Analysis of Pre-Flight Modulator Voltage Calibration Data for the … · 2013-08-30 · L Analysis of Re-Flight Modulator Voltage Calibration Data for the Voyager Plasma Science Experiment

Analysis of Pre-Flight Modulator Voltage Calibration Data

for the Voyager Plasma Science Experiment

Ognen Nastov

CSR-TR-8 8- 1

This work s u p p o r t e d i n p a r t under:

1 ) NASA C o n t r a c t 957781

2) NAGW-1209

https://ntrs.nasa.gov/search.jsp?R=19880010852 2020-03-05T21:07:08+00:00Z

L

Analysis of Re-Flight Modulator Voltage Calibration Data for the

Voyager Plasma Science Experiment

OGNEN NASTOV

Prepared Under the Undergraduate Research Opportunities Program

Supervised by Prof. R. L. McNutt, Jr.

Department of Physics and Center for Space Research Massachusetts Institute of Technology

Cambridge, MA 02139

. CSR-TR-88-1

k

April, 1988

- 1 -

1. Introduction

The digitized readout (DN = data number) of the low-voltage signal (MV) which is proportional to the potential (V) on the high-voltage modulator grids, is a part of the routine calibration sequence on the Voyager Plasma Science (PLS) experiment. Theoretically, in producing the modulator voltage, the voltage is stepped through one decade from 60 to 600 volts and then through the same steps with a multiplication factor of 10 to sweep from 600 to 6OOO volts. To this voltage is added an offset of -50 volts.

The voltage monitors measure a voltage between 0.1 and 10 volts (a constant fraction of the voltage swept) which is then input into the fast A/D converter. This voltage is converted to a binary number from which the highest order bit is discarded. Thus, for M V sweeping from 0.1 to 1.0, a number between 0 and 255 is returned; for M V sweeping from 1.0 to 10 volts, again, a number between 0 and 255 is returned. To know which decade is being read out, one must either rely on the “proper sequence” or an “educated guess.”

For more information on MVM Interpretation, see Voyager Memorandum #161 by R. L. McNutt, Jr. (attached).

2. An Outline of the PLS Calibration Data Analysis

The PLS Modulator Calibration Data Analysis was undertaken in order to check the correctness of the fast A/D converter formulas that connect low-voltage monitor signals (MV) with digital outputs (DN), to determine the proportionality constants between the actual modulator grid potential (V) and the monitor voltage (MV), and to establish an algorithm to link the digitized readouts O N ) with the actual grid potential (V). The data used for the DN-MV analysis were from the calibration tests for the PLS-Prototype run at MI” CSR (included as Appendix A of Voyager Memo #161). The MV-V analysis used the results of the power supply tests, made by Matrix Research and Development Corp. for both PLS instruments to be flown on the Voy- ager 1 and Voyager 2 spacecraft (in the further text referred to as “test results”). The Matrix data was obtained between 11/11/76 and 11/15/76 for “Flite 1,” and between 12/21/76- and 12/27/76 for “Flite 2.” Drawings containing this data are labeled “M.J.S. Power Supply” (there are no drawing numbers). These are stored in the fling cabinet next to the outer wall, second drawer from the top in N52-367. Note that there is some potential confusion about which unit is which: if Flite 1 is SN002, then it is the instrument actually flown on Voyager 1. SN002 was put on Voyager 1 and SN003 which was supposed to have gone on Voyager 2 was not flown. When these data were taken, Flite 2 probably referred to SN003 which has remained at M.I.T. as the flight spare. SNOOl was actually flown on Voyager 2.

- 2 -

All calibration tests and power supply tests were entered into a sequence of files. The method that was used in both analyses, i.e., DN-MV and MV-V, is an improved version of a least squares fit algorithm, implemented in several FORTRAN 77 programs (see, for example section 14.2 of Numerical Recipes: The Art of Scientific Com- puting, Press, W. H. , B. P. Flannery, S. A . Teukolsky, and W. T. Vetterling, Cambridge University Press, New York, 1986).

3. Description of the Files

All of the files related to the investigation can presently be found on the directory /usr4/ojn. The file suffixes have the following meanings:

.dat ......... data files

.mon ......... mongo source code files

.f ......... FORTRAN 77 source code files

The DN-MV test data and mongo files have filenames of the form:

p[ mode J[ mvr # ].[ suffix ]

The MV-V test data and mongo files have filenames of the form:

f I flite # I-[ temperature 1-[ mode I.[ suffix 3

where:

rnvr # = 1 if 0.1 S MV S 1.0 m v r # = 2 i f 1.0 S M Y S 10

flite # = 1 for Voyager 1 PLS Instrument flite # = 2 for Voyager 2 PLS Instrument

temperature = 5 for Temp = -5°C temperature = 10 for Temp = 10°C temperature = r for Temp = room-temp temperature = 30 for Temp = 30°C temperature = 40 for Temp = 40°C

(The temperature is that to which the instrument was exposed during the test.)

mode = 1 for the L-mode mode = m for the M-mode mode = e l for the El-mode mode = e2 for the E2-mode

- 3 -

suffix = dat if the file is a data file suffix = mon if the file is a mongo source code

Note that the= are no files of the form f2-30-[ mode I.[ suffix ] due to lack of this information.

The files of the form:

tl[ mvr # 1.f and tl[ mvr # ].dat

are FORTRAN 77 source code and output data files produced by their corresponding source code programs which makekontain the data from the theoretical fast A/D formulas for MV-DN conversion.

The FORTRAN 77 source code files that actually perform the analysis are:

dns-code.f ..... DN-MV analysis mvsc0de.f ..... MV-V analysis mvs-tc-code.f ..... MV-V analysis combining the data files

that differ only by temperam

These source code files use the following driver files:

dns-driver.dat mvs-driver.dat mvs-tc-driverdat

The analysis generates the following files:

dns-output.dat ..... DN-MV analysis results vs-output.dat ..... MV-V analysis results mvs-tc-out.dat ..... MV-V temperature-combined analysis results

- 4 -

4. Data File Organization

All data files are organized as matrices of numbers and/or strings. Their row structure is as follows:

p[ mode I[ mvr # ].dat and tl[ nivr # ].dat:

dn mv where: dn are the data numbers

mv are the corresponding monitor voltages

f [ flight # I-[ temperature I-[ mode 1.dat:

channel# hvdcout mvu mvl vl vu where : channel# is the channel number

hvdcout is the dc high voltage output mvYmvu are the monitor low voltages at the

lowedupper channel edge vYvu are the actual grid potentials corresponding

to the mvYmvu

dns-driver.dat:

filename mvr[ mvr # ] lines# where : filename is the name of a file.

mvr[ rnvr # ] is a character string that can be either mvrl or mvr2 depending on what range the value of M V belongs to.

lines# is the number of lines in the file.

mvs-driver.dat:

filename startline-1 endline-1 startline-2 endline-2

wheik : startline-[ chr # ]/endline-[ chr # ] arc the starting/ending line numbers for parts of the 4 I-[ 3-[ ].dat data files that correspond to different channel ranges as follows:

c h r # = l if OSchannel#S7 and m o d e = L o r E 2 .......... 0 S channel# S 63 and mode = M .......... all channels and mode=El

chr # = 2 if 8 S channel# S 15 and mode = L or E2

- 5 -

.......... 64 5 channel# S 127 and mode = M

mvs-tc-driver.dat: (Note: this file is a single-column file.)

mfilename# 1 filename#l.l filename# 1.2

mfilename#2 filename#2.1

*

when : d e n m e # [ integerl ] is a character string of the form:

f[ flite # I-[ mode ].dat which is found in the mvs-tc-outputdat file and labels the analysis results obtained by combining groups of fc I-[ I-[ ].dat files that differ only by temperature.

filename#[ integerl I.[ integer2 ] are the filenames of the f[ I-[ 3-[ ].dat data files that were combined.

dns-output.dat:

filename mvr[ mvr # ] c l c2 sqd pairs#

where : filename is the name of a file that labels the information obtained by its analysis.

mvr[ mvr ## ] is as explained under dns-driver.dat. cl, c2 are constants in the presumed function

sqd is the square deviation of the function. pairs# is the number of analyzed data pairs.

that fits the data.

mvs-outpu t.da t:

filename chr[ chr # ] mvx-vx a b sqd siga sigb cov pairs#

where : chr[ chr # ] is a character string that can be either chrl or chr2. chr # has a meaning as explained under mvs-driver.dat.

mvl-vl or mvu-vu. It specifies what monitor voltages wen analyzed - those at the upper or

mvx-vx is a character string that can be either

- 6 -

at the lower channel edge. a, 6 arc constants in the presumed function that

fits the data. sqd, siga (oJ, sigb (ob)are,

respectively, the square deviation of the function, and the deviations of its coefficients a and b.

cov (od) is the covariance of a and b.

mvs-tc-output.dat

d l e n a m e chr[ chr # ] mvx-vx a b sqd siga sigb cov pairs#

where : mfilename is the character string that labels the analysis data, and contains information what files were combined for the analysis (see mvs-tc-driver).

5. Mongo Files

All files that have a suffix .mon are mongo source code files. There are two types of such files: p[ I[ ].mon and f [ I-[ I-[ ].mon files.

Every mongo file is used to create a hardcopy plot of the data file associated with the mongo file. A plot is obtained by typing:

mongo [ mongo-filename ]

p[ mode I[ mvr # ].mon plots include both the p[ mode I[ mvr ## ].dat data (square dots) and the tl[ rnvr # ].dat data (broken line). fl flite ## I-[ temperature I-[ mode ].mon files produce four plots out of the associated f [ 1-[ I-[ ].dat data for different combinations of the channel range and choice of mvl-vl or mvu-vu pairs of data (square dots).

6. Fortran Files Files that end with .f are FORTRAN 77 source code files. There are five such

files: tll.f, t12.f, dns-code.f, mvs-code.f, and mvs-tec0de.f. The sequence of UNIX commands:

f77 -0 execfile fi1ename.f execfile

- 7 -

performs a compilation of the source code and an execution of execfile, the gen- erated object file.

tl1.f & tl2.f:

These programs generate the theoretical DN-MV pairs assuming the following exponential functions:

= eiz5-DN)tK

for tl1.f i.e. mvr# = 1 m = 10 e-w-DN)r/.F

for tl2.f, ie., mvr# = 2

where t = 4.34 ps, T = 482.55 ps and 0 S DN 2 255 (From A. Mavretic’s lab note- book labeled book 2, #254, July 1974). Note that d’’ = 10” 256.0167 for these values.

dnsc0de.f: This program does a least squares fit by analyzing the p[ ][ ].dat data files, and

presuming that MV and V are related by the following exponential relation:

(3) M V = c l e -w-m ca where c1 and c2 are constants that are determined by the fitting procedure.

The program produces the file dns-output.dat. The algorithm used is a modification of the general least squares algorithm that makes the round-off errors as small as possible.

mvs-c0de.f : This program uses the same algorithm as the previous one. It presumes that MV

and V are connected with the following relation:

V = a M V + b (4) where a and b axe constants that are determined by the fit.

The program analyses the fl I-[ I-[ ].dat data files, dividing each file into four parts using the driver file. Every part is characterized by a unique channel range and mv range (mvr). The analysis, i.e., the least squares fit is carried out independently for each part. The file mvs-output.dat contains the results from the analysis.

The formulae used to calculate the deviations of the obtained coefficients are:

where

and

with N the number of MV-V pairs. mvs-tc-code.f :

This program employs the same least squares fit algorithm as the previous two. It presumes the same relation between MV and V, and performs the same analysis as mvs-code.f does, except that it internally combines the 4 I-[ I-[ ].dat data files that differ only by temperature. The results of the analysis am contained in the file mvs-tc- 0ut.dat.

7. DN-MV Analysis

The primary goal of this analysis was to find out how much the theoretical DN- MV formulas are off from the true values. The only information available is from the calibration tests performed on the PIS-Prototype. This information was severely limited - only one test, at unknown temperature, per channel in each mode. Also, there is no information on the measurement uncertainties. Hence, the analysis results should be used with some caution. consideration.

The mongo plots of DN-MV data files @N vs. log MV) show that DN and M V are related through an exponential relation. The DN-log(h4V) pairs, as can be seen from the‘plots, lie on a straight line which, however, has a slightly different slope and y-axis intercept than the theoretical line. This suggested that there might be some nonlinearity in the A/D conversion, if one assumes that DN’s and MV’s were measured at the same time. In addition, the slope and intercept shifts are bigger for mvrl (0.1 5 MV S 1.0) than the corresponding shifts for mvr2. This implies that the multiplication factor in the DN-MV formula for this range is not exactly 10.

The investigation of the plots sugested that the presumed function for the least square fit be:

- 9 -

The program actually uses a linear least squares fit algorithm. If we take h of the

(10)

both sides of the equation, we obtain: In MV=(DN-255) c2 + In c1

If we consider t (the period of the fast A/D converter) as constant, then the time constant z might be off (bigger) 1.085% + 1.201% from the nominal value of 482.55 ps, or, on an average, by - 1.15%.

which can be considered as a linear function y = ax+ b setting: x=DN-255; y = In MV ; a = c2 ; b = In cl.

Before the data files were used for the analysis, all DN-MV pairs which were significantly off the imagined line that connects the other points on the mongo plots were deleted. They obviously represent a subjective reading, or even a writing, error.

The analysis results show that the presumed function yields a good approximation, since the square deviation is less than lod in most cases. The multiplication factor c1 ranges from:

1.02831 + 1.02864 instead of 1 for mvrl (2.831% + 2.864%) 9.91158 + 9.92186 instead of 10 for mvr2 (4.8842% + -0.7814%)

The range deviations for c1 in percentages are:

mvrl: f 0.016% mvr2: f 0.052%

is surprising that the cl(mvr2)/cl(mvr2) ratio is less than 10 ; it ranges from 9.63880 + 9.64560. This implies that PLS circuitry constants responsible for the multiplication factor might be off by -3.5% from their nominal value(s).

The constant in the exponent, c2 = t / z, ranges from :

0.888580 x 1W2 + 0.889623 x 1W2 instead of 0.899388 x (1.201% + 1 .085%).

The range deviation for c2 is f 0.058%.

- 10 - 8. MV-V Analysis

The goal of this analysis was to determine the proportionality constants between the monitor low voltage and high voltage (Le., the actual grid potential). The low voltage is (nominally) proportional to the high voltage applied to the modulator grids (at least in the linear regime of the amplifiers). The theoretical value of this factor is not known, although it can, in principle, be derived from the circuit diagrams.

The analysis uses the results of the power supply test, conducted by Matrix Research and Development for two of the PLS instruments (refer to discussion in section 2). The tests were performed for several temperatures. Although all modes were incorporated in the measurements, the files were highly incomplete and difficult to read (handwritten). As is the case with the DN-MV analysis, there is no information on the measurement uncertainties.

All of the plots of MV-V data files clearly show that MV and V are linearly dependent. Since no information was readily available for the theoretical factor of proportionality, the plots do not include any theoretical graphs. A simple linear function was presumed for the least squares fit:

Similar to the DN-MV analysis, all MV-V points with unusually large deviations from this function were deleted from the files.

Throughout the analysis, every file is divided into four parts (with an exception of mode El files which are divided into two parts with respect to mvr). Each part has a unique channel range (chrl or chr2), and monitor monitor voltage range (mvrl or mvr2). These parts are processed independently.

The analysis results (file mv-outputdat) are somewhat surprising. First, the square deviation ranges from roughly to over 350 (considering least squares fits of more than 2 data pairs). This might be evidence of harsh measurement errors during the test or strong non-linear influences in the PLS circuit. The random distribution of the square deviation values shows that the errors or influences are not dependent on factors such as: temperature, mode, voltage or channel range. The deviations obtained for a and b, as well as the covariance of a and b, show a much more stable picture; their median values are, roughly, as follows:

6, -0.1

b& - -10-2 6 b -0.1

- 11 - It is interesting that the values obtained for a and b show an obvious dependence on various factors, although, ideally, they should not. If one looks at the values for the slope a, it can be seen that in 90% of the cases:

a(mv1-vl) > a(mvu-vu)

There are only eight cases when this relation does not hold true; since these cases appear to be totally random with respect to any factors that might have influenced them (temperature, specific mode, etc.), the observed property may serve as an evidence that some circuit constants change when switching from the lower to the upper channel edge.

The ratio a(chr2)/a(chrl) randomly oscillates around the value of 10, which is the nominal value for the multiplication voltage factor when the instrument switches from the first channel range to the second channel range. The stability of this multiplication factor shows that the circuit constants for the appropriate part of the circuit are close to their nominal values.

This situation, however, changes in considering the dependence of the slope of a on the mode. The values of a for various modes vary roughly as follows:

1-mode ..... a = 60.5; 605.1 m-mode ..... a = 60.3; 603.5 el-mode .... a = 67.2 e2-mode .... a = 68.1; 672.1

The slight difference between a(1-mode) and a(m-mode) can be noticed in many cases, especially considering values of a which correspond to a smaller square deviation and larger number of analyzed MV-V pairs. On the average, a(1-mode) is 0.33% larger than a(m-mode). Following the same criterion, one notices that a(e1-mode) and a(e2- mode) differ more drastically from a(1-mode) and a(m-mode). The percentage increase from a(m-mode) turns out to be, roughly:

a(e1-mode) ..... 11.45% a(e2-mode) ..... 12.93%

This suggests that some circuit constants in both E modes differ from their nominal values.

The last consideration for the slope values is whether or not they are temperature dependent. It is important to note, once again, that the analysis was limited to five different temperatures. The values of u do not show any dependence on temperature; they tend to oscillate randomly as the temperature increases. This observation is con- sistent with the theoretical expectations.

- 12 - The values of b, which represent the y-axis intercept or the voltage offset are

much more randomly distributed than are the slope values. In fact, b values vary so much that no dependence property can be uniquely established.

The values of b vary from -40 to +40; however, most of them are between -5 to 0. In roughly 90% of cases the values are negative. Theoretically, the values of b should be close to 0. Typical values for b are (roughly):

-0.4 ... for chrl -1.5 ... for chr2

9. MV-V Temperature-combined Analysis

Since the slope values did not show any dependence on temperature, all families of files that differed only by temperature were collapsed, and the data from the corresponding parts of these files combined for the least squares fit. The result of this analysis is the file mvs-tc-outdat. First, it can be noticed that the lower edge of the square deviation range has moved toward larger values (from The square deviations for a and b, and the covariances are, for the most part, in the same range as in the previous analysis.

to

The relation:

a(mv1-vl) > a (mvu-vu)

still holds true for 80% of cases (there are three cases when it does not).

The ratio a(chr2)/a(chrl) oscillates around 10 (the theoretical value) as in the MV-V analysis.

This analysis also confirms that there is a definite dependence of a on the mode. Some spot checks suggest the following percentage differences (with respect to a(m- mode)):

a(1-mode) ..... 0.13% a(e1-mode) .... 11.75% a(e2-mode) .... 12.12%

(The median value for a(m-mode) is 60.13.)

The values of b are, again, very randomly distributed, so that no firm property or dependence can be established. For example, bme = -4.631 volts. However, if the first four large values of b are not considered (since they range from -20 to -30), the new,

- 13 - restricted average value of b becomes: b, ru = -1.153 volts.

This example clearly shows that it is very difficult to derive any major conclusion about b from the analysis results obtained.

10. Conclusions

The analysis results are surprising in that the derived conversion constants deviate by fairly significant amounts from their nominal values. However, it must be kept in mind that the test results which were used for analysis may be very imprecise. Even if we assume that the test result enors are very large, they do not appear to be capable to account for all discrepancies between the theoretical expectations and the results of the analysis. Measurements with the flight spare instrument appear to be the only means of investigating these effects further.

It is very clear that is impossible to create one simple algorithm that for given DN will return V - the actual grid potential. The MV-V slope dependence on the channel range is what was expected. However, the MV-V slope dependence on whether we have the top or the bottom of the channel, and a dependence of the MV-V slope on the mode, precludes a unique algorithm for all cases.

In order to convert a given DN into the V, we need the following auxiliary information:

1- themode 2 - the decade the modulator is sweeping through (mvr#) 3 - the channel range (chr#) 4 - the channel edge the DN corresponds to (mvl-vl or mvu-vu)

Given the stated information (for 2 and 3 we must rely on the “proper sequence” or an “educated guess”), one can pick up the corresponding constants from the files dns-outputdat and mvs-tc-out.dat, and substitute them in the general formula:

to obtain the actual grid potential V. The error in the value obtained will be as large as the errors in the test results the DN-MV-V analysis is based on.

For example: Flite# = 2 ; Mode = M ; mvr# = 2 ; chr# = 1 ; channel edge = lower (mvl-vl). The values of the constants in the general DN-V formula, are:

~1 = 9.91168 u = 60.13527

~2 = 0.888609 x lo-* b = 0.71937 x lo-’

If the value returned from the PLS instrument is DN = 73, then the grid potential is 118.3 volts.

- 14 - It is important to note that the general DN-V formula only yields magnitudes; the

actual potentials in the El and E2 modes will be negative.

11. Description of Tables

Tables 1,2, and 3 are, respectively, hardcopies of the files dns-output.dat, mvs- output.dat, and mvs-tc-out.dat.

12. Description of Graphs

The first set of graphs contains hardcopy plots of p[ I[ ].mon mongo files. The second set of graphs contains hardcopy plots of fI I-[ I-[ ].mon mongo files (see section 5) .

13. Description of Source Code used in the Analysis

Hardcopy listings (with comments) of the files dns-code.f, mvs-code.f, and mvs- tc-code.f are included.

14. Voyager Memorandum #161

Voyager Memorandum #161 is appended.

d

d , QD

In .. In 0 .. d 0

d l-l

I 4J I 4 I 4J

a

a a 4J

e :

P Q D I D P W P P Q 00000000 I I I I I I I I a l a l a l a l a l a l a l a , r n C J 1 0 r n P h l I n P

hl 0 I al Q W In m

hl 0 I al 0 Q In aD

N 0 I a, m

W P Q

N 0 I a, m 0 w OD

hl 0 I a, m rl w m

e4 0 I al u) m 0 m

hl 0 I al m N u) m

hl 0 I al m m u) Q

Q Q Q D Q Q D Q Q Q W Q Q Q Q Q D a D Q D

I I I I I I I I . .-. . . . . .

r ( h l r ( h l r ( h l d N & & & & & & & & 3 $ 3 3 3 3 3 3 E E E B E E E E

rl

3 M

d N 00

w N 0

dhl 00 00

d w d d 0 0 00

O d d d r l m w d d d r l m d r l m rld 4 N d P P m m 000 0 0 0 0 00 0 0 0 000 00 0 000000

d 0 I

d 0 I

dol d C J 00 00

d d rld 00 00

I Q) m

W w (v I- OD

I Q)

W OD m Io OD Q)

t a, d W m OD w w

I Q) (v CJ 0 0 OD QI

I a,

OD 0) CJ d d (v

I a,

Q) l- I- 0 CD m

I a, I- d. m W Q d

Q) m m m 0 w

OD w Q) w 0 I-

(v OD d W 0 0

d. m W m w (v

OD 0 w m OD m

OD I- CJ m I- d

(v 0 I- d 0) I-

0 m 0 W m w

OD m d W m m

(v 0 I- m m d

OD m 0 m W

N

l- w 0 d. 0 d

W m I- 0 m m

W l- m 0 l-

.-.1

w OD

d I- w d 0 w

6i I- I- OD m w

E E . - . . .... o o o o o o l 4 n o o o o o o o o o 000 000000000000

d d 4 d C J dd(v d d de4 dd(vd d d 00 0 00 000 0 0 00 0000 0 0 I Q)

OD Q) m m m m 0

l a, w 0 m m OD OD

0

I 0) W w w

h (v cp In w W (v

0

I a, m 0 0 m d (v

0

I a, m m

OD 0 OD OD

0

I a, d 0 I- m I- d

0

I a, (v w (v d OD d

0

I Q) Q) OD m m N I-

0

I a, (v (v OD 0 m CJ 0

1 a, w d m l- Q) (v

0

I a, m w (v W m w 0

I a, 0 0 OD m (v CD

0

I a, W 0 W I- (v (v

0

& W (v 0) OD Q) W

0

I a, l- d W Q) d OD

0

m m m d (v (v

0

w l- w (v I- d

0

0 0 m m 0 (v

0

(3 m OD CJ 0 d

0

(0 0 OD W (v

CJ

0 OD 0 W 0 d

0

0 l- w (v P d

0

d d 0 m m 0

0

d OD 0 m m d

0

w m m (v W m 0

Io Q) 0 w Q) d

0

\o m 0

cu C H

w C n

m

CJ (v 0 0

(v d 0 0

I- I- (v W CJ

In 0

OD m OI 8-4

0 d

..

..

c - $

U

a" i 0 d I

(v w

U

a" f 0 d I

(v w

a" d a, I 0 rl I 04 w

ia a d

P 0 d I

(v w

m d (v a, I 0 r 4 I

(v w

a" (v a, I 0 d I

(v w

m a (v a, I 0 d I

(v w

U

a" 4 I & I

(v w

' c , a" FI

I u I

(v w

U

a" 4 I bl I

(v w

U

a" 4 I

Lc I

(v w

U

a" F U I

(v w

c, a" 4 Li I

(v w

' U

4 4 Lc I

(v w

U

a" 7 Li I cv w

U

a" rl Q) I

Lc I

(v w

c, 4 d a, I u I

hl w

U

a" (v a, I Li I

(v w

U

a" (v a, I Li I N w

U

a" (v a, I Lc I N w

U

4 (v a, I & I

(v w

c, a" 4

I 0 w I

(v w

U

a" 4

I 0 w I

(v w

U

a" 4

I 0 w I

(v w

c, 4 rl I 0 w I

(v w

U

a" f 0 w I cy w

U

a" 4 0 w I

(v w

U

4 f 0 w I cv w

U

a" 4 0 w I

(v w

i a d m i a m i a a a a a a a . . . . . . a (v a, I 0 d I

(v w

dd(v(v(v(v Q ) Q ) Q ) a , a , a , I I I I I I 000000 w w w e w w I I I I I I

( v ( v ( v N ( v ( v w w w w w w

d

E 7 rn

r)

d d d F I N N N N d d d d d d d d d d N N N N r l d d d d F I 0000000000000000000000000000

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

. . . . . . . . . . . . . . . . . . . . . . . . . . . . I I I I I I I I I I I I I I I I I I I I I I I I I I I I

d N rlrl d d d d d d 00 00 00 00 0 0

. . . - .. . . . . . . 0 00 00000 0000000000000

d 0

I al m hl w k-l m m 0

4 d 00 d - 00 0 0

m d d 0 00 I 1 1 I 1 I I I

d d 0 0

I I

d d 0 0 I I

-.8

-1

0

................................... + ...................................

................................. + ............................___.

7 l l - r ..................................

...................................

1 . do t 7 - l l - r

....................................

J-LLL

50 100 150 200 Reference counts (DN)

-0CT-87 -n+F

.............................. , , l l l j _.._ .....

250

1

.8

.6

.4

.2

0

I I I I

.......................... ...... 1...1..1.... ......... -

2 adat n-d ...................................

f ..a‘

....-......... ........- .

.. ..-........... .................. -

-0CT-87 00 l7ll-

.......................

.................................... - 0 50 100 150 200

Reference counts (DN) 250

l - r l - r -.- .............................

..........-.... I .................

I1 .dot 19-OCT-87 1( I I I I I I I V

....-............. ......

...................................

e; 0.2 '

O,,." .''

O...'

a,.'

0. .;e.'

,/,

.-• ............................ ..

0 50 100 150 200 250 Reference counts (DN)

0

..................................

................... ..........-.

i

..................................

llll-

..... ...........

... ................................

........... -.._ ................ --.

* I .................

.... .......

&#

3 P

2 .dot 7-r-d

d 8

B P

d a' z ..... I.. ........................

...................................

-0CT-87 -nl+

.a

s#&

d J)..-.. .......................

...................................

...................................


250

, .................................. - ....................................

j

... .- ........................ +..- ... ...-.......-.--. j

.............................. ---.- .- ....................................

i

j

q,/:*’. I

-4: ...............................

....................................

u 0 50 100 150 zoo 250

Reference counts (DN)

2 .dat -n-d -0C T-87 - I l l +


I I I I

................................

32 1 . mon pe2 1 .dot 1 I l l 1 I I I I

..............................................................................

-0C T- 8 7 n - r +

,/’

‘/, , .: *,:i

,,*’

c ................................

..................................... - ...................................... i


1

.8

.6

.4

.2

0

I I I I I i j i

--t

- i

- ,

- .

j -.... -..I-...... --._-I ...... -

- I

- . j

- . j

-+ ...................................

- :

- *

!

- :

......................................

- ,

- I

- :

.................................. I_

,/

. j ,/ I

j ,/ . : i-

. j /*"

i' j */ ./'. : ,- ! /a

i ,/. : ;

_......___..........__. I_.-.... I..

2 .do t +

,I' I

i /' p'

,*. L .. ......._... ............ I- I..

..........._...............--....

u

-0CT-87 l 7 - r - P

................ I.- ..... - ........

-LLLL


250

400

300

n >

> W

4 200

100

0 0 2 4 6

Channe1s:O-7 Monitor VL (V)

6000

5000

4000 n * ’ 3000 W

2000

1000

2 4 6 8 10 Channels:& 15 Monitor W (V)

0

0

2 4 6 Channek8-15 Monitor VL (V)

Flite: 1 Temp:-SC Mode:L

500

400

h

5 300 3 >

200

100

0 0 2 4 6 0

Channe1s:O-63 Monitor W (V)

500

400

300 > W

5! 200

100

0 0 2 4 6 8

Channe1s:O-63 Monitor VL (V)

6000

5000

4000

n >

2 3000 v

2000

1000

6000

5000

4000

n > g 3000 W

2000

1000

2 4 6 0 Channek64- 127 Monitor VI., (V)

2 4 6 0 10 Channek64-127 Monitor VU (V)

Flite: 1 Temp:-SC Mode:M

.5 1 1.5 2 Channe1s:O- 15 Monitor VL (V)

0 2 4 6 Channe1s:O-15 Monitor VL (V)

2 4 6 8 Channelso- 15 Monitor W (V)

F1ite:l Temp:-SC Mode:El & E2

600

500

400 n > 3 > 300

v

200

100

0

400

300

n >

> v 4 200

100

0

r1-10-1 .dat I l l I l l 1 1 1 - - - j o

- - I ! - --+.--I) .-_--.--- + .................. ...,...----- - j j 1 j - - - - i -

- i - __ ........... ...:......... ............ i.... ................ 4 ................ .... - - - - - - - - ___. ............ i ..................... i...-.. .__.I..... ..i ............. ........ i I.-.-_ - - - - - -

............ .................... ..................... - - _... i ............._...... 4 ..--..__ - - - - - 3

- i j - _ ......- . ,...._ i..- 4 ................ -...; ..................... i ...-._._.- - - - - - - - - _ ............... i ..................... i.... ................ i..... .............. ..i ......... ..- - - - 0 i - - -

D - " J l l J l l l ' l ' J l l l

- 2 4 6 8

0 2 4 6 Channelso-7 Monitor VL (V)

I l l ,...-............... I.- . ......-.... -

i - - -

..... I... ......... 1 ....-.............. & - -

D j - - .......... ___..- ...... ......_

- - - j -

-_ ..........-.... (. ............-.... & - 1 - - -

........................................ - - - -

. ........-....-.. 4. --.- -&

i- I l l I l l 7

6000

5000

4000 n * 3 * 3000 v

2000

1000

n > W

4000

3000

2000

1000

2 4 6 8 10 Channe1s:B-15 Monitor W (V)

Flite: 1 Temp: 1OC Mode:L

n > 3 > W

50

40

30

20

10

6000

5000

4000

n * 9 3000 W

2000

1000

.2 .4 .6 .8 Channe1s:O-63 Monitor W (V)

2 4 6 8 10 Channels:64-127 Monitor W (V)

.2 .4 .6 .8 Channe1s:O-63 Monitor VL (V)

2 4 6 8 Channels:64-127 Monitor VL (V)

Flite: 1 Temp: 1 OC Mode:M

n * U

3

140

120

LOO

80

60

40

20

.5 1 1.5 Channe1s:O-15 Monitor VL (V)

.5 1 1.5 2 Channe1s:O-15 Monitor W (V)

4000

3000

n > v

51 2000

1000

0


6000

5000

4000

n * - 3000 3

2000

1000

0

2 4 6 8 Channe1s:O- 15 Monitor W (V)

F1ite:l Temp:lOC Mode:El & E2

2 4 6 8 Channe1s:O-7 Monitor W (V)

6000

4000 5000i 2oooi 1000

Channe1s:B-15 2 4 Monitor 6 W 8 (V) 10

2 4 8 Channels:8- 15 Monitor VL (V)

Flite: 1 Temp:Room Mode:L

500

400

Y 5: 300 ?

200

100

0

500

400

n 300 * W

0 2 4 8 8 Channe1s:O-63 Monitor W (V)

200

100

0 I

0 2 4 6 8 Channe1s:O-63 Monitor VL (V)

6000

5000

4000

n >

? 3000

U

2000

1000

8000

5000

4000

n > $ 3000 v

2000

1000


Flite: 1 Temp:Room Mode:M

.5 1 1.5 Channe1s:O- 15 Monitor VL (V) .5 1 1.5 2

Channe1s:O-15 Monitor W (V)

0 2 4 6 ChannekO-15 Monitor VL (V)

F1ite:l Temp:Room Mode:El & E2

500

400

200

100

2 4 6 8 Channels:O-7 Monitor W (V)

400

300

n > v

5! 2oo

100

0 0 2 4 6

Channe1s:O-7 Monitor VZ. (V)

6000

5000

4000 n * ’ 3000 v

2000

1000


4000

3000

n * W

$ 2000

1000

2 4 6 Channels:8-15 Monitor VL (V)

Flite: 1 Temp:30C Mode:L

8000

5000

.2 .4 .6 .8 Channelso-63 Monitor VL (V)

4000

2000

1000


Flite: 1 Temp:30C Mode:M

120

100

80

60

40

20


4000

3000

n + v

9 2000

1000

0



6000

5000

4000

n * 3000 s 2000

1000

0

F1ite:l Temp:30C Mode:El & E2


2 4 6 Channe1s:B-15 Monitor VL (V)

Flite: 1 Temp:40C Mode:L

500

400

n 300 > W

9 200

100

0

0 2 4 6 8 Channels:O-63 Monitor W (V)


6000

5000

4000

n >

? 3000 W

2000

1000

6000

5000

4000

n > 9 3000 W

2000

1000

2 4 6 8 Channek64-127 Monitor VL (V)

Flite: 1 Temp:40C Mode:M

.5 1 1.5 Channelso- 15 Monitor VL (V)

4000

3000

n * Y

5i 2000

1000

0


n > W

140

120

100

80

60

40

20


6000

5000

4000

n * - 3000 5i 2000

1000


F1ite:l Temp:40C Mode:El & E2

6000

5000

4000 n * ’ 3000 v

2000

1000

4000

3000

n > W

9 2000

1000

2 4 6 Channels8-15 Monitor VL (V)

Flite:2 Temp:-SC Mode:L

0 2 4 6 8 Channels:O-63 Monitor W (V)

2000

1000

i j

i .--.I .... .)-..--. ..........

i j

.----- ...+-......_........

j

-----..A .-. .............. j

i ------+- ...........

j

j

Channels:64-127 2 4 Monitor 6 8 W (V) 10


2 4 6 8 Channek64-127 Monitor VL (V)

F1ite:Z Temp:-SC Mode:M

140

n > J > W

120

100

80

60

40

20

.5 1 1.5 Channels:O-15 Monitor VL (V)

100

h

k 80 3 *

60

40

20


4000

3000

n > W

9 2000

1000

0

0 2 4 6 Channe1s:O- 15 Monitor VL (V)


6000

5000

4000

n * - 3000 CI d * 2000

1000

0

F1ite:Z Temp:-5C Mode:El & E2


i - i i

j

j -

... I ..__ ~ ___. +-- -

-

,.._._.I....__.__

- - -

....-.._.._ c-

- -

i -

i

J I I I I '

1.2 1.22 1.24 1.26 1.28 1.3

780

760

n * v

i?

740

720

2 4 6 Channek8-15 Monitor VL (V)

F1ite:Z Temp: 1OC Mode:L

.2 .4 .6 .8 Channelso-63 Monitor W (V)

50

40

30

20

10

6000

5000

4000

n >

? 3000

v

2000

1000

5000

4000

n * $ 3000 v

2000

1000

.2 .4 .6 .8 Channe1s:O-63 Monitor VL (V)

2 4 6 8 Channe1s:g.C-127 Monitor VL (V)

F1ite:Z Temp: 1OC Mode:M

Channe1s:O-15 .5 1 Monitor 1.5 W (V) 2

2 4 6 2 4 6 8 0 Channe1s:O-15 Monitor VL (V) Channe1s:O-15 Monitor W (V)

Flite2 Temp:lOC Mode:El & E2

500

400

n

5 300 3 >

200

100

n > v

400

300

200

100

0

2 4 6 8 Channe1s:O-7 Monitor VU (V)


6000

5000

4000 n > W

3000

2000

1000

2 4 6 8 10 Channe1s:B-15 Monitor W (V)

4000

3000

n > Y

9 2000

1000

2 4 6 Channels:& 15 Monitor VL (V)

I I Flite:2 Temp:Room Mode:L

500

400

300 W

i2 200

100

0

500

400

n 300 * W

5! 200

100

0

i 1 i

0 2 4 6 8 Channelso-63 Monitor W (V)


6000

5000

4000

n * ? 3000 W

2000

1000

2 4 6 8 10 Channe1s:BQ-127 Monitor W (V)

5000

4000

n * $3000 W

2000

1000

2 4 6 8 ChannelsB4-127 Monitor VL (V)

Flite:2 Temp:Room Mode:M

100 1 2 0 F

80 c- n > cl > 60

Y

20 40/7 L

.5 1 1.5 Channelso-15 Monitor VL (V)

4000

3000

n > W

5! 2000

1000

0


140

120

100

n

t. 80

s 60

40

20

6000

5000

4000

n > - 3000 ?

2000

1000

0

Flite:2 Temp:Room Mode:El & E2



2 4 6 Channels8-15 Monitor VL (V)

Flite:2 Temp:40C Mode:L

0 2 4 6 8 Channelso-63 Monitor W (V)

500

400

n 300 > W

fl 200

100

0

6000

5000

4000

n * ? 3000 W

2000

1000


5000

4000

n > fl 3000 W

2000

1000

0 2 4 6 8 Channels:O-63 Monitor VL (V)

2 4 6 8 Channels:64- 127 Monitor VL (V)

Flite:2 Temp:40C Mode:M


4000

3000

n * v

? 2000

1000

0

0

C2-40-.2.cbt OS-WV-87 17 : 38 :54 I l l 1 - I I l l I l l

- 1 0 - - j -

......... . --..- .... -..+ --._-....- i --.- - i - i

4 ...._.-._......... ;.- - - -

i n ; - +-

- ...................

j

j j - " I

- --. i- ........-.-... & .--I.----... -..+ -

- i - -

- - - 1

4 -...--------- i ..-.. ..-....... j. - j

j

....................... -

i.- i n j -

b j - ............................................ -̂ ---*.-&.I..---. + I.-

- i

1 0 f

- - - - i ;

!

i -

n j j j - -

j 0 - - - - n j i ....................... A! -. ...... A -....-..... ....... L .................... i ;.-

0 -

I l l I l l I l l I l l L

Flite:2 Temp:40C Mode:El & E2

C C C C C C C C C C

+

+

1010

1020

C C

2000

C C C C

DNSC0DE.F ana lyse^ the PLS Prototype daul --1---___-____1_--------.---------------

Author: Ognem Nastov , MlT undergraduate, Nov 1987

lease cbes a least s q u m approximation and minimizes the round-of errors. n = #of x-y pairs A - (1 / sum(t**2)) * sum (t*y) B = (sum(y) - A*sum(x)) / n where t(i) = x(i) - sum(x) / n

subroutine leastsq (filename, depth, a, b, sqd)

integer depth, dn real a, b,mv,x, y,sumx,sumy,

charact198 filename s u m = 0 sumy = 0 sumtsq = 0 sumty = 0 open (1, filefiename, fonn=’fmatted‘) rewind (1) do 1010 counter .I 1, depth

read (1, *) dn, mv

Y - 1% (mv)

sumy = sumy + y u(int(counter)) = x v(int(counter)) = y continue

t(i) = u(i) - sumx / depth sumtsq = sumtsq + t(i)*t(i) sumty - sumty + t(i)*v(i) continue

a = sumty I sumtsq b = (sumy - a*sum) / depth call sqdev (a, b, depth, u, v, sqd) close (1)

end

--------_I---___---_--------------------

t(120). u(120), v(120), sumtsq, sumty, sqd

x = 255 - dn s w l u = s u m + x

dol (nOi= 1,depth

rem

Sqdev calculates the square deviation by defmition.

subroutine sqdev (a, b, n, u, v, sqd)

real a, b, u( 120). v( 120), sqd sqd-0 cb2000i= l , n

sqd=sqdIn rem end

inTeger n

sqd = sqd + (a*u(i) + b - v(i))**2 continue

The MAIN program body. It produces one line of output for each input file. The program uses a special driver fie

integer pairs real a, b, cl, c2 characteP8 lilename characteP4 mvr

dns-driver.dat.

open (2, file-’dnsdIiver.dat’, fonn-’formatted’)

dna-cod.. f Ned Dec 2 16:52:05 1987

C 50

100

200 +

open (3, file=’dns-output.dat’, form=’formaaed‘) rewind (2) print *, ’Creating t?le: ’, ’dns-output.dat’, ’...’ read (2, *) fhame, mvr, pairs print *, ’Processing file: ’, filename, I...’

call leastsq (filename, pairs, a, b, sqd) c l = exp (b) c2 = a write (3, fmt=200) filename, mvr, c l , c2, sqd, pairs if (filename .ne. ’pe22.dat’) go to 50 close (2) close (3) print *, ’File: dns-outputdat created.’ farmat (a8, a5 a 4 ~ 2 , g12.4 tr2, g12.6, tn,

end g11.6, i3)

2

~ ~~~

lUV8-COd.. f Sun S a 10 19:58:18 1988 1

C C C C C C C C

+

+

500

600

+ +

C C C C C C

+

+ +

MVS-C0DE.F analyses the PLS calibration data for both flites.

Auhm. - Nastov, ha undergraduate, Nov 1987

leastsq das a least square approximation I sum(x**2)A + sum(x)B = sum(xy) I sum(x)A + nB = sum(y) where n = # of x-y pairs

subroutine leastsq (fiename, voltrange, startline, endline, a, b, sq4 depth)

integer channel, startline, endline, skip, depth real a, b, mvu, mvl, VI, vu, hvdcout, x, y, sumxsq, sumx, sumxy,

sumy, sumysq, sqd character*6 voltrange character*l2 filename

-------------_-___--__________________I_--------------

-_---------_____---___________I_________--------------

sumxsq = 0

sumxy - 0 sumy = 0 sumysq = 0 open (1, file=filename, form=’formatted’) rewind (1) skip = startline - 1 do 500 counter = 1, skip

read (l,*) channel, hvdcout, mvu, mvl, vl, vu continue

depth = endline - startline + 1 do 600 counter = 1, depth

S U ~ X = 0

read (1, *) channel, hvdcout, mvu, mvl, vl, vu if (voltrange .eq. ’lower’) then

else

x = mvl y = VI

x=mvu y = vu

endif

sumn = s u m + x sumnsq = sumxsq + x*x

sumxy - sumxy + x*y sumy = sumy + y sumysq - sumysq + y*y continue

call solvesys (sumxsq, sumx, sumxy, s u m , depth, sumy,

call sqdev (a, b, depth, sumx, sumy, sumxy,

close (1) rem end

a, b)

S W S ~ , ~ u m y ~ , Sqd)

lsq2 does a least square approximation and also minimizes round-of errors n = #of x-y pairs A = (1 / sum(t**2)) * sum (t*y) B = (sum(y) - A*sum(x)) / n where t(i) = x(i) - sum(x) / n

subroutine lsq2(filename, voltrange, startline, endline, a, b, sqd sqda, sqdb, cov, depth)

integer channel, starthe, endline, skip, depth real a, b, mvu, mvl, vl, vu, hvdcout, x, y, s u m , sumy,

t(201, ~(201, ~(201, sumtsq, smv, sqd, sqda, sqdb, cov, del, sumxsq

c h a r a c e 6 voltrange c h a r a c e 12 filename sUmx=o

ma-coda . f Sun Jan 10 19:58:18 1988 2

lo00

1010

1020

C C

2000

C C C

sumy = 0 sumtsq=o sumty=o sumnsq-0 open (1, fibfdenune, form=’formatted’) rewind (1) skip - stanline - 1 do 1 W counter = 1, skip

read (1. *) channel, hvdcout, mvu, mvl, vl, vu continue

depth - endline - startline + 1 do 1010 counter = 1, depth

x = mvl

read (1, *) channel, hvdcout, mvu. mvl, vl, vu if (voltrange .eq. ’lower’) then

else y = vl

x=mvu Y - V U

endif

sumy - sumy + y sumxsq = sumxsq + x*x u(int(c0unter)) = x v(int(c0unter)) = y continue

t(i) - u(i) - sumx / depth sumtsq = sumtsq + t(i)*t(i) sumty = sumty + t(i)*v(i) continue

a - sumty I sumlsq b = (sumy - a*sumx) I depth

~ q d 2 (at b, depth, U, V, Qd) close (1) del = depth * sumxsq - (sumx)**2 Sqdb

cov = - sw / del * Wd *depth I (depth - 2) return end

sumx - s u m + x

dolOZOi-1,depth

~ f l ( S U ~ X S ~ / del * Qd * depth I (de# - 2)) sqfl (depth/ del * Sqd * depth/ (depth - 2))

sqd2 calculates the square deviation by definition

subroutine sqd2 (& b, n, u, v, sqd) integer n

wd=O do20(10i= l , n

sqd=sqd /n return end

sqd - (1 I n) * sum(Ax + B - y)**2

e b, u(20), v(20), Wd

sqd - sqd + (a*u(i) + b - v(i))**2 continue

solvesys solves 2x2 linear systems using determinants

subroutine solvesys (al, bl, cl, a2, b2, c2, x, y) integer b2 real al, bl, cl, a2, c2, x, y, det, dew, dety det = al*b2 -a2*bl dew = cl*b2 - c2*bl dety = al*c2 - a2*cl X - dewdet

a1 *x+b 1 *y=cl a2*~+ b2*y=~2

y = detyldet return end

C C C

+

+

C C C C C C

+

C 50

+

+

+ +

C

+

+ + +

100

200 +

sqdev calculates the square deviation expanding

subroutine sqdev (a, b, n, s u m , sumy, sumxy,

integer n real a, b, s u m , sumy, sumxy, sumxsq, sumysq, sqd sqd = a*a*sumxsq + n*b*b + sumysq + 2*a*b*sumx - sqd = abs (sqd t n) renun end

the sum sqd = (1 t n) * sum(Ax + B - y)**2

sumxsq, sumysq, sqd)

2*a*sumxy - 2*b*sumy

The MAIN program body. It produces four lines of output for each input file. The program maks use of the lsq2 and sqd2 subroutines. It uses the file mvsdriver.dat to get the info

essential to analyse only the appropriate chunks of data.

integer stll, edll, stl2, edl2, pairsl, pairsu real al, bl, sqdl, sqdal, sqdbl, covl,

character*6 voltrange character*12 filename open (2, file=’mvs-driva.dat’, form=’formatted‘) open (3, file=’mvs-outputdat’, form=’formatted’) rewind (2) print *, ’Creating file: ’, ’mvs-outputdat’, ’...’ analyze the first channel range. .. read (2, *) filename, stll, edll, sU, edl2 print *, ’Processing file: ’, filename, ’...’ voltrange = ’lower’ call lsq2 (filename, voltrange, stll, edll,

voltrange = ’upper’ call lsq2 (filename, voltrange, stll, edll,

au, bu, sqdu, sqdau, sqdbu, cow, pairsu) write (3, fmt200) filename, ’chrl’, ’mvl-vl’, al, bl,

sqdl, sqdal, sqdbl, COVL pairsl write (3, fmt=u>o) filename, ’chrl’, ’mvu-vu’, au, bu,

sqdu, sqdau, sqdbu, cow pairsu analyze the second channel range. .. if (stl2 .eq. 0) go to 100 v o l m g e = ’Iowa’ call lsq2 (filenam, voltrange, stl2, d 2 ,

al, bl, sqdl, sqdal, sqdbl, covl, pairsl) voltrange = ’upper’ call lsq2 (filename, voltrange, stl2, edI2,

au, bu, sqdu, sqdau, sqdbu, covu, pairsu) write (3, fmt200) filename, ’chr;?’, ’mvl-vl’, al, bi,

sqdl, sqdal, sqdbl, covl, pairsl write (3, fmt=200) filename!, ’chn’, ’mvu-vu’, au, bu,

sqdu, sqdau, sqdbu, covu, pairsu if (filename .ne. ’fZ-rlO-e2.dat‘) go to 50 close (2) close (3) print *, ’File: mvs-output.dat created.’ format (a12,tr2, a4,tr2, a6.92, B.5, tr2, g12.6, tr2,

end

au, bu, sqdu, sqdau, sqdbu, covu

al, bl, sqdl, sqdal, sqdbl, covl, pairsl)

g11.6, fr2, g12.6,92, g12.6, tr2,g12.6 fr2, i2)

~

-8-tC-Cod..f Sun Jan 10 20:02:28 1988

C C C C

C C C C C C

+

+

+ +

500

lo00

1010

1020

MVS-TC-CODEF

lsq2 does a least squan approximation and also minimizes round-of errors n = # of x-y pairs A = (1 / sum(t**2)) * sum (t*y) B = (sum(y) - A*sum(x)) / n where t(i) = x(i) - sum(x) / n

subroutine Isq2(filenames, fienum, voltrange, startlines, endlines, a, b, sqb sqda, sqdb, cov, totdep)

integer channel, startlines(6), endlines(6). skip, depth(5). curfie, totdep, filenum

real a, b, m u , mvl, vl, vu, hvdcout, x, y, sumx, sumy, t(100). ~(100). ~(100). sumtsq, sumty, sqb sqda, sqdb, cov, del, sumxsq

characteP6 voltrange character*12 filenames(6) s u m = 0 sumy = 0 sumtsq = 0 sumty = 0 sumxsq = 0 curfrle = 1 totdep - 0 open (1. file=fienams(curfie), form='formatted') rewind (1) skip = startlines(curfie) - 1 do lo00 counter = 1, skip

read (1, *) channel, hvdcout, mvu, mvl vl, vu continue

depth(curfi1e) = endlines(curf-e) - startlines(cdie) + 1 do 1010 counter = (totdep + 1). (totdep + depth(curfie))

read (1, *) channel, hvdcout, mvu, mvl, VI, vu if (voltrange .eq. 'lower') them

else

x = mvl y = vl

x-mvu Y-W

endif

sumy = sumy + y SumXsQ = sumxsq + x*x u(int(counter)) = x v(int(munter)) = y Continue

SUIlu = sumx + x

close (1) totdep = totdep + depth(curfi1e) curfie = curfie + 1 if (curiile .le. filenum) go to 500 do 1020 i = 1, totdep

t(i) = u(i) - sumx / totdep sumtsq = sumtsq + t(i)*t(i) sumty = sumty + t(i)*v(i) continue

a = sumty/ sumtsq b = (sumy - a*sumx) / totdep call sqd2 (a, b, totdep, u, v, sqd) del = totdep * SumXsQ - (sumx)**2 qdb = ~ q r t ( s u m ~ ~ q / del * q d to&p/ (totdep - 2)) qda= sqrt (totdepl del sqd * totdep/ (tow- 2)) cov - - sumx/ del

1

~ -~ ~

-8-tC-C0de.f S u n Jan 10 20:02:28 1988 2

C C

2000

C C C C C C C

+

+

so

100

300 +

200

+

+

+ +

C

+

return end

sqd2 calculates the square deviation by definition

subroutine sqd2 (a, b, n, u, v, sqd) integer n real a b, ~(201, ~(201, sqd

do2000i- l , n

s q d = s q d / n return end

sqd I (1 1 n) sum(Ax + B - y)**2

sqd = 0

sqd - sqd + (a*u(i) + b - v(i))**2 continue

The MAIN program body. It combines the files that are corresponding to the same temperature, and produces four lines of output per each combination. The program uses two driver files: mvs-driver& and mvs-tc-driver.dat The lsq2 subroutine has been modified in order to be able to combine several files for analysis. integer stll(6). ed1(6), stl2(6), edl2(6), pairsl, pairsu,

filnum, i real al, bl, sqdl, sqdal, sqdbl, covl,

au, bu, sqdu, sqdau, sqdbu, covu charactex*6 voltrange character*12 filenames(6), Mile, file open (2, file’mvs-driver.dat’, fonw’formatted‘) open (3, file’mvs-tc-out.dat’, fonn=’formatted‘) open (4, file’mvs-tc-driver.dat’, form-’formatted’)

i = O read (4, *) mfiie i = i + l read (4, *) filenames(i) if (filenames(i) .ne. ’*’) go to 100 f inumai- 1

d0200i=l,filnum rewind (2) read (2, *) file, stll(i), edll(i),

stl2(i), edl2(i) if (file .ne. filenames(i)) go to 300 continue

pint *, ’Processing master-fie: ’, dile, ’...’ voluange = ’lower’ call lsq2 (fdenames, fdnum, voltrange, sttl, edll,

al, bl, sqdl, sqdal, sqdbl, covl, pairsl) voltrange = ’upper’ call lsq2 (filenames, filnum, voltrange, stll, edll,

au, bu, sqdu, sqdau, sqdbu, covu, pairsu) Write (3, fmt=U)OO) die, ’chrl’, ’mvl-vl’, al, bl,

sqd, sqdal, sqdbl, covl pairsl write (3, frnb2000) mfiie, ’chrl’, ’mvu-vu’, au, bu,

sqdu, sqdau, sqdbu, covu, pairsu

analyze the second channel range ... volaange = ’lower’ call lsq2 (filenames, filnum, voltrange, stl2, ed2,

volaange = ’upper’

if (stl2(1) .eq. 0) go to 500

al, bl, sqdl, sqdat, sqdbl, covl, pairsl)

ma-tc-c0de.f Sun Jan 10 20:02:28 1988 3

call lsq2 (filenames, filnum, voltrange, s d 2 , edl2, w, bu, sqdu, sqdau, sqdbu, con, pairsu)

write (3, ht=u)oo) d i e , 'cM', 'mvl-VI', al, bl, sqdl, sqdal, sqdbl, Covl, pairs1

write (3, fmb2OOO) d i e , 'chn', 'mn-vu', au, bu, sqdu, sqdau, sqdbu, cow paifsu

close (2) close (3) print *, 'File: mvs-tcamlat created.' format (a12.02, a4.92, a6, tr2, B.5. tr2, g12.6, tr2,

end

+

+ +

500 if (mfile .ne. 'f24~2.dat') go to 50

2000 + g11.6, h2, g12.6, h-2, g12.6, tr2, g12.6, n?, i2)

Massachusetts Institute of Technology M E M O R A N D U M Center for Space Research

September 24, 1987

VOYAGER MEMORANDUM # 161:

From: Ralph L. McNutt, J r . m To: Voyager Internal Subject: Modulator Calibrations (Modcal or MVM): Interpretation

NORMAL OPERATION

As part of the routine calibration sequence on the Voyager Plasma Science (PLS) experiment, then is a digitized readout of the low voltage signal which is proportional to the potential on the modulator grids. This is accomplished via a set of Modulator Voltage Monitors; these measurements are referred to by MVM on Summary or EDR tape logs. In many of the engineering notebooks they are referred to as "Modcal" measurements or mode (as compared to "Curcal" or cumnt calibration measurements. In producing the modulator voltage, the voltage is stepped through one decade from 60 to 600 volts and then through the same steps with a extra gain of x 10 switched into the circuit to sweep from 600 to 60oO volts. To this voltage is added an offset of 50 volts. Hence, the L and M modes sweep from +10 V to +5950 V and the E2 mode sweeps from -10 V to -5950 V (the El mode sweeps from -10 V to -- 140 V by using a different step size). Thus, the true voltage ranges across 2.8 decades.

The voltage monitors measure a voltage which is a constant fraction of the voltage swept which does not include the factor of 10 gain change. The monitors measure a voltage between 0.1 and 10 volts which is then input into a fast A/D converter. This voltage is then converted to a binary number from which the highest order bit is discarded. Thus, as the low voltage sweeps between 0.1 and 1.0 the digital output sweeps from 0 to 255; as the voltage sweeps from 1.0 to 10.0 volts, the output again sweeps from 0 to 255, although retention of the highest order bit would give a digital output from 256 to 5 11. Hence, the voltage monitor output can correspond to 4 orders of magnitude although, again, the factor of 10 gain change is not explicitly incorporated in this measurement. To know which decade is actually being read o u ~ one must either rely on a "proper sequence" (the case for nominal operation for which the MVM readout presumably yields voltages close to the nominal values) or an "educated guess" (for interpreting the MVM readouts from the PLS !nstrument on Voyager 1 after the failure).

!

NOMINAL VOLTAGE LEVELS

The nominal conversion between channel number (or "step number") is employed in all analysis (both in the VGRANL and MJSANL software packages). The conversion between

.. .

M.LT. Room 37-635, Cambridge, Massachusetts 02139 617-253-7396

channel number and voltage is discussed in Bridge et al. [1977]. Let n be the channel number, which can take on values of 1 to 16 for the El, E2, and L modes and 1 to 128 for the M mode. Let M be an integer with values of 8 for the E2 and L modes, 32 for E l and 64 for M. Then the potential +n at the lower edge ("bottom") of channel n (also referred to7 below as VL, w), which is equal to the voltage at the upper edge ("top") of channel n + 1 (also referred to below as Vu Hv), (the upper edge potentials follow from considering "chan-

2 '

nel" 129 for the M mode and '17 for the other modes), is given by i

(1) where en is in volts. This formula only yields magnitudes, of course, the actual potentials in the E l and E2 modes being negative.

It is important to note that the 50 volt offset makes the channel spacing only quasi- logarithmic; at lower channel numbers, the channel widths are larger than if they were loga- rithmically spaced.

e,, = 60 - 50

NOMINAL FAsT A/D VOLTAGE CONVERSION

Information on the analog-to-digital converter is contained in Anton Mavretic's lab note- book labeled book 2, number 254 (starting date of July, 1974). On page 142 (dated 7/25/75), we find for analog voltage v running from 0.1 to 10, the digital count number N running from 0 to 127. Anton gives the constitutive relation as

(127 - N) T - s v = VREF e

and values VREF = 10 V, T =fd = 11-57.6 lcHz _=17.36 ps, z = 482.55 ps. Transforming to base 10, this yields - ->

(127 - N ) 64.oooO8 v = 10 x 10

where the denominator in the exponent is usually approximated as the integer 64. On page 143, formulas arc given for thefastA/D conversion. Anton lists

( 2 5 5 - N ) T - , 1.0 < v c 10. s v = VREF e I

and 1 - ( 2 5 5 - N ) T

v = VR,p/ 1Oe ' , 0.1 < v < 1.0.

with T =fi& AD = 1/ 230.4 lcHz = 4.34 ps, r = 482.55 ps. Transforming to base 10, this yields the combination

(255 - N)

v = 10 x 10 256 , 1.0 < v c 10.

and -GL!9

v = 10 256 , 0.1 c v < 1.0.

Although not explicitly stated, these arc apparently the nominal, rather than measured,

- 2 -

conversion formulas.

NOMINAL MONlTOR OPERATTON

In the u s u a l d m mode, the data numbers @N) produced by running the low voltage monitor outputs through the-r are returned as a function of channel number. In the L and M modes, the DNs corresponding to V, are returned in the space occupied by A cup plasma measurements in the usual plasma measurement mode; similarly, the DNs corresponding toV,are returned in the space occupied by the B cup plasma measurements. The C cup and D cup contain currents corresponding to normal plasma currents but with DN values obtained from the fast AID convener. Hence, for the L and M modes, there is a measure of the contiguity of neighboring channels (gaps and/or overlaps in coverage).

For the E l and E2 modes, there are only 16 slots available in the data sequence for each mode, so compromises were made in deciding which DN values would be included in the telemetry stream. For odd channel numbers (counting from 1 to 16), the DN corresponding to

L i s returned. For even channel numbers, the DN corresponding to Vu is telemetered. Hence, both absolute value and contiguity are checked, but only for eveq=channel. For example, there are MVM measurements for VLl, V,, V u , Vu,, Vu, etc., but there is no Cali- bration information for the boundaries between channels 1 and 2, 3 and 4, etc.

In nominal operation, note that we should have Vu, = V,, n+z. Indication of where the VG and V,s come in the data is given in the VOYPRT appended to Voyager Memorandum 158 as Figure 6.

c_

-

RELATIONSHIP BETWEEN MODULATOR POTENTIAL AND MONITOR VOLTAGES

Figure 1 (from 3. Binsack) shows a simplified block diagram of the high voltage modulator circuit for the PLS instrument; indicated on the figure is the location of the suspected failed circuit in the Voyager 1 PLS instrument. The low voltage monitor voltages are measured at the point indicated with the m w and labeled "VB = CAL. This low voltage is (nominally) directly proportional to the high voltage applied to the modulator grids (at least in the linear r e g h e of the amplifiers). The DC gain of the Cockcroft-Walton amplifiers ("C. W.") and the AC gain of the transformer in the feedback loop arc unknown as of this writing, so a theoretical value of the proportionality factor is not known (but could be derived from the appropriate circuit diagrams, J. Binsack, private communication, 1987). Data shown in Appendix A (see below) suggests that this factor is - 60.

The s-ubcontractor responsible for the high voltage power supply, Matrix Research and Development Corporation, measured both the voltage to ground from the modulator grid leads and the low voltage monitor outputs for all three modulators (on SNOO1, SN002, and SN003) for all channels in all modes. 0 Work has begun to combine this information with other information to develop an algorithm to link the measured DNs in the MVM mode with the actual modulator potential. It is important to note that all modulator potential measurements were actually made with high impedance probes to the modulator grid leads, terminated with capa- citances to approximate the electrical characteristics of the grids themselves (R. Butler, private communication, 1987). In no case wen potential measurements made on the grids themselves with the grids in place in the cups; this is not possible as a probe cannot physically be

- 3 -

inserted into the assembly. Before the PLS instruments were to de f ived to JPL for integration with the Voyager

spacecraft, extensive calibration tests wen at the MIT Laboratory for Space Experiments using the Bench Checkout Equipment @a). The modulator was stepped through all channels in all modes and the DNs (or "decimal counts") were recorded as a function of step number, equal to the channel number minus one (channel 1 corresponds to step 0).

In Appendix A, the reference counts are fisted as a function of step number for each mode (labeled for the prototype rather than for one of the flight units). Corresponding voltages from the voltage monitors are also given. Note that these are the counts measured with the BCE. For example, the nominal value of V', for step 0 is +10 V in L and M modes and -10 v in El and E2 modes. The reference counts for the L, El, E2, and M modes are,, respectively, 54, 48, 50, 48. The fraction values for VL arc, presumably, the measured low voltage monitor outputs (they are not nominal values as the nominal value should be the same in for all of these). The applicable equation relating the monitor voltages and the reference counts is (4b). However, applying (4b) to values of 0.172, 0.163, 0.166, 0.163, yields DNs of 59.3, 53.3, 55.3, and 53.3, respectively. This suggests that the theoretical A/D algorithms are off (if the reference counts and monitor voltages were really measured simultaneously). Note that (4b) is off by about a factor of 16RM = 1.046, but not quite.

The agreement is better at the other end of the scale for V u of step 15 of E2 and L and step 127 of M, the counts are 248 and the monitor voltage is 9.314 V. Using 9.314 volts and equation (4a) yields a DN of 247.1. A m the counts and monitor voltages were measured at the same time (which is not known), then is some nonlinearity in the A D converter, not

account for the discrepancy. - I have searched through all of Anton's notebooks stored in Herb's office and through all of the files in N-52 and have not been able to locate any documentation (other than that men- tioned above) on the performance of the A/D converter. To put all of this in perspective, at the higher voltages in the M mode, the reference counts for Vu and V' arc separated by 4 DN. This implies that a determination of the true potentials on the modulator grids to an accuracy of better than 1% should, in principal, be possible.

Attached as Appendix B are similar modulator calibration data from the Flight 1 unit (on Voyager 1, i.e., SN002). Most of the measurements are dated 4/22/77, less than 5 months before the launch. Unfortunately, shultanmus high voltage, monitor voltage, and DN were not recorded. It is worth noting that for step 127 of the M mode, the measured value of Vu was 6010 volts, as compared to a nominal value of 5950 volts, a deviation from nominal of 1%. This suggests that such deviations are not uncommon.

7 z t l y acmunted for. A spot check suggests that-a simple offgef'voltage also doesxt-- -- _ - - --

MODULATOR VOLTAGES FOR THE VOYAGER 1 PLS INSTRUMENT AFTER THE FAILURE

The assumed circuit failure on the Voyager 1 PLS instrument pegged the value of VL at some value greater than 6OOO V, and caused the values of Vu to increase with step number, but in a spurious way (J. Binsack, private communication, 1987), Le., the phase of the modu- lated current is shifted 180°. Hence, the solar wind can no longer be detected in the normal plasma mode because the current is detected synchronously, and the failure introduced the incorrect phasing.

- 4 -

If the value of VL is pegged to a large value, then the solar wind is always excluded from the detector for half of a measurement. If the value of Vu is low enough, then the solar wind should be admitted for the other half, unless Vu steps to a large enough value to also exclude the solar wind.

The DC return measurements from the Voyager 1 instrument after the failure show changes in all 16 channels which vary with time; this presumably gives some measure of the solar wind flux (see Voyager Memorandum 157). There is a "break" in the DN numbers in the DC r e m mode above channel 8, suggesting that Vu for the first 8 channels (in the L mode) is always low enough to allow the solar wind access to the collector plate. The DNs decrease for the top 8 channels suggesting that more of the solar wind is gradually being excluded, but the lack of variation in this pattern could mean that this is another instrumental effect instead. The interpretation is also complicated by the fact that there is a factor of 10 gain change internally in going from the voltages for channel 8 to those of channel 9, so the break at this point may actually be related to this change.

From the L mode spectrum in Figure 6 of Voyager Memorandum 158 (referred to previ- ously), the MVM data numbers change from 190 to 93 for the first 8 channels of the L mode (at 1980-346/0103:23.735) with a "fold' occurring between channels 2 and 3. If the DNs can be interpreted as before the failure, then the fold could occur at -6, 60, 600, or 6000 V. The value of 6OOO V must be excluded on the basis of the detection of the solar wind. At channel 8, the DN of 93 could be -14 or 140 V; higher values are again excluded because the solar wind is detected.

In the top 8 channels, the DNs run from 90 to 136. If no "folds" are present (and none arc apparent), this could correspond to -140 V to 210 V or some multiple of 10 times these limits. These potentials would allow the solar wind to be admitted in all channels, but for speeds of 400 W s , similar to those detected by Voyager 2 at these distances, no decnase in DC current with channel number should be present. Multiplying the limits by a factor of 10 would exclude the solar wind from the top channel, unless its speed exceeded -600 W s , and speeds of this magnitude were not detected just prior to the failure, so something is not con- sistent.

To understand the post-failure measurements by the instrument on Voyager 1, a more detailed investigation is clearly requind. Such an investigation is now underway.

li

- 5 -

ORIGINAL PAGE IS OE POOR QUALIm,

APPENDIX A

MVM Measurements for Protatype

1 OF POOR QUUIm DEC. COUNTS

P LS -r 2 U Y b T YJ? E H I G H VOLTAGE MONITORS

L-Mode S.C. = 960 ms

PLS-PROTOTYPE H I G H VOLTAGE MONITORS

El-Mode S.C. = 960 ms

. .

- .. . . . . . _ . - . _ . - .

- .

PLS -P ROTO’TY P E HIGH VOLTAGE MONITORS

DEC. COUNTS E2-Hdde S-C. = 960 ms

DEC. COUNTS

PLS-P ROTOTYPE HIGH VOLTAGE MONITORS

M-Mode S.C.' = 960 ms 1

PLS-PROTOTYPE HIGH VOLTAGE MONITORS

+Mode

.

S.C. = 960 ms

. . 1

PLS-PROTOTYPE H I G H VOLTAGE MONITORS

DEC .

DEC COUNTS

L d3-1. - . V L . U A L r I

HIGH VOLTAGE MONITORS

M-Mdde

4

S . C . = 9 6 0 ms

5 H I G H VOLTAGE MONITORS

DEC. COUNTS M-Mode S.C. = 9 6 0 ms

APPENDIX B

MVM Measurements f o r Flight 1 (SN 002)

O O t ORIGINAL PAGE IS OF POOR QUALITY *&

DEC. COUNTS L-Mode w4 S.C. = 960 ms fi.1

0

1

2

3 I38 130

4 -20 2 I38

5

280 4 b I -

7 '

8 .

9

12

151 . 1 182 I 13 ~

14

15 I 184 21s

I

I I i

4 -

ORIG~NAL PAGE IS PLS- QA.deOo7 HIGH VOLTAGE MONITORS

OF POOR QUALITYi

DEC. COUNTS L-Mode S.C. = 960 ms

PLS-- % 4 k O O ; L HIGH VOLTAGE MONITORS

DEC. COUNTS S . C . = 960 ms

t

DEC. COUNTS M o A K , ~ ~ EZ-PIode/ / S.C. = 960 ms 7 - 1

I f R e f e r e n c e C o u n t s

L

/L 7'

'VL 'VU "?$y -Va 4 t u b iep -

0 050 -IO -30 ..

1 240

2 241

f I

4 086 ' 3 93y 20342

5 166

1 3 I

I I I I

I I

I

(sl

2 3 3 2 i3cf

I

DEC. COUNTS

P L s - n O O Z HIGH VOLTAGE MONITORS

M-Mode S.C. = 960 ms

t

4[ L"( 7- 1

t P L S - w W & O o 2

H I G H VOLTAGE MONITORS

P L S - m E * & H I G H VOLTAGE X O N I T O R S

DEC. COUNTS M-Made 4eWJ

9!!b”/n ‘3

S.C. = 960 ms

P L 3 - m SA*& @ o L HIGH VOLTAGE MONITORS

PLS- %&mL #

H I G H VOLTAGE FIONITORS

DEC. COUNTS h e a s a d M-Mode S.C. = 960 ms

Analysis of Pre-Flight Modulator Voltage Calibration Data for the … · 2013-08-30 · L Analysis of Re-Flight Modulator Voltage Calibration Data for the Voyager Plasma Science Experiment

Documents