[Elecraft] Receivers?

[Matt Maguire]

In the K3 schematic, I don’t see any quadrature path starting from the second mixer in the RX path — the two signals lines coming out of that 2nd RX mixer are a balanced (differential) output, not in-phase and quadrature signals. On the TX side, the DSP seems to produce I&Q outputs, but it seems the I signal is not used for anything and only the Q signal is used as an input to the TX mixer.

This is in contrast to the KX3 where there are I&Q signals used between true quadrature mixers and the DSP. The IF of the KX3 can be 0Hz (direct conversion), 8KHz (single conversion), or as you point out, in the case of SSB weaver demodulation the 0Hz mark can be placed in the centre of the signal passband, then the I/Q signals are passed through the optional roofing low pass filter before being unfolded/shifted in the DSP.

-- 
73 de Matt VK2RQ

On 12 décembre 2015 at 10:55:59 PM, David Woolley (forums at david-woolley.me.uk) wrote:

I don't think these descriptions are accurate, particularly the K2  
versus K3 one.  

The K3 has a double conversion superhet architecture, with an HF first  
IF and an extremely low second one. There is a selectable crystal  
filter (using commercial sub-assemblies) in the first IF, which provides  
coarse selectivity. The final IF processing is digital. There is a  
quadrature path starting from the second mixer, analogue at that stage.  
Combined with digital processing, this creates an analogue of a  
phasing design receiver to suppress the final IF image, rather than the  
audio image. As the signal continues in quadrature, the digital  
processing may also act analogously to a phasing receiver to do the  
final conversion and audio image stripping, but it may be that the  
internal logic is more complex than that - the fine details are a trade  
secret, although they may or may not have release information about that  
part of it.  

The K3X, for CW at least, implements a hybrid analogue/SDR direct  
conversion, phasing design. For SSB it may do the same, but it is also  
possible that it actually implements a final passband centre at 0Hz, and  
then does a final frequency shift to move the centre of the passband to  
the correct audio frequency (i.e. they could have implemented it as a  
single conversion architecture). Selectivity is provided entirely by  
digital processing.