To decode the High-SNR format when the exponent is 0, interpret the mantissa as a 12-bit
signed integer and multiply by 2**-24.
A complete analysis of the noise properties of the floating point codes would be fairly tricky.
For the High-SNR format, the 12-bit mantissa with hidden normalization bit vary from
2048 ... 4095. The SNR therefore varies from 66 dB ... 72 dB and we can assign a mean value
of 69 dB. Another 9 dB of useful range is contained within the code as follows:
• In a
floating point encoding format, the notion of fixed additive quantization noise is
not really correct. For a signal having a given power, the additive noise within each
instantaneous sample scales down according to the magnitude of that sample. The
ensemble of noise terms thus contributes an RMS power that is smaller than the Peak-
to-Noise ratio would imply. In the case of a sinusoidal input, this gives a 3 dB boost in
eective SNR.
• The format, of course, also represents negative amplitudes with the same relative
precision as positive values. In a fixed-point format this would add 6 dB (one more bit)
to the overall dynamic range and large-signal SNR. In the
floating format we really only
gain 3 dB (half a bit) because the RMS noises add independently on the positive and
negative excursions.
• The packed format is used to encode time series (I,Q) pairs, and it's the SNR properties
of these pairs that we're really concerned about. To a
first approximation, having a pair
of values roughly doubles the information content and adds another 3dB to the SNR.
The last of the time series output words, the Log of Power in Sample, is provided for
backwards compatibility. It can be calculated from the I and Q numbers. To convert to dBm
it requires a slope and oset as follows:
=
+ × 3584
where:
P
MAX
+4.5 dBm for 12-bit IFDR, +6.0 dBm for 14-bit IFDR, +8.0 dBm for 16-bit IFDR
V
MAX
0.5309 V for 12-bit IFDR, 0.6310 V for 14-bit IFDR, 0.7934 for 16-bit IFDR
Slope
Log Power Slope word 3 of the SOPRAM command. 0.03 is recommended.
For backwards compatibility, RVP900 produces a 8-bit fixed point time series format.
Because of the limited dynamic range available, this only shows strong signals, and is not
recommended for use. The I, Q, and Log power triplets are packed in two 16-bit output
words as follows:
RVP900 User Guide M211322EN-J
266
To Next Page
Loading...