On 4/25/2010 11:08 PM, CSM(r) Gary Huber - AB9M wrote:
Is there any phase noise in an analog LC PTO? I believe the only digital
circuit in the Corsair II is the frequency counter.
Every oscillator has phase noise. In the analog oscillator it comes from
phase modulation of the sinewave by the 1/f and flicker noise of the
active device and is usually better with a bipolar transistor than a
FET. That's because a FET inherently has far greater 1/f and flicker
noise than a good bipolar transistor. Phase noise is least with a high Q
crystal in the best of circuits, like the HP 10811 oscillator. Phase
noise in a PLL can be high from multiple sources. First the inherent
phase noise of the wide range VCO is often large, made worse by a large
frequency agility per volt of control voltage that often means kHz shift
per microvolt and its really hard to keep analogue voltages noise free
at the microvolt level, just the random noise from resistances in the
circuit contribute much noise in the control loop. Then phase detectors
and dividers add phase noise. Amplifiers can add a little phase noise
too. Oscillators also have amplitude noise from the device noise, but
generally good mixers (double balanced) ignore amplitude noise so its
usually neglected. Oscillator design has to trade off power dissipation
and the resulting drift with signal to noise ratio in the oscillator.
Oscillators not in a phase locked loop tend to have rapidly decreasing
phase noise near the signal, often dropping 20 dB per decade of offset
until reaching levels lower than -120 dBc/Hz, then they begin to flatten
off with good oscillators reaching -160 dBc/Hz at MHz plus offsets. The
wide band phase noise tends to have a flat spectrum from the broad band
noise of the active devices. This spectrum was first characterized by
Leeson and so is referred to as a Leeson phase noise spectrum. A few
really fine oscillators reach -180 dBc/Hz but they are uncommon.
Wide range VCOs as used in some synthesizers can have open loop phase
noise spectra much like the best oscillators except 40 to 80 dB
stronger. Often they are controlled with a wide band (up to a MHz)
control loop that typically gives the PLL a flat spectrum out to the
control bandwidth, then it slopes down. There is much more power in that
shape of spectrum than in the Leeson spectrum.
In the perfect world phase noise increases when the signal is multiplied
by a factor of 20 log10 n where n is the multiplication factor.
Multiplying a 100 MHz crystal to 10 Ghz where the multiplier is 100,
increases the phase noise by 40 dB. That's a minimum increase, practical
circuits increase the phase noise more. Conversely, dividing reduces the
phase noise almost as much as 10 log10 n where n is the fraction. And
that is used in the Omni V and VI where the synthesizer runs in the 400
MHz region and is divided down to 5 MHz.
I claim, and there is beginning to be some evidence though I plan a
detailed experiment some day, that at frequencies where the outside
noise is not the limiting factor (UHF and higher) that the reciprocal
mixing of broad band RF circuit noise by the phase noise sidebands of
the LO affects the minimum discernible signal. This effect is less
significant than that of reciprocal mixing if there are any strong
signals in the RF input and I claim its more significant for narrower
bandwidths and lower noise figure systems. I have to prove that. In the
past year since I brought up that theory, W1GHZ has made MDS
measurements at 10 GHz comparing a crystal LO and a synthesized LO and
found a 2 dB difference in MDS. Since the test signal for noise figure
measurements is broad band (generally 10 MHz to 18 GHz or more)
comparing NF measurements for different levels of phase noise finds no
difference in NF. There are many articles talking about the MDS effect
in general terms but none put numbers to the effects. Not even all the
articles and books about phase noise by Ulrich Rohde put numbers to the
phenomena.
Reciprocal mixing from phase noise is very serious on the lower HF
bands, far more than the effects on MDS because those bands (especially
40m in Europe) have many more STRONG signals and nearly all receivers
are limited by antenna noise, not MDS. To add complications those strong
signals also have phase noise sidebands.
When I've had my Corsair II running with either a Paragon or a Omni VI+ as
its AUX RX, the Corsair II was always quieter. Weak signal detection is
about the same or too small a difference to hear. Only in close in (2 KHz)
weak DX, strong station (multiple S9+ signals) CW pileups does the OMNI VI+
prevail.
73,
Gary - AB9M
I had the opportunity to compare Corsair II, Cubic 103 and Omni V on the
same weak 10m signals with rapid antenna and radio switching in WA0ROI's
shack one day and the differences between Corsair II and Omni V were
barely detectable, less than changes in propagation minute to minute,
but the Cubic 103 heard better. I don't know how the Cubic does in
strong signal situations. Its rare and mostly unobtainable anyway. I
chose the Corsair II because for VHF operations it runs CW on USB on 28
MHz where the Omni V and VI run CW on LSB, copying a bad ICOM idea. That
makes it impossible to run mixed mode QSOs on VHF which is common in
contests.
73, Jerry, K0CQ
|