IPR-CIC Altera, IPR-CIC Datasheet - Page 28

• Manufacturer Part Number: IPR-CIC
• Description: IP CORE Renewal Of IP-CIC
• Manufacturer: Altera

Specifications of IPR-CIC

• Software Application: IP CORE, DSP Filters And Transforms
• Supported Families: Arria GX, Cyclone, HardCopy, Stratix, Stratix II
• Core Architecture: FPGA
• Core Sub-architecture: Arria, Cyclone, Stratix
• Rohs Compliant: NA
• Lead Free Status / RoHS Status: na

4–4
Hogenauer Pruning
Multi-Channel Support
CIC MegaCore Function User Guide
f
1
1
Output Rounding
For high rate change factors, the maximum required data width for no data loss is
large for many practical cases. To reduce the output data width to the input level,
quantization is normally applied at the end of the output stage. In this case, the
following rounding or saturation options are available:
■
■
■
■
These rounding options can only be applied to the output stage of the filter. The data
widths at the intermediate stages are not changed. The next section describes cases
where the data width at the intermediate stages can be changed.
Hogenauer pruning [Reference ] is a technique that utilizes truncation or rounding in
intermediate stages with the retained number of bits decreasing monotonically from
stage to stage, while the total error introduced is still no greater than the quantization
error introduced by rounding the full precision output. This technique helps to reduce
the number of logic cells used by the filter and gives better performance.
The existing algorithms for computing the Hogenauer bit width growth for large N
and R values are computationally expensive.
For more information about these algorithms, refer to U. Meyer-Baese, Digital Signal
Processing with Field Programmable Gate Arrays, 2nd Edition, Spinger, 2004.
The CIC MegaCore function has pre-calculated Hogenauer pruning bit widths stored
within the MegaCore function. There is no need to wait for Hogenauer pruning bit
widths to be calculated if Hogenauer pruning is enabled for a decimation filter.
Hogenauer pruning is only available to decimation filters when the selected output
data width is smaller than the full output resolution data width.
There are often many channels of data in a digital signal processing (DSP) system that
require filtering by CIC filters with the same configuration. These can be combined
into one filter, which shares the adders that exist in each stage and reduces the overall
resource utilization.
Truncation: The LSBs are dropped. (This is equivalent to rounding to minus
infinity.)
Convergent rounding. Also known as unbiased rounding. Rounds to the nearest
even number. If the most significant deleted bit is one, and either the least
significant of the remaining bits or at least one of the other deleted bits is one, then
one is added to the remaining bits.
Round up: Also known as rounding to plus infinity. Adds the MSB of the
discarded bits for positive and negative numbers via the carry in.
Saturation: Puts a limit value (upper limit in the case of overflow, or lower limit in
the case of negative overflow) at the output when the input exceeds the allowed
range. The upper limit is +2
n-1
and lower limit is –2
n
.
© December 2010 Altera Corporation
Chapter 4: Functional Description
Cascaded Integrator Comb Filters