JFET preamp, battling 1/f noise

[[email protected]]

I have a JFET preamp which appears to be measuring in the ballpark of 6
nV/rtHz at higher frequencies, but with a 1/f noise corner around 500
Hz, which seems a bit high. The JFET is a 2N4393. Can anybody offer
some general pointers for avoiding 1/f noise in JFET circuits?

 

[[email protected]]

I should have mentioned, my circuit looks a lot like the one referenced
here:

http://www.jensen-transformers.com/as/as098.pdf

 

[Phil Allison]

<[email protected]>
wrote in message


** Beware - Google Groper on the loose

I have a JFET preamp which appears to be measuring in the ballpark of 6
nV/rtHz at higher frequencies, but with a 1/f noise corner around 500
Hz, which seems a bit high. The JFET is a 2N4393. Can anybody offer
some general pointers for avoiding 1/f noise in JFET circuits?

** Yeah.

Use a low noise fet instead of a switching type.

2N5457, 2N5459, 2SK30A etc .....




........ Phil

 

[[email protected]]

Thanks for the tips. Would you mind explaining "Google Groper?"

 

[Phil Allison]

<[email protected]

wrote in message

Thanks for the tips.

** No problem.

Would you mind explaining "Google Groper?"

** Been waiting for someone to ask me that.

It's no spelling mistake.

Think of someone " groping about in the dark " and then the other usual
meaning of "groping".

Seems a very apt title to give the hordes of moronic, sociopathic,
anonymous, trolling Googlegroups posters that have been recently inflicted
on usenet and are now *hell bent* on looting and destroying what little of
any value was left after the Hotmail onslaught.

BTW

If you desire even the tiniest bit of respect - learn to post correctly for
usenet.

Click on "options" and then "reply, quote the previous text and poster's
name.




......... Phil

 

[Tim Wescott]

<[email protected]




** No problem.






** Been waiting for someone to ask me that.

It's no spelling mistake.

Think of someone " groping about in the dark " and then the other usual
meaning of "groping".

Seems a very apt title to give the hordes of moronic, sociopathic,
anonymous, trolling Googlegroups posters that have been recently inflicted
on usenet and are now *hell bent* on looting and destroying what little of
any value was left after the Hotmail onslaught.

BTW

If you desire even the tiniest bit of respect - learn to post correctly for
usenet.

Click on "options" and then "reply, quote the previous text and poster's
name.




........ Phil

What he _really_ means is that he doesn't want anyone new here, so that
USENET will die a slow death from starvation, and moderated, for profit
'forums' will be the only way to find information. He's just being
cranky with Google Groups because it's bringing in a lot of newbies and
because the broken Google interface gets in the way of making an
intelligent looking reply.

But do learn what a real USENET reply should look like and how to coerce
Google into doing it for you -- read the link in my tagline for more
information.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google? See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

 

[John Larkin]

What he _really_ means is that he doesn't want anyone new here, so that
USENET will die a slow death from starvation, and moderated, for profit
'forums' will be the only way to find information. He's just being
cranky with Google Groups because it's bringing in a lot of newbies and
because the broken Google interface gets in the way of making an
intelligent looking reply.

I think what he really means is that he hates everyone, and actually
welcomes an enlarged population that he can hate.

John

 

[Phil Allison]

"Tim Wescott"

What he _really_ means is that he doesn't want anyone new here, so that
USENET will die a slow death from starvation, and moderated, for profit
'forums' will be the only way to find information.

** Where did you get THAT asinine idea from - Tim ?

Same place the sun permanently don't shine ?

BTW

I do not contribute to any on line "forums" - they are all crap AFAIK.

He's just being cranky with Google Groups because it's bringing in a lot
of newbies and because the broken Google interface gets in the way of
making an intelligent looking reply.

** It does a LOT worse than that:

It brings in .. " hordes of moronic, sociopathic,
anonymous, trolling Googlegroups posters that have been recently inflicted
on usenet and are now *hell bent* on looting and destroying what little of
any value was left after the Hotmail onslaught. "

But do learn what a real USENET reply should look like and how to coerce
Google into doing it for you -- read the link in my tagline for more
information.

** You missed the main, BAD issue with GG posts - there is often no clue
as to whom or which post they address.

When a GG post comes along late in a thread, the post appears in a list (
created by OE ) that depends on its time and date of arrival - not its
relevance to the post immediately above it.

This is why it is ESSENTIAL for ANYONE quote the prior post - so readers
know WHO is being addressed.

It is pig arrogant and damn lazy not to do so.




........ Phil

 

[[email protected]]

What he _really_ means is that he doesn't want anyone new here, so that
USENET will die a slow death from starvation, and moderated, for profit
'forums' will be the only way to find information. He's just being
cranky with Google Groups because it's bringing in a lot of newbies and
because the broken Google interface gets in the way of making an
intelligent looking reply.

But do learn what a real USENET reply should look like and how to coerce
Google into doing it for you -- read the link in my tagline for more
information.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google? See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

Thanks for that useful link about posting from Google. I had no idea,
as I assumed that all newsreaders organized threads in pretty much the
same way as Google Groups.

On an amusing note, fear of newbies ruining the Internet is as old as
the Internet itself, if not older. I remember a time when many Internet
users seriously believed that any commercially motivated use of the
'net was unacceptable. There were people who would flame anybody with
"aol.com" in their e-mail address.

 

[Phil Allison]

Thanks for that useful link about posting from Google. I had no idea,
as I assumed that all newsreaders organized threads in pretty much the
same way as Google Groups.

** Get a *real* news reader and news account and join usenet.

See just how wrong you are.

On an amusing note, fear of newbies ruining the Internet is as old as
the Internet itself, if not older.

** This is NOT the internet - pal.

This is usenet, a non-commercial, open public forum paid for by its users.

I remember a time when many Internet
users seriously believed that any commercially motivated use of the
'net was unacceptable.

** While some commercial operators are innocuous enough on usenet - most
soon become a real MENACE to the openness and original purpose of most
news-groups. Particularly in the world of audio.

There were people who would flame anybody with
"aol.com" in their e-mail address.

** Same as you would *expect* folk to heckle someone using a public park or
other public speaking arena for commercial gain. It plain stinks.





......... Phil

 

[Uwe Bonnes]

<[email protected]>
wrote in message


** Beware - Google Groper on the loose

** Yeah.

Use a low noise fet instead of a switching type.

2N5457, 2N5459, 2SK30A etc .....

The BF862 is also a good choice...

 

[Phil Allison]

"Uwe Bonnes"

Phil Allison





The BF862 is also a good choice...

** I looked it up:

http://www.semiconductors.philips.com/acrobat_download/datasheets/BF862_3.pdf


It's an RF type in a tiny SMD pack.

Likely give the OP a coronary.

Are you sure it has a low NF at audio frequencies?




......... Phil

 

[Winfield Hill]

Phil Allison wrote...

Uwe Bonnes wrote...

** I looked it up:
http://www.semiconductors.philips.com/acrobat_download/datasheets/BF862_3.pdf
It's an RF type in a tiny SMD pack.
Likely give the OP a coronary.
Are you sure it has a low NF at audio frequencies?

I'll wager is does quite well. The combination of 0.8nV with
only 10pF is rather nice, much better than a 2N5457 or 2sk30.

 

[Phil Allison]

"Winfield Hill"

I'll wager is does quite well. The combination of 0.8nV with
only 10pF is rather nice, much better than a 2N5457 or 2sk30.

** Hypothetical *wagers* made by *wankers* = voodoo science.


The OP simply needs a low noise, audio J-fet he can install in lieu of a
TO92 switching type in a musical instrument pre-amp.

The commonly available, TO92 J-fets I listed provide a 1 dB, or better,
noise figure in this kind of app.





....... Phil

 

[[email protected]]

Phil Allison wrote...

I'll wager is does quite well. The combination of 0.8nV with
only 10pF is rather nice, much better than a 2N5457 or 2sk30.

Summarizing what I think that I have learned so far: The JFET itself is
likely to be the only palpable source of 1/f noise, assuming a
reasonably designed circuit. For newbies like me (mere physics PhD),
1/f noise is something that's just there empirically, for which a
deeper understanding than a simple parametric model is of little
practical use.

I found one device, LSK170, with an impressive 1/f corner around 10 Hz.
Notably, it is an IC. It appears that JFET op amp chips also have lower
1/f corners than mainstream discrete JFETs. Any ideas as to the reason?

 

[Phil Allison]

<[email protected]...

Summarizing what I think that I have learned so far: The JFET itself is
likely to be the only palpable source of 1/f noise, assuming a
reasonably designed circuit. For newbies like me (mere physics PhD),
1/f noise is something that's just there empirically, for which a
deeper understanding than a simple parametric model is of little
practical use.

I found one device, LSK170, with an impressive 1/f corner around 10 Hz.
Notably, it is an IC.

** Notably, it is *not* an IC.

Beware of press releases with purple prose.

It appears that JFET op amp chips also have lower
1/f corners than mainstream discrete JFETs. Any ideas as to the reason?

** Their noise level at high frequencies is higher.


BTW

Why are you so concerned with this nonsense in an audio band pre-amp ?

http://www.jensen-transformers.com/as/as098.pdf

Since 95 % of the bandwidth is *above 1kHz* , that is where the audible
noise is.






......... Phil

 

[[email protected]]

<[email protected]...



** Notably, it is *not* an IC.

Beware of press releases with purple prose.




** Their noise level at high frequencies is higher.

That's a good point. If the 1/f corner is where the 1/f contribution
equals the white noise contribution, then the corner could be lowered
by raising the white noise, rather than by lowering the 1/f. And as you
say below, it's the broadband stuff that contributes to the audible
noise.

BTW

Why are you so concerned with this nonsense in an audio band pre-amp ?

http://www.jensen-transformers.com/as/as098.pdf

Since 95 % of the bandwidth is *above 1kHz* , that is where the audible
noise is.






........ Phil

That's certainly a fair criticism. I am willing to consider the droll
little audio problem to be solved, but I would not mind understanding
this stuff in a bit more generality, just for the sake of my own
education. I do some things with electronics that are not audio
related, but which involve signals in the audio band.

 

[Phil Allison]

That's a good point. If the 1/f corner is where the 1/f contribution
equals the white noise contribution, then the corner could be lowered
by raising the white noise, rather than by lowering the 1/f. And as you
say below, it's the broadband stuff that contributes to the audible
noise.


That's certainly a fair criticism. I am willing to consider the droll
little audio problem to be solved, but I would not mind understanding
this stuff in a bit more generality, just for the sake of my own
education. I do some things with electronics that are not audio
related, but which involve signals in the audio band.

** Do a Google on 1/f noise.

Plenty there to be read.

Seems to be some tie in with Chaos theory and fractals .....


BTW

Nice to see you using GG correctly now.



........ Phil

 

[Winfield Hill]

[email protected] wrote...

Summarizing what I think that I have learned so far: The JFET itself
is likely to be the only palpable source of 1/f noise, assuming a
reasonably designed circuit. For newbies like me (mere physics PhD),
1/f noise is something that's just there empirically, for which a
deeper understanding than a simple parametric model is of little
practical use.

You're correct the JFET is the primary source of 1/f noise, but
you're off base in worrying about it as a significant spectral-
noise-density issue. Consider analyzing the integrated noise
in third-octave bands, and factor in the Fletcher-Munson curves.

I found one device, LSK170, with an impressive 1/f corner around 10 Hz.
Notably, it is an IC. It appears that JFET op amp chips also have lower
1/f corners than mainstream discrete JFETs. Any ideas as to the reason?

The LSK170 is a substitute for Toshiba's 2sk170. Toshiba provides
lots of information about the 2sk170, but Linear Integrated Systems
provides little detail about their LSK170 replacement. For example,
look at Toshiba's plot of voltage-noise vs drain current: the curve
goes through 0.9nV at 2mA, which is the only point provided by Linear
Systems. After several glances over the years, I haven't seen any
advantages for the LSK170 part. BTW, in my opinion, the 2sk170 is
an excellent JFET for applications having 20 to 40pF of sensor-node
capacitance (it's the old en-Cin-w noise issue). If your capacitance
is significantly lower than 15pF another JFET may be a better choice.

 

[Pooh Bear]

<[email protected]...


** Notably, it is *not* an IC.

Beware of press releases with purple prose.


** Their noise level at high frequencies is higher.

BTW

Why are you so concerned with this nonsense in an audio band pre-amp ?

http://www.jensen-transformers.com/as/as098.pdf

Since 95 % of the bandwidth is *above 1kHz* , that is where the audible
noise is.

And a device with 1nV / sqrt Hz is essentially noiseless compared to a guitar
pickup !

Graham