Hi

Here’s the problem:

Spectrum analyzers, synthesizers, signal generatros, and the like all tend
to follow a common design approach. They take the 10 MHz reference in and
run it through a narrowband PLL. Not every one of them, but 99% of them.
The same is true of microwave multipliers and signal sources. What you care
very much about is phase noise out to about 100Hz or so. Past that, it
simply does not matter.

Why?

From the instrument side:

If you are headed to microwaves, the 20 log (N) phase noise formula is
working against you. 10 MHz multiplied to 100 MHz goes from -170 dbc to
-150 dbc. That’s not what you want to see. They long ago came up with the
approach of locking up a VHF crystal oscillator to get -170 dbc at 100 MHz.
The technique came out a long time ago (as in before I started doing this
in the 1970 … or was it the 1790’s …). Cost wise this made sense.
They bought a cheap(er) OCXO at low frequency if they needed stability, and
just ran a simple circuit with a crystal in it at VHF.

From the distribution side:

People expected that if they plugged an HPxxxx into a HPyyyy that it would
meet spec. They even expected it to work if the entire chain was not made
by HP. Generating -170 level signals is hard enough, distributing them
across a building, not so much. The designers made a simple decision, -145
to -155 dbc/Hz phase noise was “good enough” out of a distribution
system or out of a master standard. Could they have done better? Probably.
Would it have run up costs in that era? Most certainly.

From a system standpoint:

The people on both ends of the cable made decisions more or less together.
Who knows who went first or what was tried and failed, that’s buried
somewhere back in the 1950’s. Did everybody drink the same beverage?
I’m sure somebody somewhere didn’t. Every piece of HP gear I’ve ever
seen fits the approach above. Every microwave multiplier I’ve ever seen
or designed fits it. The Fluke and Comstron gear I’ve worked on or
actually seen schematics for works this way. Every distribution amp or
distribution system I’ve seen works this way. I have a nasty habit of
plugging standard lines into phase noise testers. Each time I do, the data
I get supports the decision to do things as shown above.

——————

So what’s this mean?

A simple distribution board made up for less than $10 should do you just
fine for plugging instruments together. There’s no need to go crazy over
broadband noise.

Bob

are 1) the spectrum analyzer(s), 2) a Comstron direct synthesizer, 3) the
synthesized signal generators and the test source(s) used to drive
microwave multipliers and signal sources. All these devices will see the 10
MHz phase noise (improvements) within the narrowest PLL the devices use.

use one for each instrument. Besides it would hurt to go through the
trouble of buying a low phase noise 10 MHz reference and lose it in a poor
distribution amplifier(s). Also, the advantages of running all instruments
from the same 10 MHz source are well known.

effort in the hopes someone on this thread had been there, I'll just "hit
the books" and do some prototyping and noise testing and see what I come up
with.

normally go deep inside a box of some sort. It’s a rare off the shelf
device that takes in the output of a distribution amp and requires that
sort of phase noise.

-170 dbc oscillator will give you -167. A device with a -200 dbc floor will
still “degrade” a -170 dbc oscillator. That’s a fairly big change in
circuit complexity (and cost) for a 2.9 something db improvement. The list
of devices that might make it worth spending (say) a few hundred dollars a
channel versus under a buck a channel is pretty short. That may put a bound
on this.

about multi channel distribution amps. They will add a ground loop(s) /
pickup loop(s) that you will be fighting forever and ever. Do that sort of
stuff straight off the oscillator. There is no rational amount of money
(ummm …. errrr … how much do you have?) you can spend to get around
this. A second (or eighth) oscillator is cheaper than even some of the
simple approaches that don’t work very well. The type of OCXO you are
talking about is a < $50 item on eBay.

can be a real problem since the 10 MHz to be distributed is -170 dBC/Hz at
10 KHz and needs to be preserved if at all possible.

that's out.

CMOS buffers will not add enough noise to matter at 10 MHz. That of course
also assumes that the target gear is the normal bunch of instruments that
we all play with.

https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts [4]

https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts [5]

time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts [6]
and follow the instructions there.

Links:

[1]
http://webmail.timeok.it/parse.php?redirect=https://www.febo.com/cgi-bin/ma
ilman/listinfo/time-nuts[2]
http://webmail.timeok.it/parse.php?redirect=https://www.febo.com/cgi-bin/ma
ilman/listinfo/time-nuts[3]
http://webmail.timeok.it/parse.php?redirect=https://www.febo.com/cgi-bin/ma
ilman/listinfo/time-nuts[4]
http://webmail.timeok.it/parse.php?redirect=https://www.febo.com/cgi-bin/ma
ilman/listinfo/time-nuts[5]
http://webmail.timeok.it/parse.php?redirect=https://www.febo.com/cgi-bin/ma
ilman/listinfo/time-nuts[6]
http://webmail.timeok.it/parse.php?redirect=https://www.febo.com/cgi-bin/ma
ilman/listinfo/time-nuts

Hi

Here’s the problem:

Spectrum analyzers, synthesizers, signal generatros, and the like all
tend to follow a common design approach. They take the 10 MHz
reference in and run it through a narrowband PLL. Not every one of them, but 99% of them.
The same is true of microwave multipliers and signal sources. What you
care very much about is phase noise out to about 100Hz or so. Past
that, it simply does not matter.

Why?

From the instrument side:

If you are headed to microwaves, the 20 log (N) phase noise formula is
working against you. 10 MHz multiplied to 100 MHz goes from -170 dbc
to
-150 dbc. That’s not what you want to see. They long ago came up with
the approach of locking up a VHF crystal oscillator to get -170 dbc at 100 MHz.
The technique came out a long time ago (as in before I started doing
this in the 1970 … or was it the 1790’s …). Cost wise this made sense.
They bought a cheap(er) OCXO at low frequency if they needed
stability, and just ran a simple circuit with a crystal in it at VHF.

From the distribution side:

People expected that if they plugged an HPxxxx into a HPyyyy that it
would meet spec. They even expected it to work if the entire chain was
not made by HP. Generating -170 level signals is hard enough,
distributing them across a building, not so much. The designers made a
simple decision, -145 to -155 dbc/Hz phase noise was “good enough” out
of a distribution system or out of a master standard. Could they have done better? Probably.
Would it have run up costs in that era? Most certainly.

From a system standpoint:

The people on both ends of the cable made decisions more or less together.
Who knows who went first or what was tried and failed, that’s buried
somewhere back in the 1950’s. Did everybody drink the same beverage?
I’m sure somebody somewhere didn’t. Every piece of HP gear I’ve ever
seen fits the approach above. Every microwave multiplier I’ve ever
seen or designed fits it. The Fluke and Comstron gear I’ve worked on
or actually seen schematics for works this way. Every distribution amp
or distribution system I’ve seen works this way. I have a nasty habit
of plugging standard lines into phase noise testers. Each time I do,
the data I get supports the decision to do things as shown above.

——————

So what’s this mean?

A simple distribution board made up for less than $10 should do you
just fine for plugging instruments together. There’s no need to go
crazy over broadband noise.

Bob

are 1) the spectrum analyzer(s), 2) a Comstron direct synthesizer, 3)
the synthesized signal generators and the test source(s) used to drive
microwave multipliers and signal sources. All these devices will see
the 10 MHz phase noise (improvements) within the narrowest PLL the devices use.

use one for each instrument. Besides it would hurt to go through the
trouble of buying a low phase noise 10 MHz reference and lose it in a
poor distribution amplifier(s). Also, the advantages of running all
instruments from the same 10 MHz source are well known.

effort in the hopes someone on this thread had been there, I'll just
"hit the books" and do some prototyping and noise testing and see what
I come up with.

normally go deep inside a box of some sort. It’s a rare off the shelf
device that takes in the output of a distribution amp and requires
that sort of phase noise.

-170 dbc oscillator will give you -167. A device with a -200 dbc floor
will still “degrade” a -170 dbc oscillator. That’s a fairly big change
in circuit complexity (and cost) for a 2.9 something db improvement.
The list of devices that might make it worth spending (say) a few
hundred dollars a channel versus under a buck a channel is pretty
short. That may put a bound on this.

about multi channel distribution amps. They will add a ground loop(s)
/ pickup loop(s) that you will be fighting forever and ever. Do that
sort of stuff straight off the oscillator. There is no rational amount
of money (ummm …. errrr … how much do you have?) you can spend to get
around this. A second (or eighth) oscillator is cheaper than even some
of the simple approaches that don’t work very well. The type of OCXO
you are talking about is a < $50 item on eBay.

can be a real problem since the 10 MHz to be distributed is -170
dBC/Hz at
10 KHz and needs to be preserved if at all possible.

that's out.

CMOS buffers will not add enough noise to matter at 10 MHz. That of
course also assumes that the target gear is the normal bunch of
instruments that we all play with.

https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts [4]

https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts [5]

time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts [6] and follow
the instructions there.

Links:

[1]
http://webmail.timeok.it/parse.php?redirect=https://www.febo.com/cgi-b
in/ma
ilman/listinfo/time-nuts[2]
http://webmail.timeok.it/parse.php?redirect=https://www.febo.com/cgi-b
in/ma
ilman/listinfo/time-nuts[3]
http://webmail.timeok.it/parse.php?redirect=https://www.febo.com/cgi-b
in/ma
ilman/listinfo/time-nuts[4]
http://webmail.timeok.it/parse.php?redirect=https://www.febo.com/cgi-b
in/ma
ilman/listinfo/time-nuts[5]
http://webmail.timeok.it/parse.php?redirect=https://www.febo.com/cgi-b
in/ma
ilman/listinfo/time-nuts[6]
http://webmail.timeok.it/parse.php?redirect=https://www.febo.com/cgi-b
in/ma
ilman/listinfo/time-nuts

On Nov 24, 2014, at 12:53 PM, Bill bill@hsmicrowave.com wrote:

Hi Luciano,

Thanks for sharing that data. Do you have a  link to the NIST distribution amplifier?

When you're ready I hope you'll share more data on your MDT123 design.

Regards...Bill

-----Original Message-----
From: time-nuts [mailto:time-nuts-bounces@febo.com] On Behalf Of timeok@timeok.it
Sent: Monday, November 24, 2014 7:14 AM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] Low Additive Phase Noise 10 MHz Amps

Hi,

This is a good and reasonable description of what we normally need. I agree.
Some amateur like me want some more than the normal, and If you want, more than  the necessary. This happen only for testing ours ability to improve the performance of our instruments we use as hobbist.

For this reason I have developed with the help of three friend a  new solution to distribute or separate a frequency standard. Nothing of revolutionary or incredible but a mid cost, high performance solution.
I hope the solution will be available mid/end  next  years in two versions, a single smd amplifier unboxed to mount it as separator inside an existent instrument, or a complete single/dual channel (total 12 outputs) with multiple input possibility in a 1U rack complete of an high performance AC-DC switchover power supply.

The file attached is to compare some commercial and NIST solutions with this amplifier in the single amplifier version.
I will update you using this discussion list.

Luciano
timeok@timeok.it

On Mon 24/11/14  1:42 PM , Bob Camp kb8tq@n1k.org wrote:

Message sent via Atmail Open - http://atmail.org/

time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.