Re: [Discuss-gnuradio] Practicality of DC-light RF front-end?

[Marcus D. Leech]

Johnathan Corgan wrote:

> So, personally, I'd like to see an emphasis on this.  Maybe a separate
> "pre-selection" module that can carve up the input spectrum with a
> little more granularity than is usually the case would be the right way
> to go.

Something like:

http://www.propulsionpolymers.com/radioastronomy/widebandreceiver.pdf

It seems, though, that doing a filter bank for an entire 3000Ghz swath
  would be expensive, if done with high-quality in mind.  In the diagram,
  I show 4 filter control lines, giving 16 different filters--which gives
  an average bandwidth per filter of 187.5Mhz.  Above roughly 400-500Mhz,
  use of DRFs (dielectric resonant filters) becomes practical.
  DRFs are cheap, but only when made in largish quantities.
  Below 400Mhz, LC filters work reasonably well.  And there's microstrip
  filters as well, although physical line lengths get pretty large below
  1Ghz, even for folded resonators.

The diagram obviously is missing a LOT of detail, but it gives some of my
  thoughts.  Kind of a combined SSRP and RF front end.  I think it's entirely
  practical to build something like this onto a single 4 layer board without
  too much trouble.  The MAX2116 is a nice chip, but I couldn't find a 40-pin
  QFN package in Eagle, and I'm a total newbie when it comes to e-cad in
  general, and Eagle in particular.

My particular application is radio astronomy, so some design decisions might
  be made differently depending on the application.  Radio astronomers like
  lots of bandwidth for example (since sensitivity is proportional to
  both the instantaneous bandwidth, and integration time), but for a simple
  design like this, one is constrained in ultimate A/D--USB bandwidth.
  I haven't shown any filters in front of the A/Ds--I think that the
programmable
  baseband filters on the MAX2116 can take care of this.  But if not, you'd
  have to put an anti-alias filter in front of the A/Ds.  The I2C D/A shown
  is to control the gain of the MAX2116.  Irritatingly enough, the MAX2116
  splits gain control between an *analog* GC1 control, and a digital
  GC2 register that's programmed over the I2C.  The GC2 control is used to
  control baseband gain, while GC1 controls the gain of the front-end amplifier
  in front of the mixers.  For radio astronomy, the analog gain control has
  to be *very* stiff in relation to both temperature and power-supply
fluctuation
  effects.  In total-power radio astronomy, a gain change of 0.1dB is considered
  to be very large [and methods are used, outside the scope of this project, to
compensate
  for that].

The RF signal and input to the MAX2116 is controlled using PIN diodes--I haven't
shown
  the necessary detail, but the idea is that the FX2 can supply control signals
to
  tell the PIN-diode bias logic which diodes to turn on, and which to turn off
in
  order to route the various RF signals.  I've also shown a MMIC amplifier--to
  compensate for the loss through the PIN diodes, and the inevitable loss
through
  the filters [in fact, there should probably be one in front of the filter bank
  as well--at least for UHF signals].

The filter bank is, obviously, left rather vague indeed.  Again PIN diodes could
be
  used to switch the various filter banks in and out.  I've shown 4 control
bits,
  which would be enough to control 16 filters.  In reality, probably more would
  be needed, depending on filter design, etc.  The biasing arrangement for
  the PINs here would be the same as for the RF switching of the various
downconverter/
  upconverter signals.

For signals between 5-950Mhz, a simple single-conversion upconverter is used,
with a
  950Mhz HPF after the mixer.  Minicircuits makes some excellent mixers that
aren't
  that expensive.

For signals between 2000-3000Mhz, a simple single-conversion downconverter is
show, with
  a 2000Mhz LPF after the mixer.  Again, minicircuits mixers could be used to
good effect.


Ok, so who wants to go off and "code" this up in Eagle????? :-)

-- 
Marcus Leech                Mail:   Dept W669, M/S: 04352P16
Advisor                     Phone: (ESN) 393-9145  +1 613 763 9145
Internet & Security Services
Nortel Networks                          address@hidden