欢迎访问ic37.com |
会员登录 免费注册
发布采购

PDSP16488AMAACBR 参数 Datasheet PDF下载

PDSP16488AMAACBR图片预览
型号: PDSP16488AMAACBR
PDF下载: 下载PDF文件 查看货源
内容描述: 单芯片的二维卷积器与积分行延迟 [Single Chip 2D Convolver with Integral Line Delays]
分类和应用: 外围集成电路
文件页数/大小: 30 页 / 238 K
品牌: MITEL [ MITEL NETWORKS CORPORATION ]
 浏览型号PDSP16488AMAACBR的Datasheet PDF文件第9页浏览型号PDSP16488AMAACBR的Datasheet PDF文件第10页浏览型号PDSP16488AMAACBR的Datasheet PDF文件第11页浏览型号PDSP16488AMAACBR的Datasheet PDF文件第12页浏览型号PDSP16488AMAACBR的Datasheet PDF文件第14页浏览型号PDSP16488AMAACBR的Datasheet PDF文件第15页浏览型号PDSP16488AMAACBR的Datasheet PDF文件第16页浏览型号PDSP16488AMAACBR的Datasheet PDF文件第17页  
PDSP16488A MA
BIT 7
This bit must be set if the pixel clock is greater than
20MHz. It disables the output and input time
multiplexing, and instead outputs the least signifi-
cant half of the 32 bit intermediate sum for the
complete clock cycle. When the gain control is
used, the output multiplexing will automatically be
disabled.
REGISTER C Bit Allocation
BIT
0
0
3:1
3:1
3:1
3:1
3:1
3:1
3:1
3:1
5:4
5:4
5:4
5:4
7:6
7:6
7:6
7:6
CODE
0
1
000
001
010
011
100
101
110
111
00
01
10
11
00
01
10
11
FUNCTION
Field selection defined by C5:4
Automatic field selection
DELOP = 29 + 0 clks
DELOP = 29 + 8 clks
DELOP = 29 + 16 clks
DELOP = 29 + 24 clks
DELOP = 29 + 32 clks
DELOP = 29 + 40 clks
DELOP = 29 + 48 clks
DELOP = 29 + 56 clks
Select upper 20 bits
Select next 20 bits
Select next 20 bits
Select bottom 20 bits
By-pass the gain control
Normal gain control O/P
Saturate at max + and -ve values.
Force -ve to zero.Sat.+ve values.
REGISTER B Bit Allocation
BIT
0
0
2:1
2:1
2:1
2:1
3
3
4
4
6:5
7
7
BIT 0
CODE
0
1
00
01
10
11
0
1
0
1
0
1
FUNCTION
Second line delay group fed from the
first group
Second line delay group fed from L7:0
which become inputs
Store pixels to end of line
Store pixels till count is reached
Frame store operation
Not Used
No delays on pixel inputs
4 delays on both pixel inputs
Use expansion adder
Expansion adder disabled
Not used
Use first delay in second group
Bypass first delay in second group
This bit defines the input for the second group of
line delays. It must be set in the 16 bit pixel modes,
and is set by power on reset.
These bits control the mode of operation of the line
stores. In real time systems pixels can be stored
either until HRES [ SYNC ] goes active , or until a
pre-determined count is reached. In the frame
store mode line store operations are continuous,
with a pre-determined line length.
When this bit is set four pipeline delays are added
to the pixel inputs to compensate for the internal/
external delays between line stores. The extra
delay is only necessary when a device supplied
with system video in which the first pixel in a line
is valid in the period following the first active clock
edge. See Fig 7. The delay is not necessary if the
device is fed from the output of another convolver.
When set this bit will add four additional delays to
those defined by Register D, bits 4: 2.
When this bit is set the expansion adder will not be
used. It is automatically set in a MASTER or SIN-
GLE device.
This bit controls the bypass option on the first line
delay on the L7:0 inputs. It is only effective when
an 8 bit pixel mode is selected, which also needs
more than four line delays. When L7:0 are used as
outputs it should always be reset. In the 16 bit
modes the bypass function is only controlled by the
BYPASS pin, and the bit is redundant.
BIT 0
If this bit is set, the 20 bit field selected from the 32
bit result, is defined automatically by internal logic.
BIT 2:1
BIT 3
BITS 3:1 These bits are in conjunction with Register D, bits
7:5 to define the pixel delay from the HRES input
to the DELOP pin. They are used to match the
appropriate processing delay in a particular sys-
tem. The minimum delay is 29 pixel clocks.
BITS 5:4 These bits define which of the four 20 bit fields out
of the 32 bit final result is selected as the input to
the gain control. They are redundant when the gain
control is not used, or if Register C, bit0, is set.
BITS 7:6 These bits define the use of the gain control as
given in the table. Intermediate devices in a mul-
tiple device system MUST by-pass the gain con-
trol, otherwise the additional pipeline delays will
effect the result. Disabling the scaler will reduce
the device pipeline by 13 PCLK cycles from the
delays shown in Table 4.
BIT 4
BIT 7
13