[time-nuts] 5>10 doubler

Charles Steinmetz csteinmetz at yandex.com
Tue Jan 27 19:57:42 EST 2015

Bruce wrote:

>Using the square law characteristic will inevitably increase the 
>phase noise floor particularly in the flicker region with respect to 
>just using the switching characteristic of a JFET, diode or 
>BJT  (non saturated).

Even FETs with very large transconductance and low pinchoff voltage 
do not switch cleanly in a push-push mixer -- there is always plenty 
of square-law behavior to add flicker noise.  So the difference is 
not really all that great.  Furthermore, in practice whatever 
difference there is will be swamped in many (if not most) cases faced 
by amateur time nuts by the much larger phase noise of their sources.

There are always tradeoffs.

>The only viable solution is to use better filtering of the output of 
>a switching multiplier.

For the reasons given above, I disagree that a switching multiplier 
is the only viable solution for amateur time nuts.  For NIST, 
possibly.  But not for amateur time nuts.  Given a dirty multiplier 
such as a push-push doubler with 2SK369s or BF862s, the only way to 
get the fundamental and distortion products at 30MHz and below down 
to suitable levels for critical work is to use a high-Q 10MHz filter 
plus a series of traps.  The high-Q 10MHz filter brings its own low 
frequency stability problems.

There are always tradeoffs.

>If you intend to use a diode ring based mixer configuration diode 
>connected (collector shorted to base) npns such as 2N222's are 
>significantly quieter (as shown by NIST) than schottky diodes for 
>frequencies below 40MHz or so.

Yes, that is well known.  However, the transistors need to be well 
matched or, once again, you end up with spurious products that are 
hard to remove without causing collateral damage.  (I speak from 
experience -- I have built mixers that way.)  This is mitigated to a 
large extent with matched diode rings as used in diode DBMs.  The 
advantage at 10Hz with 2N2222s is only 2 or 3 dB over a good Schottky 
ring, so unless one is comparing better sources than most amateur 
time nuts have access to, it will be lost in the source noise (note 
also that there is a crossover, above which the 2222s are about 10dB 
worse).  Given the relatively small potential gain and the effort 
required to match four 2N2222s to sufficient accuracy not to do more 
harm than good, I believe this is an exercise best left to those who 
know they need the modest advantage (again, not many of us amateur 
time nuts).  Once upon a time, Motorola made monolithic quad BJTs -- 
but I'm not aware of any matched quads similar to 2N2222s at this 
time.  [Several manufacturers still claim to make MPQ2222s, but those 
are not monolithic and not matched.]

There are always tradeoffs.

Notwithstanding theoretical objections, the described circuit works 
very nicely and performs substantially better than most time nuts 
need.  I presume that when one is willing to afford sources that 
require better performance, one will (1) know that, and (2) also be 
willing to afford an expensive doubler that meets that need.

Best regards,


More information about the time-nuts mailing list