[hpsdr] mixers/QSD

[Lawrence Stoskopf]

> I've been doing some thinking about HORTON. 
> 
> It seems to me that the biggest problem with getting the QSD to run
> fast is finding switches/muxes that are fast enough.

I've commented on this before, but again my take: 

If you go back and look at the early papers on mixers: intermod and other 
problems occur two places: at the full switch on (not usually a problem as 
diode mixers are so overdriven that signal current is overwhelmed by the 
diode on current) and what happens at turn on and turn off.  In a diode, you 
have the signal running up and down on the LO current during switching, this 
is the most non-linear area and the signal is actually modulating the LO 
switching.  Thus the effort for multiple diodes, series resistors, etc. to 
allow for greater drive power with a steeper slope to switching before diode 
limiting.  My gut feeling (as an anesthesiologist!) is that we kind of/sort 
of skirt this with the FET/QSD type mixers in that the LO is not really 
directly in the signal path so they work great in the low freq applications. 
But I don't see any way to overdrive the switch with LO to steepen the 
transitions and at high frequencies, the whole waveform starts to look like 
a sine wave and worse...... 

Faster switches in switching time are the key.  I doubt switch on pass 
frequencies are a problem in the applications these are designed for and bus 
switching speeds are adequate also for the design applications.  As always, 
we are using these in what is called in medicine, "off label applications". 
For instance, we use Topamax, an antiseizure medicine in controlling certain 
types of pain, and indication never intended by the manufactured and 
patients often lose 20-30 pounds and come back for more, even though the 
pain is still there.  Enough of that. 

Somebody with a whole lot more background help me out on this.  I still have 
a couple of 7360 homebuilt mixer receivers that work in still another way 
with basically an electronic signal beam running Class A and being deflected 
by electrodes back and forth between two plates with no switching 
linearities involved (mostly).  But you can't get them in silicon and 
Gilbert cells aren't the same. 

N0UU