[time-nuts] Ultra low noise amplifiers.

Dr Bruce Griffiths bruce.griffiths at xtra.co.nz
Fri Mar 30 19:49:37 EDT 2007


The following papers may be of some interest to anyone contemplating 
building a low noise low frequency (dc -100kHz) amplifier.
Such an amplifier would be useful, for example, after a double balanced 
mixer in a conventional phase noise measurement system.
http://arxiv.org/PS_cache/physics/pdf/0503/0503012.pdf
http://arxiv.org/PS_cache/physics/pdf/0512/0512082.pdf

In a correlation measurement system the white noise floor can be lowered 
by averaging, whilst the flicker noise cannot. Thus it is important to 
select amplifiers with low flicker noise for such systems.

Although some of the aspects of the amplifier design are questionable 
(e.g. lower noise current sources are possible), the very low frequency 
noise measurements illustrate the importance of thermal lagging and 
using a light tight enclosure. The seals around the kovar leads in the 
metal dual transistor packages transmit light into the package. All 
transistors are photosensitive unless the junctions are shielded. The 
LED in the current source is also photosensitive, however the resultant 
modulation of the current source output is a common mode signal. Since 
the LED emits light it can also generate photocurrents in all of the 
metal can packaged transistors. It would be interesting to look at the 
effect of placing a light tight shield over the LED. Another approach 
would be to use equivalent dual transistors (SSM2220) in a light tight 
package.

Other useful information is the noise data for various opamps in the 
flicker noise region. This shows that the intrinsic low frequency noise 
of devices can be significantly lower than that specified in their 
datasheets. The datasheet noise specs are usually based on measurements 
by the manufacturer using a setup which isn't optimised to minimise 
thermal fluctuations or air currents.

Since the close in phase noise of a bipolar RF amplifier is primarily 
due to modulation of collector base (and other) capacitances by bias 
voltage fluctuations, and the collector base voltages of the various 
transistors is temperature dependent, it is important to shield the 
amplifier from air currents and thermally lag the amplifier by enclosing 
it in a massive metal enclosure, if one wishes to achieve low close in 
phase noise.

Bruce



More information about the time-nuts mailing list