[Elecraft] Re: KX1 mixer discussion

[Geoffrey Mackenzie-Kennedy]

Karl Larsen <k5di at zianet.com> wrote:

>    Yes. The LO has to beefed up from 400 mV into 2k-ohms to something else 
> yet to be determined since the LO has to switch the FETs on and off. The 
> space problem is caused by the transformer, and the gain problem is due to 
> the H-mode having a loss of gain like -5DB.

Morning Karl,

The easy way to do this, and I am not thinking about a retrofit with no 
space, is to use a high speed CMOS divide-by-two flip-flop to drive the 
mixer FETs, the LO driving the flip-flop. The flip-flop provides the 50% 
duty cycle square wave outputs which the mixer wants as injection. A penalty 
which is not difficult to overcome is that the flip-flop has to be driven at 
twice the injection frequency required by the mixer. Colin G3SBI, who 
invented the H-mode mixer, came up in 2005 with a very simple addition of a 
pot and three resistors which allows fine tuning of the FET's switching 
point (time). A bonus when using a flip-flop is that it can be inhibited by 
noise pulses coming from a "noise receiver" tuned outside of the band when 
you want noise blanking without having to put a noise gate in the main 
signal path which might compromise dynamic range, and the blanker will not 
be affected by in-band signals. Certainly it punches holes in the signal 
like any other blanker, but after almost 10 years of use I am still happy 
with the system.

> Actually the noise figure of the IF stages is reduced by the loss through 
> the mixer. This is also a problem.

I think that we are saying the same thing!  Approximate overall noise figure 
= IF noise figure + mixer loss + preselector loss, all in dbs. In a HF 
receiver this is not a problem since IFs with a noise figure of 2db nominal 
and an IIP3 of  +35dbm or better have been around for many years, but they 
draw some power. At VHF a preamp would be required, or a mixer with gain if 
you can find one that has a low noise figure together with a high IIP3 and 
does not cost a fortune.

> Yes this is all true. I have heard the H-mode mixer has a poor noise 
> figure but it is the loss of gain that causes this.

> Nothing can be done without a lot of design work and breadboard building. 
> I don't see this happening.

It is true that the noise figure of the H-mode mixer is approximately equal 
to its conversion loss, and this would apply to a diode ring as well. 
Actually there has been a lot of work done with the mixer since it was first 
described in Radcom in the very early 90's, and much of the work done in 
Italy and Japan has been aimed at reducing the number of transformers and 
developing LO squarers. ARRL published details of the 'Triad' receiver using 
the mixer, and as far as I know 250 or more of the improved version named 
the CDG2002 have been built by hams worldwide.

73,
Geoff
GM4ESD