Hi

no, 13 to 15 is better.

the reason for that hassle is to better isolate the outputs. The splitter has good isolation only when all the ports are properly matched. In the case of a “40 db isolation splitter” that can mean the terminations all have 40 db return loss.

Roughly 99.9% of all those little amps degrade the phase noise of the input signal significantly when driven near their “rated” output power.

If you are after 13 dbm after 12 db loss you need 25 dbm clean signal. If you need to run the amp 10 db below it’s rated output to keep it clean, that means an amp rated at 35 dbm output. Most of those amps are 10% efficient, so you will be putting 20W into the device.

The next layer is the return loss on the output of the amp. If you are after the full isolation numbers it needs to be matched as well as the terminations. If you need a pad to achieve that match, it comes out of your power budget.

Since your test gear typically does not provide a 50 ohm match (they are high Z inputs), making this sort of setup work is fairly complex. You need to terminate at each device and terminate each unused output on the splitter. Even then

For under $20 in active parts you can build a distribution box that will do a far better job than a big amp driving a multi port splitter.

Bob

Dave
I am sure there will be many answers. But yes indeed it will work fine. All
of the outputs should have the same delay also and that may be useful.

There really are the two approaches. The big power amp thats a 1/2 Watt you
are talking and the many small amps as in the distributed mode.

Whats interesting is the telcos always do the big amp splitter and the test
equipment manufactures use precision distributed distribution amps. I guess
its a pick your poison. Or maybe the test equipment manufacturers needed
more isolation port to port. Or heavens maybe they could just sell them for
more money. Would they do that?
Humor aside each has a very good reason for doing the distribution and its
driven by the requirements.

I have several of those spitters and picked them up for $ 0 at hamfests.
Seems no one had a use for them when all of the 900 Mhz gear came out of
the sites. Mostly gone at this point. A 1/2 watt 10 Mhz amp is not that
hard to build look at the many Ham sites we have a band close to 10 Mhz/ 30
Meter.
Regards
Paul
WB8TSL

On Sat, Jan 3, 2015 at 6:09 PM, Dr. David Kirkby (Kirkby Microwave Ltd) <
drkirkby@kirkbymicrowave.co.uk> wrote:

My 2 cents. I just repurposed an RGB video distribution amp for use
with a rubidium. The board has three EL2070 (200 MHz BW) amps, each
one feeds several BNC jacks for RGB output. I only need three
separate, isolated outputs, so I am feeding the rubidium into all
three EL2070's and only taking one output from each chip, via an
isolation transformer.

This seems to be working just fine for my intended purpose. I still
need to add the DDS and a few other items to the box to complete the
project.

Joe Gray
W5JG

On Sat, Jan 3, 2015 at 8:37 PM, paul swed paulswedb@gmail.com wrote:

Hi

The telco types were used to the need to terminate all the lines correctly. It’s been part of their DNA for > 100 years. They also are used to designing cable plants rather than just tossing wires around. They also have some  needs like > 20 year MTBF, so orderly is a good idea.

If I see a properly terminated single drop standard line in a real working lab, it will be the first time I see one. BNC Tee’s seem to be a popular item when hooking up instruments in a typical lab. (Note - in this case the definition of “properly terminated”  is > 40 db return loss, so it’s as good as the splitter).

More or less - the instrument guys opted for a “tap off multiple instruments” approach. The telco guys opted for a “individual run to each instrument” approach. Did every single outfit all over the world read the exact same book? Certainly not, the real world is never that orderly. Telco’s are not 100% fixed, labs are not 100% random wired.

So what happens:

Telco has a problem, it’s a fixed plant. They track it down and do what’s needed. The problem (hopefully) does not come back.

Lab has a problem, It’s Bob over on the third bench from the end. They try to track it, Bob moves on in his troubleshooting. They never find it. Much oratory, very little labor. How often does that happen? Roughly three of four times a week where I work… We find it and fix it about half the time. The rest of the time it “just goes away” or people get tired of looking. Most common issue - can’t run a clean phase noise / ADEV / vibe plot with that silly gizmo running on (name deleted to protect Bob)’s bench. Less common issues - can’t get a clean counter reading on standard line #XXX.

So yes, a real lab does go for more isolation and better protection that a telco. Undiagnosed / random problems from shorts and opens are very real in the lab, not so much in the telco. 10 MHz threat signals and 10 MHz standards are common in the lab, less so in the telco. Different systems, different needs, different solutions.

At least in my basement, I’m a lot closer to a lab than a telco. I move stuff around, I turn gear on and off. I set things up and tear them down. I do indeed work on a lot of stuff that runs at 10 MHz. I have multiple 10 MHz sources running all the time. Troubleshooting every cable connection for issues is not much fun. Isolation does indeed matter in a situation like that.  Even with all my junk, I don’t have a need for a 16 output system. I doubt I ever have more than a dozen taps running on my Spectracom at any one time. 16 boxes is a lot of stuff.

Are there some exotic situations that come up - sure. What’s exotic to me might be normal to you. My typical might be your exotic. Do close in spurs matter to you? They do to me. Does a “burp” (phase shift) in all the standards when a cable is unplugged matter to you? It does to me. Do you run stuff 24/7 365 days a year? I do sometimes (yes yell at me for “sometimes and 365”). Do you need -175 dbc/Hz  phase noise on your standard lines? I don’t.  Do you need a range of standard frequencies (100K, 1M, 5M, 10M)? I do. Do you have some stuff (= almost everything) that’s fine at 1x10^-11 at 1 second wired one way and other stuff wired another? I do. How much do you listen to WWV at 5 or 10 MHz? Me - not so much.  Again, different needs, will drive different solutions.

Bob

Hi

Ok, so what does the math look like with no injection locking:

If you have two equal tones, obviously you have a hard time figuring out which one is the correct 10 MHz. That’s not going to be the situation with a distribution amp.

If you have a threat signal that’s 60 db down from the 10 MHz “carrier” that’s a single sideband signal. Run it through a limiter and you get a phase modulated / frequency modulated  signal. Frequency modulating your standard line is probably not a good idea.

How much FM do you get from a signal X db down?

1. It’s a function of how far off the signal is

2. It’s a function of the level of the signal.

3. It’s a function of the limiter

If you have an objective of 1x10^-11 at 1 second, what’s that come out to?

Phase sidebands 86 db down at 1 Hz off will put you at 9.6 ppt against your 10 ppt budget.

Bob

On 4 January 2015 at 02:37, Bob Camp kb8tq@n1k.org wrote:

I bought two of these splitters. One covers 1-30 MHz, the other one
5-200 I think. Anyway, the 1-30 MHz one arrived today, the
Minicircuits ZFSC-16-3.

I had my VNA on, which unfortunately does not drop down to 10 MHz, or
even 30 MHz, as its lower limit is 50 MHz. So I admit these results
are a bit shaky, as I'm using the splitter outside its range. But that
said, There seems to a  "reasonable" amount of isolation at 50 MHz,
even when all other ports are open.

At 50 MHz, the loss from the common port is 12.8 dB, and the isolation
between two ports sets of ports is either 38 or 48 dB, depending on
what ports are chosen. BUT it appears to be improving as one goes
lower in frequency. The Minicircuits data sheet makes it clear there
are two sets of isolation figures. I'm pretty sure internally this is
likely to be a pair of 8-way splitters.

When the other splitter is here, I will use another VNA that covers
300 kHz to 9 GHz, and so make some measurements at 10 MHz on both of
them. But I just happened to have this VNA calibrated in N, and so
done a few quick and dirty measurements.

I've made no allowances for losses of cables.

So although I admit these measurements are not well done, but I'm not
so convinced now that the terminations on the unused ports matters a
huge amount.

I looks to me that this unit might well exceed 40 dB isolation on all
ports at 10 MHz, as all the graphs are sloping in the right direction.

Dave