S.F. =
40
x
(
3600 **
)
*
x
1
40.00
3600
100
S.F. = 1 x 1 x 0.01
S.F. = 0.01
“Pro 5”...“dECPNt”
-
0
“tbASE”
-
2
“SCLFAC”
-
0.010
“Lo-cut”
-
0.00
The totalizer will totalize up to 999999
(999,999 Amp-Hours)
.
* This value is normally 1, but can be used as a coarse scale factor of 60 or 3600.
** Since the time period is in Hrs., the selected T.B. is 3600 (Program Select
Value = 2) which equals per hour (3600 sec.)
LINEARIZER
The linearizer feature is a series of programmable scaling points that are used to
construct linear segments to linearize the input signal. Correction for non-linearity
is accomplished by continuing with scaling points beyond
“DSP 2”
and
“INP 2”
in
“Pro 1”
or
“Pro 2”
, with
“DSP 3”
, and
“INP 3”
,
“DSP 4”
, and
“INP 4”
, etc.
The unit automatically calculates the linear segments between the programmed
coordinates. This process of entering linear segments is also known as
“curve
fitting”
. A maximum of nine segments are available. No restriction is placed on the
ordering of the display scaling points as long as the input signal scaling points are
all increasing or all decreasing. To have one or more points
“back-track”
, the
input/output
(signal/display)
relationship would not be a function and would be
undefined in that area. Additionally, consideration should be given to the location
and length of each segment to fully minimize the segment conformity error over
the desired range.
A typical curve is shown using five
segments
(six scaling points)
. Usually
it is desirable to use as many segments
as possible to reduce the amount of
l i n e a r i t y e r r o r . T h e f o l l o w i n g
program, written in GWBASIC
â
,
calculates the number of linear slopes
(segments)
required to linearize a
given non-linear relationship at
programmable error levels.
This program calculates two sets of values. One set represents percent of full
scale for the input value and the other represents percent of full scale for the
display value. These values are then used by the program user to compute the
actual input and display scaling points.
To use the program, copy it into any computer with GWBASIC
â
installed.
The program uses, in subroutine 10000, the relationship between the measured
parameter and the display reading. Of course, any non-linear relationship can be
substituted into the subroutine to yield the % of full scale input and % of full scale
display. The program will prompt for
(%)
of full scale error relative to display
readings and any other information pertaining to the process. Increasing the
conformity error decreases the number of linear slopes required to fit the
function. The IMH can accommodate up to nine linear segments and it is
generally desired to use all of them to minimize linearity error.
The program calculates the input/display scaling points
(the location of each
linear segment)
as a percentage of the full scale input and full scale display. To
obtain the actual input and display scaling points, multiply the respective percent
of full scale values by the respective full scale range for the input and the display.
“Pro 2”
is then used to enter the values into the IM unit. Certain linear sections of
a given curve may have a slope which exceeds the measuring resolution of the
instrument. The effect will be an erratic display in that part of the curve, if not
corrected. Correcting for this condition consists of three steps: increase digital
filtering to level 1 or level 2, decrease display resolution to 2 or 5 and/or add
dummy right hand zeros by programming 10 or 100 for
”round"
.
1 REM THIS PROGRAM WAS WRITTEN IN GWBASIC
â
UNDER MS-DOS 3.3
2 REM THIS PROGRAM CALCULATES THE NUMBER OF LINEAR
SEGMENTS REQUIRED
3 REM TO CURVE-FIT A GIVEN FUNCTION, SEEN IN LINE NUMBER 10000.
4 REM LINE 10000 MAY BE MODIFIED TO ANY MATHEMATICAL
EXPRESSION
5 REM INCLUDING THOSE WITH PROPORTIONAL CONSTANTS AND
MULTIPLE TERMS
10 CLS
15 PRINT “
CURVE FITTING PROGRAM”
16 PRINT “”
30 DIM PA(30)
-23-