Topband: RX 4 square

[Tom Rauch]

> This is only true if the bandwidth is determined in the 
> detector
> as happens with a DSP demodulation.  Since most amateur 
> transceivers
> set the demodulated bandwidth with a crystal or mechanical 
> filter
> early in the IF, the noise bandwidth is effected by the 
> noise of
> the IF system (particularly the last IF stage) and is 
> independent
> of the RF "window."  The effect is more significant with 
> weak
> signals as the IF stages will tend to run "wide open" and 
> wideband
> noise will be maximum.

That's true, and is often why you need a bit more headroom 
than expected with narrow filters.

The major point is the narrow bandwidth means you must have 
more gain and that additional gain must be added in a way 
that, if it does anything, improves the noise figure.

As I recall (but I might be mistaken) the original post 
implied there wasn't any gain difference, but there is a 
huge difference between a voltage follower and a 
transformer. 20-30dB typically, and that's a heap of gain to 
make up later without hurting NF.

I'm sure there are cases where a transformer works, but in 
our applications (especially with a compact directional 
array) that 20dB or more gain is almost always necessary. Of 
course you could get some of it later.

There are often many ways to do the same thing and no single 
best choice when it comes to something like this.

73 Tom