need help with measuring PLL phase noise

[robismyname]

site:www.edaboard.com pll phase noise

I attached a screenshot of my locked freq. synthesizer. I would like to measure the phase noise. How do I do this?

Do any of these screenshoots look better in terms of phase noise?

 

[biff44]

Can't tell. Lower the VBW to 3 Hz and measure the cases again. It looks like there is some wobble vs time in the curves.

If I had to guess, I would say the top most curve (probably your lowest loop bandwidth) had maybe 5-10 dB better phase noise close to the carrier. It also attenuates your clock spurs the best.

What is that spur in the top plot at 30 KHz offset? Do you have anything switching at that rate?

 

[toonafishy]

It is hard to see but i'm guessing your loop bandwidth is around 20kHz. Try putting the carrier on the left side of the screen and do a slow sweep (very narrow RBW) over 100kHz or so. If the oscillator is stable, this will help. The 10MHz reference spurs are very high. What's up with that?

 

[kspalla]

I suppose you need to measure under average mode say 100.

 

[vfone]

Use this formula to measure Phase Noise using a Spectrum Analyzer:

Phase Noise[dBc/Hz] = Carrier_Power[dBm] – Noise_Power@Frequency_offset[dBm] – 10*LOG(SA_RBW[Hz])

 

[kspalla]

The above calculations are true if and only if the noise floor of the SA is low enough.

 

[vfone]

Noise floor of the SA low enough? When actually the formula specify the measurement value for both, carrier and side-band power.

I forgot to mention that the Spectrum Analyzer should be power-on also.

 

[robismyname]

The above calculations are true if and only if the noise floor of the SA is low enough.

Well my noise floor is measured at -110dBm (is this low enough?), the carrier power (at 129.3MHz) is measured at -6.7dBm, noise power (at 129.4MHz is measured at -88.75). see attachment.

Using vfone's formula I calculated phase noise to be -120.22 dBm.

My question is: Is vfone's formula valid in my situation? Can the phase noise measurment number (-120.22dBm) be greater than your noise floor(-110)?

 

[Mazz]

rob

a shorter and real time measurement is to use the marker noise function (delta marker -> give a certain offset from CW-> use noise marker, you get directly phase noise) that is available in your Agilent Sa.

I hope it can help.

Mazz

 

[robismyname]

rob

a shorter and real time measurement is to use the marker noise function (delta marker -> give a certain offset from CW-> use noise marker, you get directly phase noise) that is available in your Agilent Sa.

I hope it can help.

Mazz

Mazz,

I used delta marker at an offset of 100kHz from cw (see attached screenshot). I measured -90.99dB (not dBm?). Is this a good value for a phase measurement?

 

[vfone]

Using delta marker is ok, but is not giving the result in dBc/Hz (only if you use 1Hz RBW option, which is very slow, or is not available).

The Noise Floor should be 10dB lower than your side-band measured power. You can use an LNA in the front of the analyzer to lower the noise floor, but the new PSA from Agilent (which you are using) do not need this.

 

[Mazz]

vfone
the measurement is in dBc/Hz. If you just see the picture there are roughly 6.5 divisions, that correspond to 65 dB + 10Log300= 90 dBc/Hz.

rob

You just pickup the spur tone at 100 khz offset. So real phase noise is roughly 10 dB better (move marker to 110 kHz to check).
-100 dBc/Hz @ 100 KHz is a good value.

To have more informations let me suggest to collect these numbers (use auto coupling RBW & VBW for simple operation and make average - 16 is enough):
SPAN Delta Marker Phase noise
3kHz 1kHz ??
30kHz 10kHz ??
300kHz 100kHz -100
3MHz 1MHz ??
30MHz 10MHz ??

Take care of that spur at roughly 20 KHz. this can degrade overall noise (it can be dominant over phase noise)- Measure it (without noise marker).

Mazz

 

[vfone]

From the picture of the SA display I can see that the RBW is 300Hz.
Just happened accidentally to be the correct number, because you just mentioned, was a reading of a spur.

 

[robismyname]

Use this formula to measure Phase Noise using a Spectrum Analyzer:

Phase Noise[dBc/Hz] = Carrier_Power[dBm] – Noise_Power(at)Frequency_offset[dBm] – 10*LOG(SA_RBW[Hz])

vfone

where did you obtain this Phase Noise formula. I would like to add it to my reference library.

 

[vfone]

Is nothing fancy with this formula. Is just the difference between two absolute power levels (carrier and side-band), and subtract from the result the 10*LOG(RBW) of the SA to obtain a value in dBc/Hz.