EMRFD Message Archive 13969

Message Date From Subject
13969 2017-06-04 23:26:46 Hans Summers Re: Digest Number 2917

Hi Craig, all
 Regarding software using the Si5351 phase field to generate quadrature on one of the channels - there are only 7 bits of resolution in the field so I don't think this is enough accuracy to do the job properly.

 The field is CLKx_PHOFF[6:0]

 I haven't measure it yet.

 Adding 20dB of phase noise plus inaccurate quadrature sounds like too much of a compromise to save a couple of dollars and a few milliamps of power.

The cost difference between an Si5351A and an Si570 is significant. About a factor of 10. It does not matter very much if you are creating a one-off homebrew project. On the other hand if you are creating a kit or a product for production, then such a cost difference becomes very important. Elecraft use the Si5351A as the LO in the KX2 transceiver, very successfully. 

In manufacturing products rather than building a one-off homebrew project, there are many trade-offs. Trade-offs are the same as other components. NP0 capacitors and 0.1% precision resistors are available. But expensive. In some places of the receiver (or whatever we are building) choosing the expensive NP0 capacitor or 0.1% precision resistor makes a worthwhile improvement to the overall receiver performance. In others it has no benefit compared to a cheaper component such a s standard ceramic capacitor or a standard 1% resistor. Sames goes for the voltage rating of components, the wattage rating of resistors, etc etc. Good engineering is a matter of selecting the components wisely and appropriately such that the overall result has high performance and is economic. 

For reference the QRP Labs VFO/SigGen kit people mentioned is here http://qrp-labs.com/vfo and the price is $33 (two outputs, 3.5kHz to 200MHz, rotary encoder tuning with configurable step size, optional GPS discipline, presets, IF offset, relay control for band switching, etc).

In the VFO kit the quadrature output mode allows configuration of +90 or -90 (called 270) phase offset. Due to the limitations of the configuration parameters in the Si5351A it works down to a minimum frequency of 3.2MHz. The upper frequency limit is the upper limit of the Si5351A i.e. 200MHz. 

The phase offset register specifies the offset in units of the number of 1/4-cycles of the internal VCO frequency. If even integer MultiSynth dividers are used in the Si5351A configuration, as per the datasheet recommendation, then the 90 degree phase offset is precise. I got 50-60dB unwanted sideband cancellation in a 40m phasing direct conversion receiver using this LO; changed to 20m and the good sideband cancellation remained; changed again to 17m and still fine. Not exhaustive tests across the entire frequency range, but a reasonable indication. Ken G4APB wrote this article http://www.qrp-labs.com/receiver/u3s4m.html about using the QRP Labs SDR receiver module http://qrp-labs.com/receiver on 4m using the quadrature VFO (not a 4x VFO divided by 4).

Regarding phase noise, again, everything comes down to trade-offs, what is important for performance, and at what price. I don't know that the Si5351A is 20dB worse than an Si570, I'm not sure where this figure comes from? The datasheets are rather poor. Anyway I am sure the Si570 IS better than an Si5351A in terms of phase noise. But a top end DDS such as the AD9910 could do 20dB better at HF than an Si570. With high cost and complexity. How much is necessary for the application and market? These are the important questions... all three (DDS, Si570, Si5351A) are better than the vast majority of LC VFOs we were happy with a couple of decades ago. 

An interesting QRP Labs article by Gwyn G3ZIL compares the Si5351A with a crystal LO in a 40m direct conversion receiver see http://www.qrp-labs.com/synth/synthnoise.html

73 Hans G0UPL
13970 2017-06-05 00:11:17 Ashhar Farhan Re: Digest Number 2917
An indirect measure of how good the phase generation is the sideband suppression in a phasing receiver. i have just built a polyphase receiver based on hans' kit. it drops every signal down into the noise when tuned to the other side. 
i will push the two tone IMD signal gen into it this week and use the attenuator to measure the exact suppression. that would clearly show if enough is enough.

- f

13972 2017-06-05 07:48:13 aa0zz Re: Digest Number 2917
Thanks, Hans, for a very well done explanation.  Good info.
Yes, trade-offs are always part of the proper solutions.
-Craig, AA0ZZ