YOKOGAWA DL750 - Page 718

App-45
IM 701210-06E
Appendix
App
Computation Flow and Internal Computation Format of DSP Channels
The computation flow of DSP channels is indicated below. The input and output of DSP
channels are 16-bit binary data (if the input is 12 bits, it is converted to 16 bits).
Since calculation is performed in 32-bit floating-point decimal format inside the DSP
channels, both input and output are converted with 1 LSB weight.
Note that the 16-bit binary data of the output is converted with 1 LSB weight that is
determined by the Value/DIV setting.
Both input and output are displayed after being normalized at 2400 LSB/DIV.
Computation Input: Conversion from 16-Bit Binary to Floating-Point
The computation source data is converted to floating-point decimal format with 1 LSB
weight at the same time the DSP acquires the data.
A (Float) = A (Binary)×(1 LSB weight)
B (Float) = B (Binary)×(1 LSB weight)
Internal Computation
All calculations are performed using floating-point decimal format inside the DSP.
Example: C (Float) = A (Float) + B (Float)
Calculation of the 1 LSB Weight of the Output
The 1 LSB weight of the output is determined from the DSP range (Value/DIV).
Since 1 DIV = 2400 LSB,
1 LSB weight of the output = [Value/DIV]/2400
Computation Output: Conversion from Floating-Point to 16-Bit Binary
The output is converted into 16-bit data according to the following calculation.
C (Binary) = C (Float)÷(1 LSB weight of the output)
Appendix 6 DSP Channel Computation (Optional)
A(Binary) A(Float)
×
Floating-
point
comput-
ation
AD
1 LSB weight
B(Float)
AD
1 LSB weight
C(Float)
C(Binary)
B(Binary)
1 LSB weight
Range value Value/Div
of the DSP channel
ACQ
Memory
Computation source
16 bits
16 bits
Computed result
16 bits
Internally, all calculations
are performed using
floating-point.
V/DIV of input
2400 LSB/div
2400 LSB/div
16bit Binary -> Float
Float -> 16bit Binary
Computation
(floating-point)
500E+18 to 100E-21 [Value/Div]
(1-2-5 steps), total of 123 ranges
2400 LSB
/div
2400 LSB
/div
Value/Div determines the 1 LSB
weight of the 16-bit output data.
±
3.4028234E+38
to
± 5.8774117E-39
± 5E+21 to
± 1E-19
CH A
CH B
2400 LSB
/div
×