Ian asks:
>> A +30dBm intercept point is no use unless the phase noise is at
least
>> -130dBc/Hz at the offset you're using.
>
>Why?
Because if the phase noise performance is poor, then signals that produce an
IMD product have produced a phase noise output as well, that may be bigger
than the IMD product. See my IEE paper of Feb 96 (available electronically
as a Word attachment for anyone that wants it). Phase noise limited dynamic
range and IMD limited need to be about the same. Assume an SSB bandwidth and
a 3kHz band width to get an rx noise floor of -130dBm, and the rest of the
figures drop out.
The assumptions made and not stated (something you shouldn't do on this
reflector!) are that we are talking about an SSB bandwidth, 10dB noise
figure, and HF operation.
>Would you agree that "sensitivity" of a system is defined by
antenna gain >and receiver noise power ratio (noise figure)?
Not entirely in this particular context, because at HF, much of the time,
the rx noise figure contributes very little to the noise power.
> Or is the noise figure changing by 3 dB, methinks not?
Again, we're talking HF, where the received noise is relatively high.
Putting in a 3dB pad, while operating without AGC, and measuring the audio
output power while running at low enough level that the rx isn't gain
compressing (ore caveats!) should show about 3dB drop in noise. If it drops
a lot more, then it's IMD noise. If it drops less, you may be running into
rx noise figure becoming significant.
>Do you mean that HF receivers have far more gain than they need, as
the
>active stages are driven into gain compression, hence the use of
the
>attenuator?
Not necessarily - they may (many do) have it in the wrong place. The gain
needed for 10meters when the band is on the way out may well need all the
gain you have. But for 80metre operation, you don't need a 10dB noise
figure. It's instructive to look at the ITU curves for noise levels against
frequency. But if the 'dynamic range' is fixed by phase noise rather than
IMD, an attenuator can help move the dynamic range to the correct starting
point. I use to feel that antenna attenuators were an admission that you
can't design a front end properly; now I'm not quite so sure - possibly that
you can't design a good enough oscillator. A really good sig gen for phase
noise is the pre WW2 Marconi TF144; high Q tank, triode pushing a couple of
watts input. It drifts like Kon Tiki, pulls like a Union Pacific Big Boy,
has more FM when modulated than KFOX, gives hernias when you lift it. But
it's good on phase noise.......
>How does one jitter discrete spurs without introducing the mother
and
>father of FM onto the carrier?
Not sure - something to do with the truncation in the DDS. I read about it
somewhere, but can't remember where.
>High power keeps down phase noise?
Yes. At room temp, the maximum C/N of an oscillator with 0dBm in the tank is
174dBc/Hz. With 30dBm, it's 204dBc/Hz. One of the problems with modern
cellular radio is that they want minimum power consumption, so want the
oscillator at -10dBm. The CNR is given by:
-174 + 3 + NF + Pt dB
where -174 is KTB, the 3dB is a 'fudge' factor to allow for two sidebands,
NF is the noise figure of the active device, usually much higher in the self
limiting mode of an oscillator, and Pt is power in the tank circuit in dBm.
This is a mdifcation of Leeson's (ex W6QHS, now W6NL) equation. From a point
offset in frequency defined by f/Q (working frequency divided by working
tank circuit Q), phase noise rises from this level at 20dB/decade until the
1/f noise knee of the device is reached: noise then rises at 9dB/octave,
steepening as other noise mechanisms come into play. There's an upper limit
in frequency, determined by Ft of the active device ( although some argue
that it's determined by Fmax)
>Varactors are bad news in oscillator circuits, one is enough
surely?
In chess, you call this a 'zugswang' - between the devil and the deep blue
sea.
2 varactors were used in series back to back to prevent forward conduction
with low control voltages. The extra ones in parallel were to give the
required capacitance swing with the available silver on glass inductor.
Measurement was at a spot frequency using a 141 spectrum analyser with a
crystal filter in front of it that had a 90dB stop band.
73
Peter G3RZP
--
FAQ on WWW: http://www.contesting.com/ampfaq.html
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
Search: http://www.contesting.com/km9p/search.htm
|