[Discuss-gnuradio] 3rd-order IMD products

[Marcus D. Leech]

Here's a snapshot of the 3rd-order IMD products out of a WBX

http://www.sbrac.org/files/3rdorder.png

It shows the 3rd-order products at roughly -50dBc, which is extremely good,
when compared against many of the commercial amateur-radio transmitters
in surveys like this one:

http://www.remeeus.eu/hamradio/pa1hr/productreview.htm

Exact results would be frequency and daughter-card dependant, obviously.

Note that the digital (baseband) gain is set at -13dB, above this, the
IMD products start to
  creep up in strength, but the fundamental tones themselves don't get
any stronger.

The point at which the analog chain starts to become non-linear will be
different on different daughtercards,
  at different frequencies. It's *not* safe to assume that a baseband
magnitude of "1" represents the
  maximum still-linear signal power. Indeed, the full-scale output of
the ADC on (for example) an N210
  corresponds to roughly 0dBm. Various daughtercards have up-converters
(quadrature mixers) with
  various conversion gains ranging from about 3dB to about 5dB. On the
WBX, this drives a GVA-84+
  amplifier, whose p1dB point is at about 20.5dBm, and whose gain is
21.5dB at 1GHz, so with full-scale
  output through the mixer, the GVA-84+ would be operating well past the
P1dB point, and be operating
  in increasingly non-linear territory.

The RFX2400 has a mixer with similar conversion gain to that on the WBX,
but with two-stages of
  TX-side RF gain, with an RF3315 as the last stage. P1dB for the RF3315
is at about +23dBm. But
  with two stages of gain on the RF output side of the mixer, I would
expect the RFX2400 to hit
  non-linear operating conditions sooner than on the WBX.

So, to recap. A signal magnitude of 1.0 out of UHD doesn't mean "maximum
RF output power while
  still maintaining excellent IMD performance". It simply means
"full-scale output from the DAC".
  The extent to which full-scale output from the DAC drives the analog
daughter-cards into
  non-linear operating territory is highly dependant on the specific
daughtercard, the frequency
  of operation, and the TX gain settings on the daughter-card. Because
of conversion gains in the mixers,
  and generous post-mixer gain, backing-off from 1.0 baseband magnitude
doesn't necessarily
  mean any significant sacrifice in RF output power.

Anyone who cares to can download the daughter-board schematics for
themselves, download the
  appropriate data sheets from the manufacturers, and confirm all of
this for themselves.

-- 
Principal Investigator
Shirleys Bay Radio Astronomy Consortium
http://www.sbrac.org