measuring settling time of a frequency synthesizer [hlp]

[No one]

Hi all;
I want to measure settling time of an RF frequency synthesizer, but I don't know how I should do it. Can anyone help me? Can I do it by a spectrum analyzer? Can anyone provide me links, papers, articles, application notes about this problem?

 

[Kripton2035]

you select a square wave at the maximum frequency your RF synth can do. you look at the signal with an oscilloscope that has at least twice the bandwith your RF has (the more bw the better you measure..) and you measure the time the square signal takes to raise from 10% to 90% of the square.. good luck !

 

[Element_115]

To measure settling time you should use a
Modulation Domain Analyzer.
See the following website (PDF) .

literature.agilent.com/litweb/pdf/5966-4092E.pdf#search='Modulation%20Domain%20Analyzer'

Hope this helps
Cheers

 

[echo47]

Yes, MDA works great. I use an HP 5373A. If your budget is small, maybe search for a used 5372A. Beware, these two instruments are confusing to operate. I've never tried a 53310A.

Or, if you have access to the synthesizer's loop control voltage, that may be all you need to look at.

 

[BigBoss]

The simplest way to set the span to zero of spectrum analyzer and sweep time should be very slow regarding to normal sweep time. If you know exactly the frequency of synthesized, you set the center of spectrum analyzer, set the span to zero and slow down the sweep. if you can, you may apply a trigger signal to start-up the sweep.

Or, simlpy connect a oscilloscope on Vtune and set ext trigger to SEN ( Send Enable ) of the PLL, oscilloscope will start to sweep coherently with falling edge of SEN and Vtune will trace the Vtune in a slow motion then it stops (because of single sweep is selected )

Or best way to use Time-Frequency Domain analyzer from R&S ( FSEB or similar ) or HP.

 

[biff44]

I like to know when the phase of the synthesizer has settled out. Most of the methods mentioned above are only good for frequency settling.

A common way is to use two synthesizers (one fixed, one hopping in frequency) and beat them in a mixer with a DC coupled IF. Watch the IF on an oscilloscope, and when the beat note goes away and the DC voltage stabilizes, you are settled. You typically have to run both synthesizers off of the same clock, and often need a manual phase shifter to get the mixer into a good operating point (quadrature).

Some modern synthesizers for some divisor ratios, like fracitonal N or DDS actually never settle out in phase, they walk slowly in phase in the steady state, so it make things a little more interesting.

 

[echo47]

BigBoss, in your first paragraph, it sounds like you are using the spectrum analyzer as a tuned receiver. That would plot amplitude versus time, and not frequency versus time. Or did I misunderstand?

Biff44, that's an excellent technique for measuring extremely small effects. I use it to monitor the stability of 10 MHz timebase references down to 10E-12 or so.

 

[biff44]

You can measure big effects too, especially if you have a digital storage oscilloscope. If you are off in frequency by 10 KHz at 1 uS delay, you will see a 10 KHz beatnote at 1 uS.

 

[echo47]

Yes indeed!

I sometimes use my MDA to measure the mixer's output. Other times an oscilloscope is fine. Once I used my old HP 8505A vector network analyzer because it has three inputs, so I could compare the phases of three oscillators simultaneously. I used it to locate which instrument on my bench had the slightly noisy OCXO.