[time-nuts] NIST isolation amplifiers

Bruce Griffiths bruce.griffiths at xtra.co.nz
Wed Nov 26 03:59:41 EST 2014

A single 2N2222 or equivalent transistor in a suitable circuit dissipating about 200mW or so can achieve a reverse isolation of 35dB with distortion of around -40dBc (output +13dBm) with a gain of unity, and an output impedance of 50 ohms with a PN floor of around -180dBc/Hz or so.

     On Wednesday, 26 November 2014 9:13 PM, Charles Steinmetz <csteinmetz at yandex.com> wrote:

 Bruce wrote:

>Another issue is that if even one output needs high reverse isolation and
>low crosstalk, then even those outputs that arent so critical will also need
>high reverse isolation and low crosstalk to avoid degrading the crosstalk
>to the critical output.

This brings up the distinction between *isolation* amplifiers and 
*distribution* amplifiers.  Most of us need a dozen or three feeds 
for various test equipment, radios, etc.  These feeds should have 50 
ohm output impedance, moderate isolation (35dB or more), and should 
not noticeably degrade the noise, PN, distortion, or xDEV of the 
source.  That is the job of a distribution amp.

I would generally not use anything like one of the NIST circuits for 
this, but rather some version of a two- or three-transistor Class A 
buffer amplifier.  There are lots of circuits to choose from.  Many 
are transformer (or autoformer) coupled, some are not (the JPL 
circuits come to mind) and can also be used to distribute lower 
frequencies.  You can get build-out the NIST way (buffer amp input 
impedance high so you parallel a bunch of them at the input 
connector), or by using one stage with low output impedance to drive 
a number of output amplifiers in parallel, or by using an amplifier 
with very low output impedance (perhaps a high-current monolithic 
amplifier) to drive a number of 50 ohm build-out resistors, or by 
fanning out with CMOS logic and following each CMOS final buffer with 
a Tee network to generate sine waves.

Then there are the times when you are making measurements of 
oscillators and must absolutely ensure that there is no interaction 
between them.  That is the job of an isolation amp.  Rarely will you 
need more than two or three feeds per oscillator, so what you need 
are several, one-to-three iso amps (one for each oscillator).  Here, 
something like the NIST amplifiers makes sense.

Note that I'm advocating distributing sine waves exclusively, NOT 
square waves or pulse trains.  You will find that it is hard enough 
keeping 1, 5, or 10 MHz from getting into everything in the shop (and 
radio room), without adding the much-increased difficulty of keeping 
all of the harmonics under control.  Also, you would like the 
harmonic content to be rather lower than is often thought because (i) 
even harmonics cause asymmetry, which can cause phase modulation when 
the signal is AC coupled or feeds a comparator-type zero-cross 
detector, and (ii) variations in the phase of harmonics in relation 
to the fundamental cause phase modulation (this is "harmonic 
dispersion," which is caused by temperature changes and other circuit 
variations such as modulation of semiconductor capacitances by low 
frequecies).  NIST published a paper on this (see Walls and 
Ascarrunz, "The Effect of Harmonic Distortion on Phase Errors in 
Frequency Distribution and Synthesis").

Best regards,


time-nuts mailing list -- time-nuts at febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.


More information about the time-nuts mailing list