Red Wave Radio

Ive built a 20 meter VXO an am using it to drive a sbl-1 LO to produce
DSB. The VXO output waveform is not the best, in spite of all my
efforts.  The question is, how important is LO purity into these diode
modulators.  Should I lowpass filter the VXO output before feeding the
sbl-1.  Does it matter since the diodes switch on and off anyway?

Gary

W4AF

hello
a DBM must have a perfect purity output , ratio 50/50 square or sinusoidal
the output filter is for avoid harmonics and not avoiding spurius due to
bad ratio
excuse for approximative english or US language
Serge

The distortion you are referring to is a symptom of a problem with the
oscillator, and probably means the signal has excessive sideband noise
that will affect its operation more significantly than its harmonic
content.

The output of an oscillator grows until it hits a maximum limit.  Most
simple bipolar oscillators limit out by hitting a voltage saturation
point.  From what I read, this mechanism causes undue sideband noise due
to a temporary drastic reduction in Q as it hits this voltage saturation
condition.

For bipolar oscillators, current limiting operation is much more
desirable, as it maintains the Q of the resonator.  Oscillators with
this behavior take some design expertise to make sure it works right and
tend not to be the simple, straight forward, single transistor designs
that are most often seen.

Simple FETs oscillators tend to work much better than bipolar because of
a diode that is often used at the gate.  This diode allows a charge to
build up on the gate capacitor as the oscillation increases, reducing
the gain of the FET until the ideal loop gain of one is achieved.  The
use of the diode hurts the sideband noise a little bit, but the
resulting "gain limited" oscillator signal tends to be much cleaner than
than the output of the correspondingly simple bipolar oscillator.

Bipolars can do very well, but usually require more design expertise to
make sure it is operating under the correct conditions for optimum
results.

The ARRL book "Radio Frequency Design" spends a lot of time on the ins
and outs of various flavors of oscillators, including VXOs and is very
recommended reading.  You do not need to understand everything that is
written there to take away some very useful information.

Beyond the sideband noise issue, any mixer will generate a lot of
harmonics on its own as part of the switching/mixing process, so the
harmonic content is not as much of an issue as sideband noise.  If
nothing else, a low pass filter can be used to clean a harmonic problem
up especially if the signal has a symmetry problems.

- Dan, N7VE

Quote:> The distortion you are referring to is a symptom of a problem with the
> oscillator, and probably means the signal has excessive sideband noise
> that will affect its operation more significantly than its harmonic
> content.

There are two distinct issues here:  harmonic content and phase noise.
I believe the original poster was talking about harmonic content.  In
a crystal oscillator, the close in phase noise is normally determined
by the intrinsic noise of the crystal, and the far out noise floor is
normally determined by the buffer amplifier following the oscillator.
Hence, oscillator design has little effect on phase noise in general,
and in any event, even the worst crystal oscillator still has phase
noise that is quite adequate for a 20 meter radio.  It is far more likely
that if there was a phase noise problem, it would be due to noise
on the tuning voltage of the varactor used to pull the oscillator.

Harmonic content in oscillators can frequently be minimized by
taking the output through the crystal.  For example, if the oscillator
is designed so that one terminal of the crystal is grounded, then the
connection from the crystal to ground can be replaced by a grounded
base amplifier, which has a few ohms of input impedance.  The crystal
then acts as a crystal filter.

Quote:> Beyond the sideband noise issue, any mixer will generate a lot of
> harmonics on its own as part of the switching/mixing process, so the
> harmonic content is not as much of an issue as sideband noise

Here is a frequent misconception we see on the net.  For optimum
performance, a mixer such as the SBL-1 should driven with a sine
wave LO.  Any even order harmonics in the LO unbalance the mixer
and degrade IMD performance.  Mixers don't generate even harmonics,
so the argument that "any mixer will generate a lot of harmonics" doesn't
apply.  So why not drive the mixer with a square wave?  Well, first
of all, you have to guarantee that the square wave is exactly "square"
ie, it doesn't contain even harmonics.  Even if that constraint is met,
you then have the problem of getting the square wave to propagate
through the balun transformer in the mixer without distortion.  The
phase shifts at different harmonic frequencies will in general be different,
so the diodes will not see a square wave, but rather a distorted mess.

Quote:> If nothing else, a low pass filter can be used to clean a harmonic problem
> up especially if the signal has a symmetry problems.

Now here is some good advice.  Simply put a low pass filter
between the VXO and mixer.  You will find that the mixer
works best if filter element nearest the mixer is a series
inductor, rather than the more common shunt capacitor.
A little known fact.

Rick   N6RK
www.n6rk.com

Quote:> ... You will find that the mixer
> works best if filter element nearest the mixer is a series
> inductor, rather than the more common shunt capacitor.
> A little known fact.

An explanation, please?

73, John - K6QQ

Quote:>Ive built a 20 meter VXO an am using it to drive a sbl-1 LO to produce
>DSB. The VXO output waveform is not the best, in spite of all my
>efforts.  

Did you measure the VXO waveform driving a resistive load or an actual
SBL-1 mixer ? The mixer LO port does not necessarily represent a clean
constant resistive load during the whole LO cycle during the diode
switching periods. In the worst case you may have a variable
resistance and/or variable reactance in parallel with the frequency
determining reactances of the oscillator, causing waveform
deterioration and phase modulation (phase noise). A buffer amplifier
or at least an attenuator pad will reduce the effects of this variable
load to the oscillator.

Paul OH3LWR

Yes  ive terminated it with 50 ohms ....it looks ok,  but i was
expecting a perfect sine wave out of a crystal osillator.  Its a
2n2222 osc lightly coupled to a fet into another 2n2222 with a tuned
output and link coupled into the sbl-1.  I was putting about 13dbm
into the LO input, ive padded this down to 7dbm and now the sbl-1
output looks much better, and looking at the LO input with the scope,
it looks much cleaner.  I read you could put more than 7dbm in, but
when I do, with no IF modulation I see a ragged approximation of the
LO on the output.  Im building a phasing ssb rig.  Got the two sbl-1s
feeding a psc-2-4.

Gary
W4AF  

I'm not sure what you expect to see. When you're properly driving an
SBL-1 you should see a pretty square wave at the LO port, and a chopped
looking waveform at the output. Most of the buffers I use following an
oscillator have an output impedance considerably lower than 50 ohms, so
I generally connect a 47 ohm resistor between the buffer output and the
mixer. The buffer output usually looks fairly clean, but the SBL-1
input, on the other side of the resistor, is squared off due to hard
diode conduction. The mixer needs to be driven hard enough to cause the
diodes to switch, for proper operation.

Roy Lewallen, W7EL

Quote:> > ... You will find that the mixer
> > works best if filter element nearest the mixer is a series
> > inductor, rather than the more common shunt capacitor.
> > A little known fact.

> An explanation, please?

> 73, John - K6QQ

It has to do with the termination of the harmonics.  A ***
output filter (recommended) terminates the harmonics in
an open circuit; ie it behaves as a current source.  A capacitor
output filter terminates the harmonics in a short circuit; ie it behaves
as a voltage source.  In the former case, the filter is forcing
a sine wave current through the diodes.  The time varying impedance
of the diode is proportional to the logarithm of sin wt, which is
a well behaved function that approximates an ideal switch.  In the
latter case, the filter is trying to force a sine wave voltage across
the diodes.  In this case, the time varying impedance of the
diode is proportional to the anti-logarithm of sin wt.  This is
NOT a well behaved function and it does NOT approximate an
ideal switch very well.  Hope that helps.  It's a little hard to
explain, but if you look at the voltages and currents in a real
circuit, it is really obvious which version works better.

Rick N6RK

Thanks Richard; that is something that I wasn't aware of.

Pete