Peter,
You are certainly right...
Making a medium to high power HF transmitter with "close-in" noise sidebands of
-160 dBc/Hz would be a very difficult task indeed.
It is not only the spectral purity of all oscillators involved that count, but
also the aggregated noise properties of the amplifiers and level control loops.
Having been deeply involved with military "co-located" HF systems in the past,
I fully appreciate the efforts put into "high-end" HF hardware by their
designers,
i.a. the necessity of building the input end of an HF power amplifier in the
same way as a receiver low-noise amplifier, not to degrade the overall SNR.
Despite these efforts, it is still uncommon to find better noise floors than
-150 dBc/Hz at 20 kHz spacing
for "production" HF equipment.
An account of the design philosophies required to approach these levels of
performance can be found
in a paper "A New Generation of HF Power Amplifiers" by K G Nygren in the
Proceedings of the Nordic HF Conference HF92, which goes into the design
considerations for the ITT-Standard Radio SSA1020 1 kW MOSFET amplifier.
Having accomplished transmitter and receiver sideband noise floors of -160
dBc/Hz, at, say, 10 kHz spacing, the next major task would be finding a way of
reducing the IM sidebands of tunable SSB transmitters to corresponding
levels....
73/
Karl-Arne
SM0AOM
----- Original Message -----
From: "Peter Chadwick" <g3rzp@g3rzp.wanadoo.co.uk>
To: "Harold Mandel" <ka1xo@juno.com>; <g3rzp@g3rzp.wanadoo.co.uk>; "'Frank
Goenninger'" <frgo@mac.com>; <amps@contesting.com>
Sent: Sunday, July 23, 2006 9:14 AM
Subject: Re: [Amps] 3cx3000A7 driver
> Hal said:
> >This radio is built for the future.
>
> In tomorrow's world -160dB(c) per cycle may actually be
> a force to contend with......<
> It may be, although I have severe doubts about the technology needed. This is
> because, in the final analysis, low phase noise requires high power. S/N
> becomes osc power over thermal plus implementation noise. Modern IC
> technologies are pushing to smaller geometries, which mean lower voltages and
> so SNRs are tending to drop. Implementing a DDS to go fast without needing a
> BIG heatsink (e.g. the old Plessey SP2001 500MHz ECL DDS) becomes a bit of a
> difficulty - you need to do it in CMOS. But reduced geometry CMOS has reduced
> SNR, because of the lower voltage swings. It also has leakage problems, which
> can also degrade SNR, as well as lead to higher current consumption - which
> is why in the company I work in, we're still pretty well confining ourselves
> at the moment to using 0.18 micron CMOS, especially as we need RF analogue
> characterisation. In any case, until everyone else on the band has rigs of
> equal performance, there's not that much advantage, and as far as IMD goes, f
or
> the majority of people, need. See my 2002 QEX article.
>
> 73
> Peter G3RZP
> _______________________________________________
> Amps mailing list
> Amps@contesting.com
> http://lists.contesting.com/mailman/listinfo/amps
>
>
> --
> No virus found in this incoming message.
> Checked by AVG Free Edition.
> Version: 7.1.394 / Virus Database: 268.10.3/395 - Release Date: 2006-07-21
>
>
--
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.394 / Virus Database: 268.10.3/395 - Release Date: 2006-07-21
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|