Martin,

Very interesting.

When designing these circuits that need extremely low noise at long 
periods I believe that there are a couple of additional factors that 
need to be considered beyond the device noise specs.  These are 
factors which are usually thought of as DC issues, but which at low 
enough frequencies can look like noise--and sometimes lots of 
noise.  I'm thinking in particular of supply voltage sensitivity and 
temperature effects, since both the temperature and supply voltages 
will be varying slightly with time.

A discrete transistor pair will not be perfectly matched in 
base-emitter characteristics and they will necessarily be at slightly 
different temperatures.  Then, their temperatures can vary over times 
of seconds to minutes, with the same being true for supply voltage 
variations.  When you look at the numbers, a few thousandths of a 
degree mismatch can cause voltage fluctuations which are as large or 
possibly much larger than the noise specs.  The same effects occur in 
IC's though on a single chip the matching and temperature differences 
can be made better/smaller.  But even with IC's you need to look 
carefully at the specs for their response to temperature and supply 
voltage fluctuations.  Sometimes it's a surprisingly large effect 
when compared with their noise specs.

It's tough to get a really quiet amplifier at very low frequencies.

Regards,
Brett

At 08:46 AM 8/13/2010, you wrote:
>Hi All,
>
>
>This note concerns the flicker noise in pre-amplifiers which may be 
>of interest to those who have a coil resistance that is greater than 
>5 K ohms. Chopper amplifiers such as the LMV2011 etc.are said to be 
>free of flicker noise but the shot noise voltage is large at 35nv / 
>root Hz. For a 5k ohm source resistance the noise figure (N.F.) is 
>about 12 db and for a band width of 0.05 to 5 Hz gives a total noise 
>of about 100 nv.
>The LT1007 op amp has a flicker noise voltage level of 10-17 volts 
>squared / f and a flicker noise current of 2*10-23 amps squared / f 
>, this gives an optimum source resistance of about 700 ohms and a 
>N.F. of 15.5 db at 0.05 Hz, though at 5 K ohms this rises to 21 db. 
>This particular IC has two pairs of transistor and 6 noise current 
>sources which give it the high current noise value and are unchangeable.
>The Hitachi transistor 2SC2545 noise performance is very low indeed 
>with regard to flicker noise and the data is given as N.F. contour 
>on a log of collector current versus log of source impedance graph 
>at 1Kc/s and 10 c/s.  It has been possible to extract the principle 
>flicker coefficients from this for a simple model. The flicker noise 
>voltage remains constant but the flicker noise current depends on 
>the collector current squared and at a collector current of 0.055 
>ma, gives an optimum source impedance equal to 5K ohms. A pair of 
>these transistors  would give a spot noise at 0.05 Hz equal to the 
>chopper amplifier but the total noise for a 0.05 to 5 c/s bandwidth 
>of about 1/6.  This bandwidth is calculated for a simple RC high and 
>low pass filter.
>It may appear a small improvement but every bit may help.
>
>Martin


__________________________________________________________

Public Seismic Network Mailing List (PSNLIST)

To leave this list email PSNLIST-REQUEST@.............. with 
the body of the message (first line only): unsubscribe
See http://www.seismicnet.com/maillist.html for more information.