Listen to Latest SolderSmoke Podcast

Monday, September 7, 2015

Si5351 and the Spectral Purity Mask

I was thinking about spectral purity standards and the Si5351 chip. I realized that I didn't even know what the FCC standards for "close in" noise are.   The standards for spurious emissions ARE well known, but these are for harmonics and parasitic emissions relatively far from the desired signal.  What about unwanted signals CLOSE to the desired signal? 

My old 2002 ARRL handbook indicates that the FCC has not established firm standards for this "close in" noise.  (They call it "out of band" noise, but are clearly referring to noise that is close to the desired signal but spreading out beyond the desired bandwidth.  Phase noise would be in their category.)

In the course of my Googling, I found the above spectral purity mask.   I don't know where it comes from, but I think it is the kind of graph that would be very useful to us as we evaluate the merits and shortcomings of various frequency synthesizers.  Would our DDS or PLL rigs fit in this mask?   I think an Si5351 rig WOULD.  According to KE5FX's measurements, at a mere 100 Hz from the center frequency, the Si5351 phase noise is already -90 db.

Does anyone have a similar mask showing current standards?

I still don't understand why so many folks believe that the Si570 is a useful part for homebrew rigs, but the Si5351 is not.  Look at the numbers:

Clifton Labs  measuring at 30 MHz carrier. At 10kHz from carrier:   -109.6 dbc/Hz
Silicon Labs web site (carrier freq not specified) At 10 kHz from carrier:  -116 dbc/Hz

KE5FX measuring at 19.99 MHz. At 10kHz from carrier:  -127 dbc/Hz
Silicon Labs  measuring at 156.2 MHz. At 10 kHz from carrier   -112 dbc/Hz.

Can anyone out there explain the technical basis for the belief that the Si570 is a useful part while the Si5351 is not?   

It is important to keep things in perspective.  ALL of these noise numbers represent VERY small noise levels.   Let's keep is simple and assume a 100 watt carrier signal and a phase noise of -100 dbc/Hz.   That means the phase noise per hertz would be .00000001 watts.  That's watts/hertz.  How much "noise power" would that represent in a typical SSB passband?  Multiply by 2500 Hz and you get 25 microwatts.  That's really low noise levels. Not enough to worry about.  And as we've noted, we've happily used rigs with LC VFOs and crystal oscillators for all these years without every once measuring their phase noise.

Our book: "SolderSmoke -- Global Adventures in Wireless Electronics" Our coffee mugs, T-Shirts, bumper stickers: Our Book Store:

1 comment:

  1. I don't think that's how it works, you are applying that noise directly in the first mixer. Assume you have a receiver with -121dBm to -23dBm dynamic range (typical S1 to S9+50 on a true S-meter), assume you have a +7dBm VFO (usual signal used for ring mixers) that has -130dBc/Hz noise, assume you have a 2500Hz filter. -130dBc/Hz for a +7dBm level is -123dBm/Hz noise injected directly into the first mixer, wich with the typical 6dB loss in ring mixers means a -129dBm/Hz at the mixer output. This noise over a 2500Hz range means a -112dBm noise floor (you multiply noise per hertz with the square root of the bandwidth to get overall noise), wich unless you add a good preamp before the mixer means you will have a S2-S3 noise floor just from the VFO, without even connecting the antenna (remember your receiver path dynamic range starting at S1=-121dBm). Considering an AGC circuit compresses the dynamic range further, you will have a pretty bad noise in the speakers, far from a quiet receiver.

    You should be looking at RMS phase jitter, Si570 has 0.61ps and Si5351 has 3.5ps, the differences are significative, that's why it's much more expensive and people prefer it.

    Razvan (YO9IRF)


Designer: Douglas Bowman | Dimodifikasi oleh Abdul Munir Original Posting Rounders 3 Column