In the video referred to above, Marc shows a large phase noise at low output frequencies as a result of the 3325A's (fractional-N) PLL-architecture. It's really quite significant, completely obscuring less significant setting digits for LF signals observed over intervals of only a few periods. He shows a frequency jitter on the order of several ppm for output frequencies of a few Hz. Imagine what it's like at a few uHz....
I have tried to find specifications of this but haven't found any. Since it's quite a significant feature, I'd like to know if such specifications exist or may be derived for specific models.

I tried putting a comment re. this in the comments section to the videos but found out that I have been banned after expressing my doubts about the way Marc feeds the frequency reference into the clock. My two polite posts have been made invisible to others (logged in I still see them) and I haven't received any response from Marc after trying several emails to different email addresses. Not important but annoying.

So, I'm addressing this group, hoping for more about the phase jitter.

In case anyone is interested, my posts addressed the following:

1. Marc concluded that a 400 Hz carrier with a 2 Hz modulation (100%) is needed at the frequency reference input. The Soyuz clock seems to use a 2 Hz reference inside, right after the input section.
Although I like Rube Goldberg solutions at times and enjoyed the setup with a cesium clock (5061A), a synthesizer (3325A) and a 400 (800?) Hz signal generator (8116A), I found the need for the rather complicated signal onboard the Soyuz enjoyable but unlikely. Just the 3325A, for fun's sake driven by the 5061A would seem enough. Maybe the input circuit in the Soyuz clock is broken...

Looking at Ken's (?) hand-drawn schematics, I concluded that just a 1 Hz square wave with appropriate edge speed and amplitude should do, since the full-wave bridge rectifier at the signal input would produce same-polarity output pulses from the input circuit at 2x the input frequency already. Using the 400 Hz carrier and (2 Hz) 100% modulation *kills* the internal frequency doubling, so by using that, a 2 Hz modulation is needed.

2. In a second post, I suggested that the diode in the top middle of the drawing probably was drawn upside down.

As said, I was - and still am - banned and no contact has been possible.

I have built up an input circuit exactly like Ken's in the video, except with the diode inverted as I suggested and with a small (pulse) transformer at the input. It operates exactly as anticipated: 1 Hz in, nice 2 Hz pulse out, duty cycle configurable. I used only a 5V supply and a 4-5V input signal. You may try for yourself if you like. Hint: The bridge output *positive* terminal is at gnd.

I wonder why I was banned, resulting in my posts being invisible...

Anyway, I'm interested in quantitative info on the phase noise for specific generators if possible or ways to derive it from specs that *are* available.

Raymond

On 14 Apr 2020, at 01:04, george edmonds via groups.io <G6HIG@...> wrote:

?

Hi Colin

Are you looking at a common hardware problem that effects the SA and power meter not a calibration issue? Just my thoughts.

My main support role is with Marconi kit where calibration problems usually are frequency dependant.

Good Luck with your 8920A

George G6HIG

On Tuesday, 14 April 2020, 00:31:34 BST, Colin Smithers <colin@...> wrote:

Hi George,

?

It is reading around 10% low, compared to the HP438, measured at 1W and 20W, 50MHz.

?

I have not got as far trying to cal this, but again there is no obvious method.

?

Rgds

?

Colin

?

From: [email protected] [mailto:[email protected]]On Behalf Of george edmonds via groups.io
Sent: 13 April 2020 23:56
To: [email protected]
Subject: Re: [HP-Agilent-Keysight-equipment] HP 8920A spectrum analyser amplitude calibration

?

Hi All

Don;t want to distract from the excellent replies, BUT what about this line in the original posting.

"Also trying to calibrate the power meter. I have a calibrated HP438 power meter to guide me"

WHAT ARE THE POWER METER ERRORS?

George G6HIG.

On Monday, 13 April 2020, 22:28:37 BST, Colin Smithers <colin@...> wrote:

?

?

Thank you Robert and everyone else replying. I am collecting all the suggestions and working out which route to take, or whether its simpler to just keep on entering the offset or just remember the error as I have been doing.

?

But most of all I am filled with simple disbelief that this instrument was left so incomplete in self-cal capability, unless it is just well hidden, in which case why is it so well hidden?

?

The point about it being close so it’s very wide spec is the soft option, as the error is VERY constant, across frequency, time and temperature. I think it could hold +/- 0.5 dB if done properly. The instrument reminds me of the schoolchild that ‘could do better, if only he would only try’. A few decades ago I used a different Spec An that was wonderfully calibrated. Then it went off for cal and on return the horizontal span cal was miles out; signals near the edge of the screen were almost 10% off. I complained and the answer came back from the 3^rd party test lab ‘its within spec’. But someone had fiddled as it was not like that previously. However, that is another story.

?

Keep it coming, as its also crystal clear that I am far from not alone with this problem.

?

Kind regards to everyone.

?

Colin

?

From: [email protected] [mailto:[email protected]]On Behalf Of Robert G8RPI via groups.io
Sent: 13 April 2020 22:08
To: [email protected]
Subject: Re: [HP-Agilent-Keysight-equipment] HP 8920A spectrum analyser amplitude calibration

?

Good timing, see latest message from Jeff on the HP 8924 Family group. He;s written some code to do it via RS232 :-)

Robert G8RPI.

You are standing on the precipice of becoming a Time Nut. Fair
warning. ;)

A mostly-related question:
I have an older GPS module (Navsync CW12-TIM) that (apparently) functions as a GPSDO. The output freq is programmable, up to 30 MHz.
The CW12 doc includes (for example) the following: "The graph below demonstrates the ability of the CW12-TIM GPS to continue to provide a GPS disciplined output frequency with the GPS aerial located completely inside a building...". The doc also says the CW12-TIM was designed to meet the form and functionality of the Motorola M12+.

I have limited knowledge of these things.

My question is this: Is this module producing a signal that is equivalent to what you folks are talking about earlier in this thread? Or is there more to it?
Would it be correct to say that my GPS module outputs a signal that has excellent long-term "accuracy", but the short-term "precision" (including jitter and such) may not be so good?

Thanks,
Pete

A mostly-related question:
I have an older GPS module (Navsync CW12-TIM) that (apparently) functions as a GPSDO. The output freq is programmable, up to 30 MHz.
The CW12 doc includes (for example) the following: "The graph below demonstrates the ability of the CW12-TIM GPS to continue to provide a GPS disciplined output frequency with the GPS aerial located completely inside a building...". The doc also says the CW12-TIM was designed to meet the form and functionality of the Motorola M12+.

I have limited knowledge of these things.

My question is this: Is this module producing a signal that is equivalent to what you folks are talking about earlier in this thread? Or is there more to it?
Would it be correct to say that my GPS module outputs a signal that has excellent long-term "accuracy", but the short-term "precision" (including jitter and such) may not be so good?

Thanks,
Pete

?