[Lowfer] LOWFER receiver conundrum.

[JD]

Sounds like you're reviewing all the essential issues pretty carefully, Ben.

I'd forgotten about the Si570 kit...had looked at it one before, but wasn't 
able to obtain the IC data sheet at that time, so it slipped from my mind. 
Your mention prompted me to look again, and I also found Jack Smith's 
detailed review at Clifton Labs:
http://www.cliftonlaboratories.com/si570_kit_from_k5bcq.htm
Its spectral purity is better than typical affordable DDSes, and with the 
K5BCQ kit there's little or no additional circuitry to mess with.

One thing I notice, though, is that the phase noise (while better than many 
alternatives) is still not ideal very close-in to the oscillator frequency. 
The closer you plan to tune to zero frequency, the tighter the phase noise 
must be controlled, or you will detect a significant noise increase as you 
tune lower and lower.

For my own future receiver design, I've set a target of -130 dBm for Minimum 
Discernible Signal level in CW bandwidth at a dial frequency of 10 kHz.  I 
realize that's a bit arbitrary, as local noise sources are going to be 
significantly stronger than that with my current antenna setup, but it's my 
goal from an engineering aesthetic standpoint.  To achieve it, I will need 
an oscillator that exhibits better than -155 dBc/Hz phase noise (rough 
offhand estimate) at 10 kHz from the LO center frequency.  That's a 
challenge even for a well designed crystal oscillator, and the Si570 doesn't 
come anywhere close to that around its output frequency.  Of course, for 
VLF, the Si750 can be operated at a high multiple of the desired LO 
frequency and divided down to it, shrinking the noise sidebands as well. 
That's what enables the Softrock to do a credible job.  But there are limits 
to what frequency division can accomplish, as the dividers introduce a 
little phase jitter of their own.  Down around 10 kHz dial frequency, I'm 
not sure how well it will do.

The other thing about the Si570, unfortunately, is that the chip that comes 
with the kit is only rated +/- 50 ppm stability over the operating 
temperature range.  The reference oscillator is built into the IC and you 
can't override it.  While it might be feasible to stabilize the entire 
oscillator assembly reasonably well by enclosing it in a sorta-oven, it'll 
never be possible to lock it to an external reference.  That's not 
necessarily a deal breaker for QRSS, WSPR, etc., but would be an issue for 
slow bitrate BPSK.

Just a couple of things to factor into the decision.

John