Why Multiplier's spectrum noise floor raised?

[tony_lth]

I have made a X14 Multiplier with SRD, the base freq is 57.55M. The X14 multiplier works well, the phase noise loss about 23dB, almost without any additional loss. But I have output spectrum like attached photo, why that happens?
I guess maybe:
1. The Amplifier I used is in saturated status. For driving SRD, I have to push power higher.
2. Maybe the noise floor of the amplifier is high. But almost none vendor issue the noise floor of amplifier, how can I know it?
The amplifier I used is Mini's PMA5452 and ERA-4SM+.
Thanks in advance.

 

[biff44]

-133 dBc/Hz at 10 MHz offset. Thats not too bad. What's your problem?

What is the source phase noise at the input frequency at that offset?

 

[tony_lth]

I made freq source for weather radar. When I tested, the input source is Agilent Source generator, it's phase noise as following: -132dBc/Hz@1KHz, -137dBc/Hz@100KHz... The tested output phase noise is -109dBc/Hz@1KHz, -114dBc/Hz@100KHz...
23dB loss for 14 multiplier.

In radar system, the OCXO I used should have -155dBc/Hz@1Khz. So the output should have -132dBc/Hz@1KHz. And the target phase noise should be better than -140 dBc/Hz @1MHz or 10MHz. But the noise floor raised, so I am afraid that the phase noise can't reach the target. Because I have no chance to install my module tested in radar, I don't know what will happen in the real system. BTW, the system chain NF is about 22dB from OCXO to my output.

Is the noise floor raised by AM noise? When I simulated with ADS, I met several times that AM noise worse than phase noise, but I modified the setting, and ADS simulated result show the AM noise should better than -140dBc/Hz @ >1KHz. In my new design, I change the amplifier to higher P1dB, the AM noise simulated result improved more than 10dB.

Anyone have any ideas?

 

[biff44]

Well, we can't tell from that measurement. Redo the measurment with the span at say 2 MHz, with the resolution bandwidth at 100 Hz, and repost it, then we can talk.

 

[tony_lth]

OK, I will test it next Monday, thanks.

 

[tony_lth]

Hi, I tested the spectrum as your suggest.

E8267D Signal Generator at 57.55M_20K_RBW3Hz_VBW1Hz_12dBDiv.jpg for test source, pls notice that a div is 12dB.

E4440A PSA 2M_RBW100Hz_VBW30Hz.jpg for test Multiplier output.
E4440A PSA 20K_RBW10Hz_VBW3Hz.jpg for detailed test Multiplier output. There are some spurs, but they are generated by source. When I tested phase noise with E5052A source analyser, I can see it clearly.

 

[biff44]

Ok, so now what is your question.

You see that the big flat hump of noise from 0 to +/- 5MHz from your original post is now gone. It was a quirk of the resolution bandwidth.

In your last post, the final plot, is that the signal source, or the signal source PLUS the amplifiers?

 

[tony_lth]

Rich,
Thank you very much. The last post is the signal generator of Agilent E8267D, no Amplifiers.

 

[tony_lth]

I have tested the multipliers module with OCXO input, the result proved that biff44 is right. Thanks Rich again.

 

[biff44]

I am? well...of course! Right about what, that the phase noise originally measured was more a quirk of the way you measured it, instead of the actual phase noise of the multiplier chain?

Tight phase noise specs are more common today, so one does have to pay attention to all of the details of the measurement system!

 

[zl1ujg]

Hi,
The noise on the first plot is a function of Bandwidth settings of the Spec An (470 kHz) and multiplier function (x14) and noise floor of the Oscillator/Multiplier Output.
In high order multiplication, such as with SRD the level of the output can be many dB down and this deterioates the noise floor above that of the Oscillator and Multiplier figures. The level has to be made up with amplifiers, but the noise floor has already worsened. By using using small multiplier ratio's such as x2 and x3, the deterioation is kept to managable levels.
Yes having a 14x multiplier is great for having output, but one doesnt get something for nothing.
Kevin