EMRFD Message Archive 515

Message Date From Subject
515 2007-03-12 06:51:55 Hans Summers Re: H-Mode Mixer as tested by PA3AKEuls
Allison,

this isn't strictly true. if you use the usual kind of XOR doubler in
which the incoming signal (non-square-wave) edges are detected to
produce a 2f signal, then you have a problem.

it doesn't matter that the 2f is not 50 percent duty cycle, This
problem is removed by the divide-by-2. but it DOES matter if you just
detect edges such that the 2f signal isn't even regular. imagine the
original signal is 75:25. when you detect the edges, you get
pulse,pulse wait... pulse,pulse wait... and so on. you have the right
number of pulses for 2f, but they aren't regular.

divide-by-2 brings you right back to 75:25.

so: you need to some LC filtering to get a clean 2f sinewave. only
then divide by 2.

phase noise reduces by 6dB when you divide by 2 which won't compensate
for the degradation an RC type delay will cause.

73 Hans g0upl
http://www.hanssummers.com



Also you don't care that the doubled signal is asymetric as the
following /2 removes that. the idea is to get around the need
for an oscillator at 2x (or higher). Also any amount of division
reduces the phase noise as well.

Allison

>
> ----- Original Message -----
>
517 2007-03-13 05:19:22 Allison Parent Re: H-Mode Mixer as tested by PA3AKEuls
518 2007-03-14 07:29:14 Hans Summers Re: H-Mode Mixer as tested by PA3AKEuls
> > it doesn't matter that the 2f is not 50 percent duty cycle, This
> > problem is removed by the divide-by-2. but it DOES matter if you just
> > detect edges such that the 2f signal isn't even regular. imagine the
> > original signal is 75:25. when you detect the edges, you get
> > pulse,pulse wait... pulse,pulse wait... and so on. you have the
> > right number of pulses for 2f, but they aren't regular.
> >
> > divide-by-2 brings you right back to 75:25.
>
> I suggested the XOR as some peple try to use digital functions for
> analog work and not always with the best result. With a suitable
> delay line 50% is certain.

Yes but such "suitable delay line" would then become the determining
factor in the duty cycle of the final divided-by-2 result. Which would
mean it would only be exact at one particular frequency. To me this
sort of removes the whole point of doubling and dividing by two in the
first place.

50% is not a requirement at the 2f frequency. What IS important is
that the pulses at 2f are regular, and this can only be ensured using
an XOR doubler if the original VFO is a precise 50% squarewave - which
we already know it isn't, or we wouldn't be trying to fix it.

> A simple push push diode doubler (EMDRF) and a buffer would be
> cleaner and achive the same result simpler.

Yes... some things better done in analogue!

> > so: you need to some LC filtering to get a clean 2f sinewave. only
> > then divide by 2.
>
> That helps remove some jitter, but not much.

Surely that depends on how much filtering is done? It wasn't jitter I
wanted to remove, but unwanted frequency components. Ok, same
difference, if it's a digital signal. But if filtering a 2f signal
from a doubler, I think it should be possible to properly remove
enough harmonic (and other) components.

Having said all that - I still think the fundamental mode "squarer"
has a lot to be said for it! A very neat circuit! I like the way it
steers itself onto 50/50.

73 Hans
http://www.hanssummers.com
519 2007-03-14 10:54:30 Allison Parent Re: H-Mode Mixer as tested by PA3AKEuls