Bob says:
>What about feedforward???
For HF, it has some interesting problems in getting the correct phase match over
the whole spectrum. Inherently, that also requires accurate matching in the time
delays, too - effectively putting a very tight differential group delay spec on
the system. Is the improvement really worthwhile for the amateur service,
though?
How far down does the IMD need to be? Realistically, probably -36dB on PEP is
adequate, purely because of limitations in drivers and in the quality of the
receiver that is operating in the adjacent channel.
On that basis, the pair of 3-500s or 3CX800 or whatever are likely to do all
that's needed, quite possibly with the IMD defined more by the driving
transceiver. If we look at solid state amps, feed forward doesn't allow really
heavy clipping to be cancelled, so we're not going to win much in output power,
only in IMD at a given output. If we drop the output of a conventional amplifier
by 3dB, we may gain 6 -12 dB in IM ratio, depending on what the actual transfer
characteristic is, and where the gain compression point falls. Feed Forward
amps see a lot of use in multi-carrier applications, such as cellular, where IM
ratios of 70 and 80dB and more are required. Those applications also in practice
tend to reach the full theoretical PEP only occasionally because of the
statistical nature of the multiple signaIs applied (see the paper 'Power Rating
of Multi-Channel Linear Amplifiers' by Bennett, Kenington and Wilkinson, given
at the 7th IEE European Conference on Mobile Personal Communications, Brighton
UK, December 1993, IEE Conference Publication no 387, pp 26 - 30). I suppose
that 3dB is worth having, since the error amplifer is perhaps one fifth or one
tenth of the power of the main amp, but I have worries about the relatively wide
bandwidth required in the phasing and delay networks for HF - basically 4
octaves. FF amps seem to have most use in fractional octave applications.
It could be a very useful approach (and probably has been used) for receiver
multi-couplers.....especially if you did a band segmentation to get over
differential phase problem.
The Cartesian or Polar Loop approaches seem to my mind to offer rather more at
HF, especially if a high efficiency amplifier is required. There's also the old
Doherty approach......
73
Peter G3RZP
--
FAQ on WWW: http://www.contesting.com/FAQ/amps
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
|