Text preview for : Julie_Research_Labs_VDR1067_VDR307_Kelvin_Varley_Voltage_Divider_Operator_Manual.pdf part of JRL Julie Research Labs VDR1067 VDR307 Kelvin Varley Voltage Divider Operator Manual . Rare and Ancient Equipment JRL Julie_Research_Labs_VDR1067_VDR307_Kelvin_Varley_Voltage_Divider_Operator_Manual.pdf



Back to : Julie_Research_Labs_VDR10 | Home

INSTRUCTIONS FOR USE OF

PRIMARY STANDARD

VOLTAGE DIVIDER

VDR-106/? and VDR-307




This instruction manual applies to the other Julie Kelvin-Varley

Dividers except for impedance level, specific ratios or other

special modifications (such as \lDR-307 H, VDR-327 J, VDR-327 P).



For these special dividers, the individual specification sheet will

be an addendum to this instruction manual.
INDEX



SECTION TITLE PAGE

I. Description

vDR-106/? I

VDR_307 2




II. Specifications



III. Connections



IV. Applications

A, Resistance Measurement 5

B. Voltage Measurement 7


C. Currenr Measurement I
D. Operation as a Voltage Divider IO




V. Calibration IO




\,T, Maintenance 11




VII. Precision - Vol. IV No. I 12




VIII. N. B. S. Letter l:l
I. DESCRIPTION




The VDR-106/7 is a precision voltage divider of the Kelvin-Varley

type. The two major characteristics of this instrument are a) a


constant input impedance of 100 kohms and b) an output voltage that

is a decimal part of the input voltage as indicated by the settings of the

seven decade dials of the divider. A ratio of output to input voltage of

unity is indicated by dial settings of 9-9-9-9-9-9-10.



The output indication has maximum accuracy under circuit conditions

which al1ow no current to pass through the output terminal of the circuit.

Leakage is held to such a small value that it does not appreciablv affect

the accuracy of the device.



Two buttrns are provided for the protection of the divider and the equipment

under test (such as standard celts). The "Io sensitivity" key inserts a


i meg resistor in the output circuit and the I'high sensitivity" key inserts a


I/l}A|n amp instrument fuse in the output circuit (its resistance being

about 100 ohms which has negligible effect on the circuit)-



A RATIO switch, when set in '?DIAL" position connects the output terminal

to the switching circuit of the divider. When set to rrltr. the output tcrminal

is connected to the input high terminal. When set to ttgtt, th output
terminal is connected to the input low terminal. The use of the rtor! and

rlrr positions i', standardization procedures will be discussed subsequently.
3?n
T)a
.a
., \,. I'A"6o
j"t'{ty9)
i tMEG


JO.-"




The VDR-106 is an older version of this divider with six decade dials

and no Ratio switch. In place of the Ratio switch is a "mediutn
sensitiVity" key which inserts a 150 kohm resistor in the output

circuit. Some of these dividers have a 27K resistor in place of the

1/100th amp instrument fuse. A11 older models can be modified to

conform to the specs for the VDR-106/?.



The \lDR-307 is not supplied with the Ratio Switch and has the "High"

?'Medium", and r!Lowt' Sensitivity keys as described.



'Ihe fact that the circuit employed in this instrument contains neither

potentiometers not verniers, that all decades are switched over the

entire ten million to one range, and the fact that the precision matched.

accuracv assuring decade resistors of the divider are oil-immersed

and hermetically sealed essentially ensures the permenance of accuracy

of this instrument.



Both styles divider are supplied for standard 19 inch rack mount.

The VDR-106/7 requires 14'' of rack height and is 6" deep. the VDR-307
requires 3 I/2" of rack height and is 6't deep.
II. SPECITTCATIONS




Range: Every integral ratio from 0.0000000 to 1.0000000



Resolution: Seven decades or ten miliion equal. discrete steps

of 0.0000001 each yield a I/I}th part per million

resolution.



Accuracy: Ratio indicated is accurate to 0. 000T(/o of full scale

(l PPM) absolute at 25"C with negligible power.



Stability: 0. 00017r, per year.




Maximum
Excitation: -
VDR-106/7 1.100 volts
VDR-307 - 700 volts




Overall
Resistance: 100.000 ohms + 0.01%



Temperature
Range: For rated accuracy 20"C to 30"C on VDR-106/7:

T. C. 0.000057c oC on VDR-30?
I I I. CONNECTIONS



These dividers are provided with heavy copper terminals that are either

silver or gold plated. If measurements are affected bv thermal EMF's
generated at these connection terminals, they will be minimized by
using similar connecting ieads. The "Input Lowr and "Output Low'r

terminals are internally connected permanently.



Heavy connecting leads should be used whenever lead resistance mav

become a factor in a measurement, if lead compensation is required.

the JULIE LEC-307 lead compensator is available. Care should be

exercised to assure proper insulation of all connecting leads so that

unanticipated leakage and inaccuracy is eliminated. The nul1 detector

used in precision measurement circuits must be adequatel.y insulated
to withstand such voltages as may be introduced between the circuit and

the case and keep leakages below such values as would introduce errors

to the measurements. Under some conditions it is adviseable to use a


batterv operated null detector that is effectively isolated from ground by

mounting it on a styrofoam pad. The use of guarding circuits on the null
detector is also of help in some instances.
IV. APPLICATIONS



A) RESISTANCE MEASUREMENT

1. Ratio or Relative Resistance

Refer to figure 1. V is a suitablv stable voltage source

and G is a galvanometer of adequate sensitivity. R1 and R2

are the equivalent elements of a ratio set such as a
potentiometer" a voltage-divider or any other conductirre

voltage-dividing network. For relative resistance measuremcnts.

R1 is the unknown and R2 is the resistor relative to which its
value is to be defermined. K is the dial setting.




P:Kt.000001
The relative resistance of Ry with resJrect to Re is given by
K l_ 0.000001
Ot: O, (
I-K + 0.000001 )

I Absolute Resistance may be measured bv replacing R2 by a


resistance standard of known absolute value. The percentage

error in the measured value of R1 at nuli is:

0.0001
(/c


K (1-K)

This expression has a rYriri;.rnum value at K - 0.5. hence. for

highest accuracy it is desireable that R2 be of the sarne order

or magnitude as R1. so that K may be some value near 0.5 at

nuli.
5
The following pl'rt (Fig. 2) of percentage error in R1 as a

function of the value of K at which null is achieved will be

useful.




(o .00loc/c
o
t- .0008o/c
GC
69
>tso
.0006%
b;
cc,
oo
Fig. 2
+E .ooo4c/"
J
Po
L(,
+c .ooo2"h
to
o9
-6
6s, .r .2 .3.4 .5 .6 .7 .8 ,9
fi.s Diol Sotting X




, Sensitivity:

Galvanometer sensitivity requiremcnts are dictated ltv thir

accuracv and the impedance levels involved:




GALVANOUITEA SNSITIVI'Y
REOUIREMEH;S FOR USE

E RUL CINCUIT IMPEOANCE


La



;i
2-i
li *'
!t
ot
2t

i!



otLEcrloN sNslrlvlrY
rlutrttta! {r ar! xa r@v alrrct trctT^rlQ
The curve of Fig. 3 indicates the required gaivanometer
current sensi'uivityfor desired scale sensitivity per 100 volts

of bridge excitation, for a range of external circuit impedance

of the device under test. This is the maximum irnpedance
presented by the device under test in series with the galvano-

meter and voltage source, not including the impedance of thc

VDR-106. The curve is plotted for the 50(/a ratio-setting of
the VDR-106 at which point it exhibits maximum internal

impedance. The voltage source is assumed to have negiigible

impedance. and the "high" sensitivity button on the VDR-1O(j

is assumed to be depressed.



4. Leakage:

Measurements to lppm accuracy require extreme carc to

prevent leakage errors, Leads should be short and teflon-

insulated or air-supported. Thc battery or other volilrge

source should be isolated from the galvanometer through a

minimum leakage resistance of I00. 000 megohms. Thc

bridge circuit is normally grounded at one of the galvanometer

Ieads.



B) VOLTAGE MBASUREMENT

1. Absolute DC Voltage Measurements - Method I

(for voltages. higher than that of a standard cell):

Refer to Fig. 4. V is the unknown voltage (for example. the

potential drop across a zener diode), E is the known EMF of

a standard cell or other reference voltage, and G is a
galvanometer of adequate sensitivity (see discussion altove).




Fis. 4




Where K is the diai setting, The uncertainty to which tr is
known must be allowed for, also, Il is interesting to note that.
with a primary standard ce1l set. such as the SCO-106 illustrated

below. and a value of V near 6 volts, the absolutc accuracv of

measurement is about 0.0003%.



2. Absolute DC Voltage N{easurement - Method II

(for voltages lower than that of a sLandard cell):

Refer to Fig. 5. V is the unknown (small) voltage. E is the
EMF of the standard cell. G is a galvanometer of adequate

,.#tr/1
sensitivity (see discussion above) B is a suitably stable
.\-j-.'


external power supply, R is a suitable rheostat. and S is a


SPDT switch.




Fig.5




PRIMARY-
STANDARD
AESO tUTE
VO TTAGE
REFERENCE
f hree primor'1 cells
in o !0 0l"C
oven. Eureou of
5tondords Cerli-
fied to l ppm
The measurement process requires two steps:

1. With S in position 1, K is set to the ratio between thc
standard cell EMF and some convenient voltage value.

Er such as 10 volts (B must be greater than l0 volts;
and 'tR?' is adjusted so as to null G. Note: By selecting

values of E" such as l0 volts. 100 volts. etc. K will be

direct-reading in the cell EMF, except for the decirnal

point.

2. With S in position 2. K is readjusted to null G. Assuming

negligible drift in Es during the readjustmenl of K. the

value of V is given by:

KZ
V=E +0.000002 (Es).
Kt

where K1 is the ratio setting employed in step 1. and

K2 is the ratio setting arrived at in step 2. The un-

certainty to which E is known must be allowed for. also.

If the SCO-106 is used for E (see below) and Es is set to
l0 volts. the total uncertainty in the determination of a

one volt I'V't is 2l microvolts.


C) CURRBNT MEASUREMENT

Current measurement are made by using a knowrr

resistance (shunt) in series with the current circuit

being measured and reading the voltage drop developed.

The curuent determined is calculated by Ohm's Larv and

the accuracy of the caiculation is determined by the
arithmetic sum of the error contributions of each

component in the circuit.



D) OPERATION AS A VOLTAGE D]\TDER ONLY

In the Kelvin-Varley circuit there are a number of shunt

paths, As a result. the unit may be used as a potential

divider only. The resistance between the "Output High"

terminal and either end of the divider ls nol a linear

function of the divider setting and the unit must not be

used as a rheostat or variable resistance device.



E) MEASUREMENTS AT ZERO ARM CURRENT

The divider calibration is done under zero arm current

conditions only. A current in the output circuit of

2x "
l0-A x divider current will, at worst condition.

introduce approximately 1 ppm error,




V. CA LIBRATION



The JULIE DI\TDER carries a 5 year warrantee on its linearity. It

will remain within I PPM of absolute linearity without correction. The

stabilif.v and accuracy of this device is such that ratio calibriltion can be

meaningfullv checked and correction factors be developed to use it for
.1 PPM measurements. Procedure for this calibration is detailed in

the JRL 'tPreci.sion'r Journal Vol IV. No. I which is included in this

manual. Enclosed also. is a letter from NBS in which thev statc their

approval of this method. I0
W. MAINTENANCE



The only field maintenance required on the voltage divider is replzrccment

of the instrument fuse in the high sensitivitv key. If the divider operates

on low or medium sensitivity but not on high, a blown fuse is indicated.
This replacement must be a l/lOOth amp LITTLEFUSE type: MICIIO

1/1004 P/T 278.0T0. In replacing the fuse in the divider. extreme care

must be used to prevent short circuits from occuruing due to bending

of air insulated busbars or resistor leads. If ratio of the divider is

far off from its proper value. a visual inspection of these lcads mav

reveal such short circuits. Gentle bending of the busbar to eliminatc

this type of problem is permissable. MORE EXTENSIVE REPAIRS NIUST

BE MADE AT THE FACTORY.




11
lJRilPRrorsrolt
.. . an aperiodic journal deuoted to the difficult art of one part per million

-
VOLUME IV NO. I JUNF 196I




ESTABLISHING RATIOS TO ONE PART IN TEN MILTION
HISTORY OF ULTRA.PRECISE RATIO MEASURE- INTRODUCTION
MENTS: 1956 to 196l
In previous issues we have discussed the O.OOO1% VDR-106
Primary Standard Voltage Divider and the DMR Series of
Julie Research Laboratories has been using the uniquely Decade Resistance Sets indcpendently. In this issue we
preci.se and stable NII-1* resistor design in DMR Primary will e'xpand the previous discussions by explaining how the
Standard Resistance Sets since 1955 and in VDR Primarv DMR Series Resistance Sets can be used to check ratio
Standard Dividers since 1956. accuracies with uncertainties of one part in ten million
This company has followed a policy of conservative rating (0.00001%)'and, concurrently, how the DMR Series may be
of these stanclards and instruments and has gone so far as used to verify the 0.0001% accuracy of the VDR-I06.
to develop nen' techniques for verifi.cation of the unusually
high accuracies specified for its equipment. RESISTOR ERROR
The information in this edition of Precision is presented in
the hope that it will facilitate the measurement and verifica- In Volume II, No. 2 of Precision, we explained the self-
tion of ratios by standards laboratories. checking feature of the DMR Series which allows determina-
*Patent applied for tions of relative accuracies of individual resistors to within



COLUMN A COLUMN B COLUMN C
RE SISTOR R ESI STANC E CUMULATIVE IATIO ERROR
NUMBER D EVIATION DEVIATION IN PPM
IN PPM IN PPM


6loKvloK7 I .I -t - 0.1

2 -l -0.2
0 -2 - 0.2

4 loK loK 9 4 +l -1 - 0.2
qn^o o-rnf o.l
CALIB. 5 0 -l - 0.2
3 loK l0K l0
o--rn M o-\nAo 6 +l 0 -0.t
2 loK IOK II 7 -1 -I - 0.2
o-rAAo o-\Nk 8 +2 +l + 0.0
t loK l0K l2
o-rM-o o-rAf 9 0 +l 0

t0 +l +2

ll +2 +4

t2 +l +5



FIGURE I FI GURE 2
DMR CALIBRATION CHART
Theoreticol Rolio to I port per 10,000,000




m
ur
u,
3
p
tL
one part per million by using two of the twelve resistors in to obtain ratio conections to an accuracv of the order of
the I)MR as ratio arms and one of the remainingten as a conF 0.00001% (one part in ten million).
parison standard.t We will use such data to establish the
ratio accuracy of a DMR-105 Resistance Set, By using the data of column A, Figure 2, ratio eror in parts
per million may now be calculated as follows:
Sirnilar resistance deviation data may be obtained with Accumulated sum of Resistance Deviation,
thr: use of any stabLe bridge with limited accuracy, but with ^bp t1 + a2 +
resolution capable of measuring resistance deviation to = '.. + ak fData shown as column B_]
=bk-l + ak (1)
within 0.0001%. In such a case the deviations of the indiv-
idual resistors are obtained, starting from a suitable initial Ratio Error (ppm) = 1 f bk -_1. (h ) I (2)
bridge balance. It is important that all bridge settings ex- nn
cept for the last place decade must remain fixed when inter-
comparing individual resistors in order to insure that the As an example, the ratio error of the first nine resistors in
deviations observed are significant, In addition all resistors the typical DMR-105 shown in Figure 1 is calculated in
should be measured twice in order to insure that the devia- Column C of Figure 2. The error calculation from equation
tion readings are significant to within one part per million. (f) for a ratio of J/9 is as follows:
Ratio Erro; = Ii9 -t -5/9 (+ 1) i = -0.2
Tvpical results obtained by either of the above methods are =
sirown in Column A of Figure 2. The deviations are in 0.01 These corrections may be calculated and tabulated in the
ohm units (ppm) for each 1O,OOO ohm resistor of the DMR- spaces left in the chart of theoretrcal ratios shown in
I U5 Sel. Figure 3, Note that the errors obtained for this typical
nine-resistor string are less than 0.2 ppm before correction.
fNote that in Figure 1, we are using a DMR-105 Set, com- Completion of the above chart for all values of ratio de-
prising trvelve 10,OOO ohm primary standard resistors. In this sired establishes a Primary Ratio Reference DMR-10.5
resistanc,-e range the combined residual ratio error intro- with ratio certainty of the order of one part in 1O,O0O,0OO.
duc'erl by interconnection wiring and insulation leakage
will tre less than one part in 10,000,0O0 (for a series string The temperature coefficient matchin!l and stability of a
of ten resistors) if resistance uncertainty is less than 0.01 DMR Set is such that in a typical laboratory environment, a
ohms anil leakage uncertainty is greater than 125,O00 single calibration will be usable for several months. When
megohms. Copper links are available for all DMR Sets required, recalibration by the method given is relativelv
which irerniit series connections with uncertainties of less fast and simple.
than 0.tlu1 ohms. External wiring for all tests should have
insulation of the quality used in Julie Research Laborator- APPLICATION OF THE PRIMARY
ies I.lr:sistance Sets and Primary Standard Divider VDR-1O6;
i.t'., t c,ranric, Teflon@, polystyrene or g1ass. RATIO REFERENCE
Maintenance of the O.OOO1% (one part per million) rated
accuracy of the Julie Research Laboratories Primary Stan-
THEORETICAL RATIO dard Divider VDR-106 is based on complete production and
measurement tests described in an earlier Precision Vol-
'lhe Dl\{R- 1O-5 Set permits series interconnections of its ume III, No. :1. A quick and independent check of this per-
twelve Primary Standard Resistors to obtain 66 ratios. formance is possible using the DMR-105 Set as a Primary
"I'lrese ratios are defined by the fraction r - k,/n where n is Ratio Reference. The test circuit i.s as shown in Figure 4.
tlie total number of input resistors in the string and k is the
number (less than n) across which the output voltage is To ensure that a combination of power supply leakage re-
developed. Additional ratios may be obtained by parallel as sistance and galvanometer leakage resistance does not
well as series connection of resistors, but for the purpose introduce substantial errors in the ratio determination, one
of this paper we will confine ouiselves to the simple series side of the galvanometer circuit is grounded as shown, and
connections. The 66 ratios are shown in Figure 3. It will be a battery and battery switch properly floating on polystyrene,
noted that approximately half of these ratios are distinct glass or Teflon insulation are used as a source.* With this
and give unusually good coverage of the ratio interval from circuit set-up it is only necessary to select a ratio, adjust
0.0833333 to 1.0000000. The remaining points are redundant the dials of the VDR-106 for no deflection on the galvano-
and afford an opportunity for multiple checking of the same meter when the battery switch is thrown and compare this
ratio point. figure to the ratio reference calibration chart of Figure 2.
The VDR-106 data is taken to one part in ten million by
recording the setting of the six dials and estimating the
seventh place from the galvanometer deflection.
RATIO ERROR _ DMR-I05
*A suitable test for leakage effects may be made with
The closely matched resistors of a DMR-105 Set usually any high range megohmeter. Before the cali.bration test
permit establishment of the ratios of Figure 2 to accuracies of a VDR-106, it is possible to make an operational
of 0.0001 to 0.0002% uithout need for coffection or calibra- check of the compJ.ete circuit by splitting the DMR-105
tion. Ilou,ever, we n'ish to show here how the resistance Set into two groups of six resistors each and intercom-
deviation data obtained in the earlier section may be used paring them in the circuit of Figure 4.
BATTERY FLOATING
TO Nrh RESISTOR (LEAKAGE RESISTANcE
r-- I 00,000 + MEGS )
i
L_J
t2

ll




L-___gtot_
oooooo
_J

TO KIh RESISTOR
NULL DET.


FIGURE 4
CONNECTION DIAGRAM - DMR-I05 ond VDR-I06
(Notc: All leods must be insuloted 100,000 megohms
or more obove ground. Teflon insulotion is suitoble.)



STABILITY OF ULTRA.PRECISE RATIOS
The stability of resistance ratios with temperature, voltage,
humidity and time is largely a function of the stability of the
basic resistors used and of the design of associated inter-
connections, insulation and switching components. primary
Standard Dividers manufactured by -|ulie Research Labora-
tories achieve unique accuracy and stability through the
use of the type NB-1 resistor and consistently meticulous
design of all associated components.

Of over 100 VDR Dividers in production and laboratory use
since July 16, 1956, only one has been returned to the
factory out of accuracy specification. Other units checked,
including Serial Nos. 2 and 16 maintained here at Julie
Research Laboratories, are still within one part per million
as specified in our literature. This complete absence of
drift prevailed despite instances of years of service under
extreme environmental conditions in production testing. It
should be noted that these units have no facilities for ad-
justments or recalibration. As with all JRL Dividers, no
adjustments are required to maintain the stated accuracy.

This reliability history permits Julie Research Labora-
tories to offer a factual Written Perfolmance_Gxarantee
of 0.0001% ratio accuracy for a five year period for the
Model VDR-106 Primary Standard Divider.
U. S. DEPARTMENT OF COTUVERCE
NATIONAL BUREAU OF STANDARDS


ADORiSS REPLY TO IN YOUR REPLY
REFER TO FILE NO
NarroNAL BUREAU oF fiANoAFros
1.1
WASHINGTON 25. D. C,


Gentlemen t


Refr Volrage Dlvlder
Our Tert No.
Your Order No.
I{e havc rccclved your order and/or the Voltage Dtvlder refarred to
above, whlch lnltrument tr belng rubnttted to ua for callbratlon.

The caltbrat,lon of rarto devlcea lnvolvee no reference to Natlonal
Reference standarda of any kind, auch ag standard reetatora. By
lntercompartng closely-matched reBlstors any user can hlmaelf
egtabl{ah accurately known reel-stance ratlos on a gcale ln etepe of
J, where N ls the number of reetstora connected ln sertea.
N

rn the past our ugual tesl on apparatug of thlg type hae tnvolved a
tLme-consumlng check of each poaltion of each dlal agalnat our caltbrated
preclelon ratlo devlce knosn ag a I'unlvereal Ratlo set". The fee for
thla teat varies frorn $154 to $181, dependlng upon the number of dtele
lnvolved. ltre purpose of thls letter ls to call your atEentlon to
the advantages of a slmpler, faeler, and leaa expehalve t.est vhLch,
whl1e not covering aa nany dtal settlnga, glvea reasonable aaaurance
regardlng the accuracy of adJurtment of the dlvlder. For thta tert
a $,rouP of ntne cloeely--otched recLrtors ts enployed to establleh
accurately knonn retlos on the scalc of nlnthe, L/9, Z/9, 3/9, ete.
lftc Natlonal Bureau of Standarde nll1 callbrate pronrpcly real.atance-
type voltage dlvider! on the acalc of ninthe under Teat Fee Schedule
2oL.Lo2z for a fec of SIl.* Ttre acele of nlntha ie particularly
veluable becauee cach poeitton of evary dial ie tnvolved at aoue
ttagc ln tha tert. At theee dlrcrate aettlngr .lll11r, .222222,
Gtc. thc accuracy of callbratlon lr rorncwhat better than vhet can
be reallzcd wlth our callbrated unlverr.rl ratlo devlee. For rrell-
adJurtcd dlvldere our tegt document ylli rtate the devlat{on obnenred
at the tlme of tert to the neareet dlgtt ln the rtxth declmal place.
rn your reply pleaae rtete lf you want the elmple $45 teet or rlre
ful1 teat
Slneerely youra,



Retlatance end Reectnnr.e SectLofl
Electrtclty Dlvl-al on




* 945.00
Clrcle 5-2727




lp.l JuLtt
211 wEsT
REsEARct.t LAB0RAToRtEs,
5rsr sTREET, NEw yoRK, N y. 10023
rNr
Jl\L



w ARRANTY




JULIE RESEARCH LABORATORIES, INC. guarantees to the original
purchaser that this instrument, Model \IDR 106/?, shall be free from
defects ln material and workmanshlp and shall malntain its rated accur-
:acy of 0. 000170 for a period of five (5) years after date of initial shipment.
Our liability is limited to repairing and replacing any defective part with
the exception of vacuum tubes, panel lamps, fuses, choppers, and
batteries. The warranty lapses, if upon our investigation, we judge that
the instrument has been abused in any way.




"Firsls in Accurocy"