[time-nuts] Distribution divider/amplifier for 10MHz GPSDO
kb8tq at n1k.org
Mon Oct 23 13:26:16 EDT 2017
The correct answer to any real question like this is “that depends”.
For anything that I normally run as test gear, noise outside a very narrow bandwidth really
does not matter much. The test gear *assumes* (by design) that the reference signal going
into the “ref in” jack is not very clean. It does various tricks with filters and PLL’s to “scrub”
If we are talking about the reference into one side of a phase noise test set, then
the situation is a bit different. The test set is simply going to tell me what the combined
noise is on the two inputs. If one is significantly more noisy than the other, that’s pretty
much all I will see. In this case, my answer is “don’t use a distributed signal”. Use a
stand alone source as your reference and isolate it from the rest of the world.
In any case, making a super duper distribution gizmo and feeding it with a noisy signal
is not going to make the signal any better. Most GPSDO’s have relatively noisy outputs.
Some are better than others. None that I have seen on the surplus market are what
I would call quiet at the output jack of the GPSDO. They either have an ocean of spurs
or a lot of phase noise. Some have both ….
Any time you boost a bunch of signals up to high levels, you create “crud” running around your
lab / shack. One of the most basic questions should always be “do I really need this signal?”. Next
should be “how can I have a shorter run?”. I have many pieces of gear that are rarely used.
They use odd references. When I need to use them I rig a reference. That gets shut down
once the gear goes back to storage. …. no more birdies every 100 KHz …. No need for
tripple shielded coax ….
Square up the 10 MHz (or whatever) by matching it into a 5.5 V powered high speed CMOS
gate. The NC7SZ series is one of many families you can use. A NC7SZ125 is not a bad gate
to pick. Distribute the square wave to however many output amps as you need. Each one
is another of the same gates with the output matched via a 50 ohm to 50 ohm lowpass Tee network
with a low Q ( < 2). Likely pad down the output a bit to keep it at a rational level. Build up however
many you need for however many frequencies you require. Very normal linear regulator chips
are fine for the power. Careful bypassing and solid ground planes are always a good idea.
Parts cost wise, postage is likely to cost you more than the components. There are …. errr…
many thousands …. of multi output amps of this basic design out there …. they seem to
work pretty well.
Yes, there are *lots* of possible twists and turns to this. I’m only guessing about the gear you
are trying to run and what you are trying to do with it.
> On Oct 23, 2017, at 12:45 PM, Tom Van Baak <tvb at LeapSecond.com> wrote:
> List -- Don is having email trouble, but here's his posting:
> From: donaldbcollie at gmail.com
> Date: Tue, Oct 24, 2017 at 1:38 AM
> Subject: Distribution divider/amplifier for 10MHz GPSDO
> Hello group. I`m intending to distribute, via 50 Ohm coax, frequency
> reference signals to my test equipment in my test bay [no relation to eBay,
> except that most of the equipment came from there]. I`ll be using RG58/U
> coax, and 50 ohm terminations, with the highest reasonable signal level
> reticulated. Given that the name of the game seems to be to avoid any
> severe reduction in SNR of the 10MHz signal comming out of the GPSDO, by
> the logic dividers, and impedance lowering buffer amplifiers, what
> considerations should be made regarding the choice of logic families, and
> transistors to be used? The frequencies required by the test equipment vary
> from 500kHz to 10MHz, and amplitudes from 100mV P-P sinewave, to 5V peak
> squarewave. How good must the PSU be to stop the rot getting worse, and is
> 1/f noise in the active devices important? Your thoughts will be
> P.S.: How accurate is the Trimble Thunderbolt for this
> 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