[time-nuts] Distribution amplifier for 10MHz

[Charles Steinmetz]

Taka Kamiya wrote:

> I am thinking about distribution amplifier of a sort.  Not a kind with 8 or more output but just 2 or 3.
>    *   *   *
> But how do I do this on 10MHz side?  I'd like to have minimal distortion (sine wave) and high isolation.  I acquired a few signal splitter and monolithic amplifiers.  Split first and amplify?  Amplify first and split?  Remember it's only 2 to 3 channels.  Amplifier module has too much gain, so I'll probably have to use attenuators.
>
> Have anyone run into this kind of thing?  Is there an existing design?  Anything anyone would like to share?

I have previously posted some documents you may find helpful:

<http://www.ko4bb.com/manuals/73.134.121.2/Tutorial_Distribution_Amplifier_Architectures_STEINMETZ_rev1.pdf>

<http://www.ko4bb.com/manuals/73.134.121.2/Extron_ADA_6_modifications_for_use_as_10MHz_distribution_amp_STEINMETZ.pdf>

The first one is a general tutorial on sine-wave distribution amplifiers 
that describes the various approaches and provides generalized design 
details.

The second one is a "modification manual" that gives specific 
instructions for modifying one particular family of commercial 75-ohm 
video distribution amplifiers (found inexpensively on eBay) for optimal 
use as a 1MHz to 100MHz sine wave distribution amplifier [with way more 
outputs than you say you need -- but note that you may find your "needs" 
develop and expand if you have more outputs.]

Finally, consider carefully how much isolation you really need.  There 
are two basic kinds of "distribution amplifiers".  The first is what the 
term is generally used for, while the second ("isolation amplifiers") is 
a specialized subset of distribution amplifiers with specific, enhanced 
performance requirements.

Distribution amplifiers are intended to provide high-quality output 
signals to a number of instruments (e.g., a frequency counter and a 
spectrum analyzer) simultaneously.  However, they are not expected to 
have their various outputs compared differentially or 
interferometrically with respect to delay time (phase), harmonic 
content, noise, etc.  Output isolation figures are typically from 40dB 
to 80dB.

Isolation amplifiers must be used when such comparisons are to be made, 
for example sending two copies of a signal down different paths and 
comparing them differentially or interferometrically at the downstream 
ends -- for example, to determine the residual phase noise of devices 
under test).  These techniques are capable of much greater resolution 
and are used to compare signal parameters at levels very close to the 
metrological (or even the physical) limits, so much greater fidelity 
between the two initial copies is required.  Isolation should be at 
least 125dB at the fundamental and first 11 harmonics, with other errors 
(phase noise, harmonic distortion, etc.) at similar levels.

True isolation amplifiers are therefore substantially less forgiving to 
design and build than even very good distribution amplifiers, and most 
of us don't need them until we have reached at least the journeyman 
level of time and frequency analysis.

My personal advice is to walk before you try to run.  Build or acquire a 
very good distribution amplifier, and use it until your measurements are 
pretty much always limited by the DAs.  *Then* augment your lab with a 
proper isolation amplifier.

Best regards,

Charles