic element
though.

Speaking of interfet, don't they do sk170 / sj74 replacements?

T

 

[gerhard]

I'm working at .25nV where have you been? There's no commercial products there just lab stuff. I asked a while back,"How would you make a useful .1nV amplifier". I still don't have an answer.

I've recently seen a paper of a guy who read out SQUIDs with a
single Infineon SiGe BFP650 at cryo temp. That was somewhere in
this realm IIRC. 1/f corner was 1 KHz or so, not bad for an ft of +-60
GHz for the process. The transistor costs cents, but is probably too
exhilarated for the audience here.

But then, the term "useful" suggests that there should be a use.
If the equiv. noise resistance is much lower than the source
resistance: Where is the advantage? Even moving coil cartridges
have an Ohm, and so do the ring mixers I use for carrier phase
noise measurements. OK, a superconducting loop will do.

We can parallel a dozen Interfet IF3602 at room temp, or even
more NXP BF862s or MAT03s (that will have less input capacitance).
The principle is well known but the MAT03 prices hurt.
What would such an amp be good for after a moving coil pickup?
Better reproduction of the vinyl grain noise?

regards, Gerhard

(who just ordered 250 BF862 from DigiKey)

 

[scott wurcer]

Ho, de, do!

Have you heard some of the Cab Calloway transfers from the original stampers with the reverse cartridges? Those RCA ribbons what a sound!! Hey De Hey!