[time-nuts] Measuring power supply noise
Charles P. Steinmetz
charles_steinmetz at lavabit.com
Sat Sep 4 22:19:26 UTC 2010
>HP 3552A Transmission Test Set. Designed for testing telephone
>lines, its basically an AF signal generator, level meter, frequency
>meter, monitor amplifier and line iterface in a portable box.
>Ignoring the generator, the line interface will allow direct
>connection to supplies of up to 75V and the level meter measures
>noise down to -90dBm (600ohm) with low pass filters available.
>* * * Its specified for 40Hz to 60kHz but responds down to 10Hz.
>* * * A quick check of a switching supply gave -42dBm at 450Hz
>off load and -45dBm
> at 1.5KHz at 28mA. A nice general purpose audio tester that is not
> well known. Cost was less than $40.
Brings back fond memories of the HP and other TTSs. For power
supplies used to run precision oscillators, though, one probably
would like something that performs well below 10 Hz and with noise
levels in the tens of nV range rather than the 3552's noise floor of
tens of uV. A nice, quiet preamp will solve the latter problem, but
you'd have to tear into the 3552 to conquer its LF limit.
When one gets serious about quiet regulators for oscillators, one has
to think about what's inside the can. Often, there is an internal
regulator with 10-1000 times the noise of your carefully-designed
regulator (and, in the case of a GPSDO, digital circuitry trying to
put noise on the oscillator supply, as well).
More information about the time-nuts