Topics

#qcx Si5351A Phase noise measurements #qcx

Hans Summers
 

Hi all


I have published an article on phase noise measurement techniques and in particular, relating to the phase noise of the Si5351A in the QCX. Having developed the equipment and skill to make this measurement is also expected to be very useful for final performance testing of the forthcoming QSX transceiver. 

I am grateful to Kevin ZL1UJG and the 100+ correspondence emails we had on this topic; some information from Kevin is also presented on the page and agrees with my conclusions. I had always feared phase noise measurement and thought it out of reach of mere mortals such as myself but was pleasantly surprised; to be sure it isn't trivial but neither is it as difficult as I had feared ;-) 

The QCX phase noise performance was found to be GOOD, and my measurement matched ARRL to within 1dB in their QST August 2019 review (ARRL measured -135 dBc/Hz at the transmitter output, at 10 kHz and 50 kHz spacing). The review is reproduced with permission, at http://qrp-labs.com/qcx . Various things were tried to improve the phase noise performance (capacitor and inductor filtering of the supply to the Si5351A) - none made any difference; the conclusion is that the QCX is already as good as it gets and performance is determined by the PLL of the Si5351A chip. 

Other conclusions from the experiments are relevant to anyone using the Si5351A. Specifically, heavy loading (50-ohm impedance) damages the phase noise performance of the Si5351A quite significantly. This can be partially mitigated by using a 0.1uF DC-blocking capacitor. A clean supply is also important (Kevin's results). Previous acquired knowledge on Si5351A output loading indicates that heavy loading will also cause crosstalk between outputs, as well as deteriorate the output waveform and reduce its amplitude. 

Therefore anyone using the Si5351A in applications where high drive levels are required, such as a diode ring mixer, should make sure that they buffer the Si5351A outputs, not use them directly connected to the mixer or 50-ohm attenuator pads. I have seen some Si5351A breakout boards that use small 50-ohm transformers at each Si5351A output and this is also wrong (from the perspective of phase noise performance). 

73 Hans G0UPL

Richard G4TGJ
 

Hi Hans
A very interesting article.

In the section "Heavy loading at 50-ohm" the photo is missing. It just says "{gallery}qcx/phasenoise/13/gallery}".
--
73
Richard
G4TGJ

Hans Summers
 

Thanks Richard, 

I just fixed it!

73 Hans G0UPL

On Wed, Oct 2, 2019 at 12:43 PM Richard G4TGJ <rpt@...> wrote:
Hi Hans
A very interesting article.

In the section "Heavy loading at 50-ohm" the photo is missing. It just says "{gallery}qcx/phasenoise/13/gallery}".
--
73
Richard
G4TGJ

jmh6@...
 

Hi Hans :),

Nice piece of work.

John

On Wed, 2 Oct 2019, Hans Summers wrote:

Hi all
http://www.qrp-labs.com/qcx/phasenoise
I have published an article on phase noise measurement techniques and in
particular, relating to the phase noise of the Si5351A in the QCX. Having
developed the equipment and skill to make this measurement is also expected
to be very useful for final performance testing of the forthcoming QSX
transceiver. 
I am grateful to Kevin ZL1UJG and the 100+ correspondence emails we had on
this topic; some information from Kevin is also presented on the page and
agrees with my conclusions. I had always feared phase noise measurement and
thought it out of reach of mere mortals such as myself but was pleasantly
surprised; to be sure it isn't trivial but neither is it as difficult as I
had feared ;-) 
The QCX phase noise performance was found to be GOOD, and my measurement
matched ARRL to within 1dB in their QST August 2019 review (ARRL measured
-135 dBc/Hz at the transmitter output, at 10 kHz and 50 kHz spacing). The
review is reproduced with permission, at http://qrp-labs.com/qcx . Various
things were tried to improve the phase noise performance (capacitor and
inductor filtering of the supply to the Si5351A) - none made any difference;
the conclusion is that the QCX is already as good as it gets and performance
is determined by the PLL of the Si5351A chip. 
Other conclusions from the experiments are relevant to anyone using the
Si5351A. Specifically, heavy loading (50-ohm impedance) damages the phase
noise performance of the Si5351A quite significantly. This can be partially
mitigated by using a 0.1uF DC-blocking capacitor. A clean supply is also
important (Kevin's results). Previous acquired knowledge on Si5351A output
loading indicates that heavy loading will also cause crosstalk between
outputs, as well as deteriorate the output waveform and reduce its
amplitude. 
Therefore anyone using the Si5351A in applications where high drive levels
are required, such as a diode ring mixer, should make sure that they buffer
the Si5351A outputs, not use them directly connected to the mixer or 50-ohm
attenuator pads. I have seen some Si5351A breakout boards that use small
50-ohm transformers at each Si5351A output and this is also wrong (from the
perspective of phase noise performance). 
73 Hans G0UPL
http://qrp-labs.com

zl1ujg
 

Hi, 
What Hans didn't mention, that after countless plots this end of the QRPLABS VFO, and other units, in their various guises, and the final plots I had a whoops...
The PSU powering the VFO unit, had come up on 13.8v, not 5v.. and I summarily cremated the ATMEGA328 in an instant..

With a surname of Murphy.. it was overdue..Lol

Kevin

Shirley Dulcey KE1L
 



On Wed, Oct 2, 2019 at 4:16 AM Hans Summers <hans.summers@...> wrote:
 
>I have seen some Si5351A breakout boards that use small 50-ohm transformers at each Si5351A output and this is also wrong 
> from the perspective of phase noise performance).

Prototypes of the Etherkit breakout board had transformers. It was originally offered through an Indiegogo campaign and the board pictured on that site had the transformers. But Jason Milldrum NT7S found that they didn't provide the expected benefits so they were omitted in the final version of the board; the Si5351A outputs come directly to board pads or optional SMA connectors. The boards shipped to Indiegogo backers and later sold on the Etherkit site have no transformers. All versions of that board are Creative Commons designs, so it's possible that somebody made the version with transformers in quantity.

The Adafruit Si5351A breakout board, which is probably the most popular one, has never had transformers. The clones from China, most of which have purple PCBs, are the same design with no transformers.

Those two boards and the one from Hans (QRP Labs) are the only three reasonably priced designs that I know about.