Rohde & Schwarz R&S FSL3 - Oversampling

Bluetooth Measurements (Option K8) R&S FSL
1300.2519.12 2.42 E-11
the displayed deviation value is increased by 3.2%, but without the filter the displayed deviation value
c
an increase dramatically due to interference from adjacent channels. Generally the result is more
precise, if the displayed deviation is lower with filtering than without filtering. In these cases the
inaccuracy caused by the adjacent channel interference is higher than the systematic inaccuracy
caused by the filter.
0
-30
-50
-10
-2.0 -1.6 -1.2 -0.8 -0.4 +2.0+1.6+1.2
dB
-60
-20
-40
MHz
+0.8+0.40
-60
-40
-20
0
Fig. 2-30 Selection of digital filters
Dashed–dotted curve: Standard filter with 4 points / symbol
Solid curve: Optional measurement filter, independent of the points / symbol setting
Oversampling
The number of samples per symbol is equivalent to the sampling rate in MHz (due to the symbol length
of 1 Rs).
Digital
bandwidth
(flat area)
Points per
Symbol
Sampling rate
10 MHz 32 32 MHz
8 MHz 16 16 MHz
5 MHz 8 8 MHz
3.0 MHz 4 4 MHz
1.6 MHz 2 2 MHz
According to the RF Test Specification an oversampling factor of 4 is required at minimum. For Basic
Rate measurements, this oversampling factor can be selected as oversampling factor in a range from 2
to 32. For EDR–measurements, the oversampling factor is fixed to 4 which is also the default value.
Although possible but not recommended is a value > 4. It increases the measurement time due to the
extended calculation effort. Additionally the resulting bandwidth will be larger than required, which leads
to lower measurement accuracy, unless the optional measurement filter (Meas Filter On) is used as
described in section Bandwidths.
The spectrum analyzer uses a timing offset correction in order to move the samples to the zero
trespassing points. As a result there is one sample per symbol time, which is especially important for a
0101 symbol sequence in order to obtain the precise value for the peak frequency deviation.