Hardware Model - Analog Devices SHARC ADSP-21368 Hardware Reference Manual

Hardware Model

The output rate of the low-pass filter of
× 192 kHz = 201.3 GHz. Sampling at a rate of 201.3 GHz is
20
rate, 2
clearly impractical, not to mention the number of taps required to calcu-
late each interpolated sample. However, since interpolation by 2
involves zero-stuffing 2
the multiplies in the low-pass FIR filter are by zero. A further reduction
can be realized by the fact that since only one interpolated sample is taken
at the output at the f
formed per f
S_OUT
for each f
S_OUT
interpolation.
The difficulty with the above approach is that the correct interpolated
sample needs to be selected upon the arrival of f
possible convolutions per f
must be measured with an accuracy of 1/201.3 GHz = 4.96 ps. Measuring
the f period with a clock of 201.3 GHz frequency is clearly impossi-
S_OUT
ble; instead, several coarse measurements of the f
made and averaged over time.
Another difficulty with the above approach is the number of coefficients
required. Since there are 2
ter, there needs to be 2
requires a total of 2
ROM, the SRC stores a small subset of coefficients and performs a
high-order interpolation between the stored coefficients. So far the above
approach works for the case of f
the output sample rate, f
the ROM starting address, input data, and the length of the convolution
must be scaled. As the input sample rate rises over the output sample rate,
the anti-aliasing filter's cutoff frequency has to be lowered because the
Nyquist frequency of the output samples is less than the Nyquist fre-
quency of the input samples. To move the cutoff frequency of the
ADSP-21368 SHARC Processor Hardware Reference
Asynchronous Sample Rate Converter
20
– 1 samples between each f
rate, only one convolution needs to be per-
S_OUT
period instead of 2
sample is sufficient to suppress the images caused by the
period, the arrival of the f
S_OUT
20
possible convolutions with a 64-tap FIR fil-
20
polyphase coefficients for each tap, which
26
coefficients. To reduce the amount of coefficients in
S_OUT
, is less than the input sample rate, f
S_OUT
Figure 10-2 is the interpolation
S_IN
20
convolutions. A 64-tap FIR filter
. Since there are 2
S_OUT
clock period are
S_OUT
> f
. However, in the case when
S_IN
20
sample, most of
20
clock
S_OUT
,
S_IN
10-7