[time-nuts] Any reason not to use one power amplifier and splitter for distribution amplifier?

Charles Steinmetz csteinmetz at yandex.com
Sun Jan 4 03:39:48 EST 2015

Dave wrote:

>I was looking to make a 10 MHz distribution amp to feed test equipment with
>the output of a GPSDO.
>      *   *   *
>16-way Minicircuits splitter on eBay which I got for $40. I guess the loss
>is around 12 dB.
>Is there any reason not to just drive that with 22 dBm or so of power to
>get 10 dBm at each of 16 ports?
>Is 10 dBm an optimal value?
>I see several distribution amp designs witb one amplifier on each output,
>but is it just a lot less hassle to split a higher power amp.

It all depends on what you need.  If the job is feeding the time base 
inputs of test equipment from a GPSDO, the splitter approach should 
work fine if you bear a few things in mind.

First, the splitter achieves its rated isolation only if it is 
matched well at the input and *all* outputs.  That at least means 
using dummy loads on unused outputs, but consider that the REF inputs 
of most instruments are not 50 ohm loads (usually ~ 1k ohm).  So, you 
would need to provide proper termination for the in-use outputs, 
too.  Ideally, it would be in the form of a 52.6 ohm pass-through 
load at the instrument, so the coax is terminated in 50 ohms at the 
load (this assumes the instrument has a 1k ohm input -- if not, the 
pass-through terminator would need to be recalculated -- but note 
that only 50 and 75 ohm pass-through terminators are likely to be 
available as commercial items).

Second, the amplifier should put out about 25dBm from a 50 ohm 
source, so each load will receive the standard 1Vrms = 13dBm (10dBm 
would probably work OK for most instruments, but they usually are fed 
13dBm).  If the amplifier is not naturally matched to 50 ohms, there 
will be additional loss in the matching network.  Amplifiers that put 
out 25-35dBm with harmonics below, say, -50dBc are not trivial to 
design or build.  (Not terribly difficult, but not trivial.)

I used the splitter topology for the multicoupler that multiplexes 
antennas to eight receivers.  In that case, an amplifier with 
extremely low noise and high dynamic range, with a natural 50 ohm 
output -- the sort of amplifier people use these days for post-mixer 
amps in HDR receivers -- works very well.  But the radios are all 50 
ohm loads to begin with, so I didn't need to muck about with odd 
value pass-through terminators.

For distributing 10MHz to test equipment, I find it much easier to 
get good results with a distributed amplifier approach.  The circuit 
I posted back on Nov. 26 is about as simple as it gets, but can be 
made even simpler.  As drawn, each stage uses a 1:1:1 transformer, 
with the output taken from one of the windings.  If the output 
winding is deleted, the transformer becomes a simple 1:1 and the 
output is taken through a 10nF blocking capacitor straight from the collector.

Best regards,


More information about the time-nuts mailing list