Transformer thermal noise / *small* signal

[Fred Bartoli]

Hello,
For a wide band (0.1Hz-1MHz) low impedance small signal source I need to
design a room temperature ultra low noise high gain preamp.

The specs are about 50nV rms (yes!) integrated over the 1MHz band, which is
50pV/root(Hz) noise density.

Some system specificities allow me to build the preamp like this :

.-----------------.
| |
| 200pV/root(Hz) |
IN | |
--------+-----------| G=10000 |----. .----------------.
| | | | | | |
| | | | | | |
| | | '---| LP | |
| '-----------------' | | |
| | | | OUT
--- |----------| SUM |-----
--- .-----------------. | | |
| | | .---| | |
| 1:4 | 200pV/root(Hz) | | | HP | |
| | | | | | |
'---. ,-----| |----' | | |
)|( | G=10000 | '----------------'
)|( | |
-' '- | |
| '-----------------'
===
GND
(created by AACircuit v1.28 beta 10/06/04 www.tech-chat.de)

having a 200pV/rt(Hz) noise density in the low frequency region and
50pV/rt(Hz) in the high frequency region (expected crossover about 10kHz).

At this level there are lots of noise sources to chase, one being the
transformer thermal noise. Windings noise will be low but core magnetics
thermal noise might be a pb (magnetic domain noise).

Any experience with this phenomena?

One hour googling didn't give anything useful (like orders of magnitude, PSD
shape, ...)

 

[James Meyer]

Hello,
For a wide band (0.1Hz-1MHz) low impedance small signal source I need to
design a room temperature ultra low noise high gain preamp.

The specs are about 50nV rms (yes!) integrated over the 1MHz band, which is
50pV/root(Hz) noise density.

At this level there are lots of noise sources to chase, one being the
transformer thermal noise. Windings noise will be low but core magnetics
thermal noise might be a pb (magnetic domain noise).

Any experience with this phenomena?

Just a little. I'd recommend looking at amorphous, MetGlass, core
material.

Jim

 

[John Larkin]

Hello,
For a wide band (0.1Hz-1MHz) low impedance small signal source I need to
design a room temperature ultra low noise high gain preamp.

The specs are about 50nV rms (yes!) integrated over the 1MHz band, which is
50pV/root(Hz) noise density.

Some system specificities allow me to build the preamp like this :

.-----------------.
| |
| 200pV/root(Hz) |
IN | |
--------+-----------| G=10000 |----. .----------------.
| | | | | | |
| | | | | | |
| | | '---| LP | |
| '-----------------' | | |
| | | | OUT
--- |----------| SUM |-----
--- .-----------------. | | |
| | | .---| | |
| 1:4 | 200pV/root(Hz) | | | HP | |
| | | | | | |
'---. ,-----| |----' | | |
)|( | G=10000 | '----------------'
)|( | |
-' '- | |
| '-----------------'
===
GND
(created by AACircuit v1.28 beta 10/06/04 www.tech-chat.de)

having a 200pV/rt(Hz) noise density in the low frequency region and
50pV/rt(Hz) in the high frequency region (expected crossover about 10kHz).

At this level there are lots of noise sources to chase, one being the
transformer thermal noise. Windings noise will be low but core magnetics
thermal noise might be a pb (magnetic domain noise).

Any experience with this phenomena?

One hour googling didn't give anything useful (like orders of magnitude, PSD
shape, ...)

Just a random thought, but amorphous (Metglas) cores aren't
crystalline so may not have the grain boundaries that make core noise.

Where do you get the 200 pv/rthz parts?


Here's something similar:

http://www.thinksrs.com/downloads/PDFs/Manuals/SR554m.pdf

The PDF doesn't have schematics (those pages are blank) but the old
printed manuals probably do.


Further random thought: a bank of narrowband, tuned step-up matching
networks could drive a lot of separate amps, whose output was
combined... but not this wideband, of course.

John

 

[Jeroen Belleman]

Hello,
For a wide band (0.1Hz-1MHz) low impedance small signal source I need to
design a room temperature ultra low noise high gain preamp.

The specs are about 50nV rms (yes!) integrated over the 1MHz band, which is
50pV/root(Hz) noise density.

In a low noise pre-amp I built some years ago, I used Vacuumschmelze Vitrovac
amporhous alloy cores, both in the feedback and in the input noise matching
transformer. The transformers did not contribute any noise that I could see.

The amplifier was a 300pV/rtHz 20kHz-75MHz device with a gain of 26dB and
50 Ohms input impedance. The input transistors where three Philips BF862 JFETs
in parallel.

Maybe you can draw some inspiration from the description of the following
65pV/rtHz 5Hz-100kHz amplifier:

J. Lepaisant, M. Lam Chok Sing, D. Bloyet
Low-noise preamplifier with input and feedback transformers for low source
resistance sensors
Rev. Sci Instrum. 63(3), March 1992, p2089


Regards,
Jeroen Belleman

 

[Winfield Hill]

Jeroen Belleman wrote...

In a low noise pre-amp I built some years ago, I used Vacuumschmelze Vitrovac
amporhous alloy cores, both in the feedback and in the input noise matching
transformer. The transformers did not contribute any noise that I could see.

The amplifier was a 300pV/rtHz 20kHz-75MHz device with a gain of 26dB and
50 Ohms input impedance. The input transistors where three Philips BF862
JFETs in parallel.

Those are nice parts, not the lowest e_n by any means, 0.8nV at 100kHz,
but pretty damn low for a medium-sized JFET with Ciss only 10pF.

Maybe you can draw some inspiration from the description of the following
65pV/rtHz 5Hz-100kHz amplifier: J. Lepaisant, M. Lam Chok Sing, D. Bloyet
Low-noise preamplifier with input and feedback transformers for low source
resistance sensors, Rev. Sci Instrum. 63(3), March 1992, p2089

A nice reference. It should be noted that these fellows used ordinary
Siemens N48 ferrite pot cores in their transformers. They also used
Toshiba 2sk170 JFETs, with slightly-lower e_n, but 3x higher Ciss = 30pF.