[time-nuts] Phase noise test set reference articles

Gerhard Hoffmann dk4xp at arcor.de
Tue Mar 29 18:28:18 EDT 2016


Am 29.03.2016 um 16:53 schrieb Charles Steinmetz:
>
> (10)  Phase_detector_with_low_flicker_noise_BARNES_etal_NIST_2011: 
> Describes a DIY double-balanced mixer phase detector using 
> diode-connected 2N2222 transistors.  [Note that only the flicker noise 
> is improved -- the white noise floor is actually significantly higher 
> than with DBMs using diodes.  NB: There are much better transistors 
> than the 2N2222 for this application.]
.. and they are NOT really used as diodes. They are used as switching 
transistors with most current flow C-E, saturation enforced.
That turns faster on than the diode exp-law.
Somewhere they also say that they use 50 Ohm load.

I really wonder where all that ring mixer noise is to come from. The 
diodes are just switches, the transformers
have close to no loss and behave properly in power dividers, and even 
when the diodes are resistors for a moment,
their noise is only half-thermal.

I find it easy to believe that high power mixers produce more noise. In 
that app note by Watkins-Johnson that everybody copies,
there it is clearly shown that they may use resistors to generate bias 
voltages.

If I use two 1:4 Wilkinsons, 4 low power ring mixers and put the outputs 
in series, will it also turn worse? I don't think so.

Are there anywhere musings about the equivalent noise resistance of a 
ring mixer IF output?
If it is really something like 500 Ohms, even a single AD797 may be 
excessively over-optimized
for voltage, and not current noise.

regards, Gerhard


More information about the time-nuts mailing list