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

Dr. David Kirkby (Kirkby Microwave Ltd) drkirkby at kirkbymicrowave.co.uk
Thu Jan 8 10:11:21 EST 2015

On 8 January 2015 at 10:03, Charles Steinmetz <csteinmetz at yandex.com> wrote:
> Dave wrote:
>> Yes, but I was aware of this, and that's why I got two different isolation
>> figures.
> What I was pointing out is that there will be *4* different isolation
> figures from any one output port, not just two.  The lowest will be to the
> one electrically adjacent output, next (a bit higher) will be to two outputs
> at one remove, then (a bit higher yet) to four outputs twice removed, and
> finally (the highest) to eight outputs thrice removed (these eight are the
> ones you are calling "on the other 1:8 splitter").
> Perhaps a diagram is in order (see below).  Each dot represents a 1:2
> splitter.  (I've drawn only 1:8.  A 1:16 is just this plus another 1:8,
> connected to the two outputs of another 1:2 splitter.)
> To find the worst case output-to-output isolation, you need to identify two
> electrically adjacent output ports -- either by measurement or from the
> manufacturer's published data.
> Best regards,
> Charles

The Minicircuits data sheet

basically shows there are two different isolation values, which it
calls "adjacent" and "opposite". These are typically 32 and 48 dB @ 10
MHz. I had a quick play yesterday, almost randomly putting the two BNC
connectors in different sockets, and there does appear to be only two
values of isolation values. What I did notice though is that if the
common port is not terminated properly, that dramatically changes the
isolation. That might be an issue, as I don't know what the return
loss (S22) of the amplifier will be, although that is something I can

I need to order up a keyboard and mouse, and then I can use this
Agilent N3383A VNA properly. Then I'll be able to make measurements at
10 MHz. (I bought the thing, then sold it a week later to a friend's
company. He is working in Singapore, so he said I can borrow it until
he gets back. His contract ends in July, but it might be extended, so
this is a semi-permanent fixture here! Where else could you hire a 9
GHz VNA for 9 months for free?)

If there was a design for an isolation amplifier around with a PCB
available that worked well, I would build it. But for now at least,
this might be good enough for what I want, which is basically to
ensure my spectrum analyzer, signal generators have a common frequency
reference. I don't think I have a need for great isolation, as I don't
intend this to be used for the sort of precision measurement time-nuts

BTW, I was looking at the design someone posted for one which used a
discrete transistor and transformer.  I forget which transistor it
was, but I could get 100 delivered from China for £1.00 (about
$1.60)!! I would not buy from China via eBay however, as the chances
of them being original transistors are pretty remote, so I doubt their
characteristics would be like those used to model that design.  The
transistors are only £0.07 from Farnell, so buying them is easy &
cheap enough. I just don't fancy the hassle of making a PCB if this
16-way splitter will do.


More information about the time-nuts mailing list