Beruflich Dokumente
Kultur Dokumente
TECHNOLOGY
MD
//
Bulletin
OF THE
Bureau of Standards
S.
W.
STRATTON, Director
Volume
(Nos.
1,
2, 3, 4)
1907
ON THE GEOMETRICAL MEAN DISTANCES OF RECTANGULAR AREAS AND THE CALCULATION OF SELF-INDUCTANCE.
By Edward
B. Rosa.
1.
THE FORMULAE.
The formula
rectangular section
mutual inductance twice over the area of the section of the coil. If dxdy is an infinitesimal element of the section ABCD at P and dx' dy' is a second element at Q, the mutual inductance of the two circles of which these elementary areas are sections is given by the
following formula of Maxwell:^
(I)
where a
is
is
a^y
x
r,
is
the dis-
the distance
infinitesiinal sections, is
Jx^j^y\
we
by the area
is
of this
AxisofX
Yvx- 1
large ring
ABCD.
we then
dxdy
shall
ABCD
we
is
ABCD
on
which
is
The
current
sup-
Electricity
and Magnetism,
II, ^ 705.
Bulletin of the
Bureau of Standards.
[ I'^ol.
J,
No.
I.
it
may be
taken to
wire,
b}-
The
same
7i
space.
Hence by
inserting n^ as a
factor in the expression previously derived the formula for the self-
turns,
mean
radius
<7,
axial
width
tion
b to
by combining several terms which depend only on the ratio of c into two terms, y\ and y.^, the values of which could be taken
b}'
him
8^
Stefan's formula
is
S'^'
+A
+
the
1 6/^
^J
(2)
where
the
and
c are
coil as stated
c
above, a
of turns,
mean
y,^
radius,
;/
is
number
and y^ and
.v
are taken
,,
from the
table,
using
^ or
c
-7
as arguments,
x
is
This table
article.
is
When
stein's
Fig. 2
square Wein-
formula
much simplified
as follows,
where
c is the
L ^Trn Vjlog
Formula
nuich
'^
(3)
(2)
may
l)e
any
ratio;
formula
(3)
Neither
a.
is
Wied. Annalen, 21, pp. 353-354; 1884; Maxwell, felect. & Mag., 3d ed., II, p. 350. Wied. Annalen, 22, p. 107; 1884; aud Sitzungsbericlite Kais. Akad. der Wiss., Wien, 88 (2), p. 1201; 1883.
^
Rosa.'\
Geometrical
2.
Mean
Distances
and Self-Indttctance.
As
supposed to be distributed uniformly over the entire cross section of the coil. This is a condition impossible to realize
is
in practice.
When
the coil
is
made up
is
of
Maxwell pointed
(i)
distinct
reasons
The
self-inductance
is
of
each turn
of
round wire
wire
when
9 % ii9 V9 w g^ S 9 V 9 mp w %m 9
IP
rig. 3
it is still
mean
distance
of a square
from
itself
is
greater than that of the inscribed circle, and of course (2) greater than that of a smaller circle.
For
the
square,
log
i?5
= log /)+3
2
log 2H
3
^
^2*
is
For
the
R^,
log d log
where
square and
is
The
2
difference
log
+ ^
3
TT
II
6
(4)
= log
-1-0.1380605 d
Hence, the excess of the self-inductance of the coil of round wires due to the greater self-inductance of each of the n turns is
(5)
But in addition to this, the mutual inductances of the round wires on each other are different from the mutual inductances of the square wires on one another. Maxwell states ^ that "the induc*
^
Electricity
Vol. J,
No.
I.
nnder
than that of the corresponding- eight square wires on the square wire in the middle by 2 X 0.01971 " per unit of
consideration
is
less
length, Fig.
may may
''The correction for the wires at greater distance be neglected," sa}s ]\Iaxwell, and hence the total correction
4.
JL
This value
of
47r;/^7
log^^-ho.11835'
(6)
JL
to
Fior.
stein's
formula w-as published in 1884, and in this paper he stated that ^laxw^ell's value of the absolute term in the correction JL He did (0.11835) was not right, but that it ought to be 0.15494.
'^
not give his derivation of this constant, but the context suggests
that
it
is
obtained
b)-
for self-
ofeomet-
mean
distance.
mutual inductances of round wares and of square wires is recomputed by Mr. Chree, using the values of the geometrical mean distances given by IMaxwell in his paper "On
to the difference in the
the Geometrical
Mean
Distances of
Two
Figures in a Plane."'
He
is
The
and Stefan undoubted h- so understood it. The difference between Maxwell and Stefan was as to the correction for the mutual inductances ^J/, where ^irauE (7) According to IMaxwell, i^"= .01971 According to Stefan, +.01688
right,
JM
That
is,
although of the same order of magnitude numerically. As Stefan did not show how he obtained his \alue of the correction, and Maxwell
'Wied. Annalen, 22,
"Transactions of
llie
p. 116;
IVSS4.
l'",(liiil)uri^h,
Roval
Socict\" of
2(>, p.
729;
1S72.
i?osa.]
Geometrical
Mean
Distances
and Self-hiditctance.
show how he obtained the values of th-e geometrical mean distances from which his value is computed, the discrepancy has not been explained. Accordingly, some authors give Maxwell's value and some give Stefan's. I shall, in what follows, derive the correction first by Maxwell's
did not
it
by means
of formulae of self-inductance.
I shall
show
that
Max7) is
well's values of
the
geometrical
mean
distances of
neighboring
E (equation
wrong; that Stefan's value is right as to sign, but only a first approximation as to magnitude; and that the value of this correction term given
by Stefan
but
is
a function of the
number
Maxwell
that
is
the
geometrical
mean
distance
of
two
.99401 times the distance of their centers of gravity, and that the g. m. d. of two squares corner to corner is
1. 001 1
He
does not,
Gray^ gives an excellent treatment of the subject of geometrical mean distances, but there are a number of misprints in equations 104, 109, III, and 113 which it is necessary to correct before using
them
in numerical computations.
(8)
'^
Equation
in place of b'
^Absolute Measurements, Vol. II, Part I, pp. 288-306. ^The sign of the first term of equation 104 should be + The signs before p'^ in the coefficients of the log in the first four terms of equation 109 should all be minus; [(-)'-/>']. Similarly in thus X(/3'-/>'),-X('^'-/'),-X[(-/3)'-/'"^], equation iii the coefficients of the first two terms should be Yz {^"^p'^) and %,
+X
In equation 113 the coefficient of j3* in each of the first four terms should be I instead of }4 and the first term should have log [{p-\-l>-\-l>^)'^ -\- /3"^] instead of
{a^p'^).
logiip+d-^yy-p'].
Biilletin
[Voi. j, jvo.
i.
p distance
as
shown
in Fig.
is
the geometrical
mean
distance between
the rectangles.
J(/+/p+0^j^^-^^-"t|f)!|_giog^(^+^+^)^+^A
(the
same
series of
terms with
y5
replaced by a)
(the
same
series of
terms with
yS
replaced by a)
(8)
Equation
and
d^
(S^-r)
g.
m.
d. for
The
to zero. Fig.
Rosa.']
T
V _.i
^>,
1
1__^
cc
>
I*
a
Fig. 5 a
-H
(the
same
series of
terms with
/3
replaced by a)
+ ^(^+.)^tan-Y|-^ + f (^+.)tan- ^^
^b^ tan -,P ^b^ _,b V-^^tan ^
'
P
terms with
/3
(the
same
series of trigonometrical
replaced by a)
_^/^2_aA|/^+A'-^^-^4-i^^^^^^
(8^)
8
Bulletin of tJie
Bureau of Standards.
[roLs,A'o.r.
of side a, a
d cd^ a = o, ^=a,
and
<7*
log
B.^
= - log5<:r
24
log<?^-|
log4^
24
8^*
,
2a\
tan
3
_i ^
23
,
2a\
'2 tan _i
A.a\ ^^^tan
3
^i
2s ^ 12
a'
. ~*
.-.
log
7v
=log
a-{-
- log 5 24
12
(9)
^4
-ylog 5 = 0.4694194
log
2
= 0.6931472
270898618
tan~^ 1^=0.9272952
Fig. 6
0.0065285 =
a2
= log "^
(10)
.-.
logi?.,
<7)
R^= 1.0065498
<7,
by Maxwell)
Therefore, the g. vi.d. of tivo adjacent squares the distance between their centers of gravity.
It
is
1.006^4^8 times
to
(9) is
required
b\-
number
of decimal places.
is
The degree
ordinarily necessar)-,
above is greater than but not greater than is desirable for the
of accuracy
present purpose.
4.
two squares,
as
and
3,
metric
mean
distance
(8):
is
given
from equation
Rosa.]
log ^3 = log a
log 5
+ --- log 3
log
+ 8 tan~^
i_25
4 tan'
2
12
(^0
Substituting numerical values as before, logi?3 = log ^
+ .6936576
I
= log
.'.
(2.00I02I2<7)
(12)
a
Fif.
a
7
Therefore, the g. m.
tJiird
squares
is
1.000^106
thnes the distance betzveen their centers ofgravity. Since the g. m. d. between the circles inscribed in these two
squares
is
0.6936576 .6931472 = .0005104 = ttg. Thus ttg is the naperian logarithm of 1.0005106 = ^3 -f- 2 We may obtain a check on the values of a^ and a^ hy the following
<i'.
method.
(Fig. 7)
We
first
mean
distance of square
of squares 2
and
3,
by means
of equation
The
log R^ log
+4 + ^^
tan
tan
^-3
w 22
^^^3-/2
log
4 10- ^5
48
log 5
.'.2
log
7?.,
The
o-eometric
mean
we
have,
when
(7=i,
Thus,
log
Above we found
= log i+log 2 = .69314718 ^,. 2 log i^,.= a2+a3 = 0070389 a2 = 0065285 a3 = .0005I04 Sum =a2+a3 = 0070389
.
This exact agreement between the results of two independent calculations assures the correctness of a.^ and a^
lo
Bulleti7t
ivol^.no.i.
and fourth squares we will obtain b}' the Inequation (8^) substitute b a^ second of the above methods. cia, da, and we find the g. m. d. of the square ABEF from the
The
g-.
111.
d. of
the
first
rectangle
BCDE.
3..
In this
way we
find
3 lo^
=3
log
^+'4"
3
+^
2 ^
log
5-
161
24"
log 17
(13)
D
Va; %
^B
.-b-Fi:r.
c
-^
S
By
mean
from the square is one-third the sum of the logarithms of the g. m. d.'s of each of the squares 2, 3, and 4, which make up the rectangle, from the first square. Thereof the g.
m.
d. of
the rectangle
BCDE
fore, since
and log
log
= 3 log ^,>. 4 (log A^^ + log R^ \og a-^ 1. 0987 1 48 = log (3.000307^),
i^4
.s.
R^^ we have
.-.
A^^=
1.
000102 X
<T'
The
fore
g.
m.
d.
in
square 4 from
the
circle
inscribed in square
log
A^^^
log
log A4
log
find
23
Proceeding
in the
i
,
same way
for the g.
111.
d. of
squares 5 and
6,
with
respect to square
we
og
A".,
f^
A'c
c)g /\'
Rosa.]
Geo7netrical
Mean
Distances
and
Self-Indicctance,
II
In the calculation of the corrections due to the differences in the mutual inductances for round wires from their values for square
wires the differences in logarithms of the g. m. d/s are used, and not the g. m. d.'s themselves. Hence the differences a,, a. etc., are the
etc.
h\\
the
6
*5
cc^
same row we have found to be positive, which shows that the mutual inductances of straight round wires
at a
S
2
'^e ^^5
a^ (..) a^
3 a^
than for square wires at the same distances. The same would be true
for the squares in the
?
^3
same column
^5
0^6
with square number i. Thus, we have found all the corrections to be Pis:. 9 applied to the middle wire for the 20 wires in the same row and column for a group of 121 making up a These corrections decrease rapidly as we go away square 11 by 11. from the wire in question, the first correction term being 10 times the sum of the next four.
5.
If
we
find the g.
of square
second row,
g.
we can
m.
d. 's for
the corrections
/3i,
^^^
^63,
M H
etc., as
we have found
shown
a.^,
Taking two
First,
squares, as
in Fig. 10,
we
tance
PQ
P and
Q.
we
inte-
12
\_v0i.3.
no.
i.
JK, and find the g. m. d. from Q to the line JK. Second, integrate with respect to O along Third, integrate L^I, and so find the g. m. d. of L^I from JK. with respect to JK along AB and find the g. m. d. of L]M from the Fourth, integrate along EF and so find the g. m. d. square ABCD.
grate with respect to
line
of
P along the
The formula
follows:
for the g.
m.
d. of
two squares
as
shown
in Fig. 10,
is
as
|_
4<^^
24^*
24J
^^
^
'
'
2a
12a
12J
'^
'
-(^-'f
log
(p+2.)+^^+'f
2^ 2a
loo.
(/+.)-
A
J
log/
3[_ ;[_
^ a
^7' a""
J
,,-,
-^la 3|_ ^
'
^7' ^7
2a
O ^ ,_ (/+
3|_
<'?
-'Vl
/+!-_ +
<'7
2^7
3|_<'7
^7
<^
_ir(/>+2.7)'^_j/>+^n'-'
An_ii
^7~J
SL
a'
"
^7'
(17)
Rosa.]
Geometrical
Mean
Distances
and
Self-Indttctance.
13
is
A^
1
' \
/ /
^'
^
p/.
---,
Fig. 11
log i?/3r-=log
5+7r-4
tan-^
^-^
(18)
is
aj2
There-
i?'^^
= log
a-^~ log
= log ^ + .3465736
log
is,
i^'/3i
= . 002301 9 =
(instead of
in. d.
/3,
= 0.997701 X
That
<i'-y/2
i.ooiix^^Y^ as given by
corner
to corner^ is
jMaxwell).
the g.
of
tzvo squares.^
^=
<^
in equation (17)
we
find the
m.
d. for
/Sg-
the next square in the second row, and also the correc-
tion,
Thus,
2-^
.
log 3
I
.
+ 2^
S
log 5
,
+
25
12
log 13
(19)
8 tan"'
2
<^
.
-+ s
.
_i
tan"'
l""
tan-^ 3 2^^
h
log
8047190
log ^^,
log
R' ^..=
.0000895 =
/Sg
14
Bulletin of the
Bureau of Standards.
\voi.3,no.i.
In a similar manner
log R^.^ log
we
find,
putting/=2^
27
4
log-
in equation (17)
a
,
~ log
\X-\
^ ^
^ ^
24
loor c
^ ^
IIQ ^
24
161
^
^
'
log;
24
_i
log 17
14 tan ^-
10
= log
tan
_i
20 tan
+ 16 2
tan
25 (20) 12
^7+1.1513161
+^
log io = log
^+1.1512925
log
It is to
/Sg
is
positive,
whereas
/S^
and
CHECK ON VALUES OF
/3^,
^8.,,
/S3.
We
/3i, ySg,
/Sg
by formula
Q
by means
of
formula
(8).
The logarithm
H
/Si
/52
sum
m.
of the
D
az
CCi
Ct4
logarithms of the
g.
d.'s of
making up
a^
EFGH.
Since
a^
0.3,
are the
B
/?!
the inscribed
circles,
we may
F
Vvx.
1-2
{Sa)
\'alues
compute the sum of these differences by means of equation found. This will give a check on
Substituting in equation
(8<7)
a=:i, /Q=2,
Fig. 12
bi^
we
R for the
.
adjacent squares
ABCD
and
log
EFGH,
2
9 logy? = 9 log ^
+7
I
+ 24
;''
119
lo g 5
24
I
log 13
161 + -TV
,
24
2
log 17
tan~'
10 tan~^ 4
3
7^
\-7r
(21)
/?osa.]
Geometrical
Mean
15
we have
log 5
(22)
/32
+ log 3 + 2
=9
..9
log ^ log
+ 6.3969297
i?'
log
i?-9
(23)
This is the algebraic sum of the as and /3s for the nine squares as given by means of equation (8). We have alread}' checked the values of the as if the as and /3s sum up to the above value we shall have also a check on the ySs.
;
Thus
a.^
a^z=.
.0065285
a^ 2^^
.0005104 .0001024
.0000471
+
2/^1=
.0071884
.0046038
(24)
This complete agreement with the value above (23) checks the correctness of the calculations of the y3s made from formula (17). Instead of finding /S^ by equation (17) we may use equation (8<2),
<
-Ua
>f
1
G
^1
^3
^4
C
a
D
cc^
^3
JI
B
^1
^2
^3
^4
M
Ficf.
13
Find log R for square A with respect to square EFGH, and log R' for circle A with respect to the 1 2 circles inscribed in the 1 The difference gives the sum of the as and y3s. squares of BFGH.
thus:
22261
07
i6
Since
all
Bulletin of the
are separately
B tireate
except
of Standards.
/5^,
[ ] 'ol.
s, -^ o. /.
known
28
known.
we have
R\2
161
Xo'ga^
log
-,
log 13
'-
lop-
24
17
^
24
102:29+20 tan
^
^
40 tan
4
+ 70
12
For
circle
tan-^--34 tan
'
-^
IT-
25
(25)
log (^+10.6189076
1
and the
2 circles of
EFGH
we have
12 logi^'=:I2 log
log 5
+ logi7
J^ 2{^,
= 12
.M2logi?-I2 log
log
ie'
=
.*.
By
(23),
.0024056 = a2+a3 + a,
.0000646 = 05+2 P^ .0000325 = 05
.0000321
2
/3,
(3^-\- 13^)
But
.0000160
7.
TWO LAYERS
mean
OF WIRE.
ABCD
D
made up
of 25 smaller squares,
the geometric
distances of the
y.
^
0^
'3
^2
^x
middle square A from the rectangle CDEF and from the rectangle FGHI
are given
by formula
(8<7).
If
R^^
is
:^2
'3^
,3^
f"^ k y
G
"2
0^3
the
geometric
mean
distance of
A
the
CDEF
(made up
A^^
is
m.
d.
of
FGHI (consisting of
li
small squares),
= 6.68642698
=0.70018606
^
FiL^ li
log
7?2
Adding
g.
of the
from the whole space outside A inclosed within the larger square, that is, of A from the 24 small squares outside A. Calling this latter g. m. d. I\^^ we have
m.
d. of
/?osa.]
Geometrical
g.
Mean
Distances
and
Self-hiductance.
17
The
in
m.
d.
from a
circle inscribed
any
centers.
The
log of the g. m.
of the circle in
24 logie2, = 4 log
+ 12
log 2
+4
log 5
(27)
Thus
^^, = 00073785.
.
The
total
surrounding wire
is
^J:/2=.OI7708
The
first
layer
surrounding
(a2-|-y5i)
Thus
z/il/i
Thus
8.
THE CORRECTIONS,
We have
all
The sum
of the
is
one-fourth of
^M^
2)
This
is
equal to
2+3+A4-2/32+7l
2
+ C^3=
.0070389
0024810
0045579 0044271
7i=
0001308
g.
m.
d. of
square
get-
ABCD, and
i8
Vol. J,
No.
I.
ting the excess over the log for the corresponding inscribed circles
we
find the
sum
All of them
calculated
D'
having
except
been
7.,,
previously
Thus,
7,= -.0000347
/5i
P2
/33
AB'CD'
g.
to get
d. of
73,
have
h
72
computed the
P*
m.
the square
A
The
B'
with
respect
to
the
of
5
rectangle
squares.
is
BB^DD', consisting
/3,
B
Fij?.
as
15
follows
5 log
^'=59
loR
+ 27
log
3-
^5- log
5-^^
5
log 13
+ --- log 17 + ^D 24 ^ s6
5 log
i'?'
log 29
+ 80 tan-^"^ + 64tan-^;^ ^
tan^'
3
tau-'"^
70
^
tan"-'^
10
^-^
12
(28)
^ ^
Difference
= 7.2152930 = 7.2152400, for the corresponding = 0.0000530 = ar,-\- 2/5^ + 27^ a. = .0000325
2/3^-=
.a. 4-
inscribed circles.
.0000321
2/5^=
.0000646 .0000530
Sum
of 5:=
27.5=
7.,=
.0000116 .0000058
method of differthe logarithms and inverse tanby
this
To
The above
expression for
between large positive and negative quantities (225.24224423 218.026951 18), and then
5 k>g A* gives that quantit)- as the difference
is
This small
J^osa.]
Geometrical
is
Mean
sum
Distances
and
Self-Indiictance.
19
difference
then the
known.
That these
final
The value
calculated for
of
h^
is
.0000256.
F.
W. Grover
f^,
has kindly
verified the
me
the values of
and
Figure (16) gives the 30 corrections for one quadiant; the other quadrants would of course have the
calculation of all the others.
same corrections
for. squares in
corresponding positions.
left
It will
be
below the square A are positive, whereas the corrections for squares at 45 with the principal axes as /S^, 7^, 8^, e^, f^, are negative. On either side of the 45 lines the corrections are also negative, as ySg, 72, 73, etc. Nearer the axes the corrections become positive, as ySg, /8^, -Sg, 7^. The correction h^ is four times 73, although the squares are almost equally distant from A, because 73 lies at a lower
angle,
somewhat
^1
^2
off
the diagonal.
<^3
y.
^6
+8.4
+0.7
-4.0
53
-4.5
e2
-3.1
^1
^5
V4
^2
^
^.
6,
ys
A
A
IK
^5
+ 16.0
^4
-5.8
V.3
-11.3
60
-8.1
-4.5
2
^3
y^
+23.6
-34.8 V2
-25.6
5l
-11.3
52
-4.0
^3
rs
X2
ri
^3
-89.5
180
-130.8
yi
-34.8
72
-5.8
V3
+0.7 74
A.
'6
ft.
A
3
A
2
-2301.9
-89.5
/32
+23.6
+16.0
+8.4
/35
^.7
^3
a.
A
^2
+6528..
+510.4
+102.5 4
+ 32.5
+ 13.6
16
A
y.
^3
IK
A A
Yi
ri
fK
A
A,
A A
Ki
V'.
A
r4
^3
/s
^2
^3
K
S.
y-z
A
ft.
'4
^2
^3
'^^
^2
^1
rs
a.
A
A5
^1
c.
^2
*3
y.
ft.
^6
Fie. 16.
^4
^2
^1
5 8
20
9.
Bidletiu of the
Bureau of Standards.
3
[Voi.3,no.i.
LAYERS AND
LAYERS OF WIRES.
first
We
layer of 8 wires
The
may now
easily be calculated.
(29)
XIO-'
z/iT/3 is
the correction for 3 layers of 48 wires. Fig. 16 gives the correction terms for each of 120 squares on the
11 unit squares.
The
170
1589
I6I4
1619
1621
1622
1622
184
184
185
187
:9.
217
636
1589
1777
1793
1800
1804
1804.
1.4
1-5
13
29
217
I6I4
1793
1796
1800
1803
1804.
1.2
1
I.
10
13
192
I6I9
1800
1800
1801
1803
1804. 7
1
1-3
187
I62I
1804
1803
1803
1804
1805
185
1622
1804.5 1804.
1804. 7
1805
1806
1-3
1.2
1-5
184
1622
1805
1806
1-3
1.2
1-5
184
I62I
1804
1803
1803
1804
1805
185
I6I9
1800
1800
1801
1803
1804.7
1-3
187
I6I4
1793
1796
1800
1803
1804.
1.2
1.2
10
i-^
192
1589
1777
1793
1800
1804
1804.
1.4
1-5
13
29
217
170
1589
1614
1619
1621
1622
1622
184
184
185
187
193
217
636
Fig. 17.
To
find
20 wires constituting
d. of
five layers
we
m.
the square
outside of
it
Rosa.^
Geometrical
Mean
21
of 1 20 small squares.
Then
m.
d. of
in the square
120 squares.
The
result is as follows:
log ^120 =i-3567495i> for the square log 7?\2o = 1.35659900, for the circles
Difference =
SUMMARY OF CORRECTIONS.
For I st layer of For 2d layer of For 3d layer of For 4th & 5th layers
8 = 016906
.
layer.
of
layers.
layers. layers.
mutual induction for a single turn of wire in a coil surrounded b}' other wires, assuming the wires are of small section and the curvature small. Such wires as lie on or near the surface of the coil will evidently have smaller
of the correction for
values for
JM.
COIL.
To
made
add together the corrections due to the neighboring wires. Thus, let NOP be a portion of the boundarv' of the section of a coil, one turn of wire occupying each square. Let us
of rectangular section
The
NjOiP
is
one-
The
give corrections
0.01170 which
corner ware.
is
The
0.018061-^4 =
^1
c,
^2
C^
For A,
A3
A,
etc.,
^,
B2
^2
Total correction
mm
o,
^8
^4
Mot. 18
22
IV01.3.N0.1.
In Fig.
y>
^,
the values of
left for
the differ-
ent wires, and the differences between the respective valnes and
.AT
.01806, the
right, the
on the
numbers on the ^
D^
C2
^.
c,
The
la}-er
0,
differences
are practic,
^,
0,
^,
^2
corner.
^3 ^4
The average
turns
IS
value
^,
A,
^^"'
^3
^o
^4
10x10
.00093, i^a-k-
JM=^irnaX .01713.
(30)
10.
is
given by
= log a^-
^3
mean
log
2+- ^
3
(31)
12
i?=. 447049 X a
distance and a
is
where
square.
is
the geometric
itself is
^^|'
+i
(32)
,b
2 b
-
tan
_M
<7
^ \ 2S 12
<7
where a and
tively.
/^
When b= 2a,
R=zo.6yo8o7,xa
^0
Maxwell,
II,
092.
J^osa.]
Geometrical
Mean
Distances
and
Self-Inductance.
23
Suppose ABCD, Fig. 21, represents a rectangle in which the length is two and the breadth one, and EF divides it into two unit squares. Let R be the g. m. d. of the entire rectangle from itself, R^ the g. m. d. of the square from itself, and R^ the g. m. d. of one square from the other. Then i?= .670803, R^ .447049, and R^ is to be determined.
By
mean
distance
= log i?i + log or i?' = i?i R^ 2 logio ie = 1.65 3 1 900 logio ^1 = 1-6503551 .-.logio i?2 = 0.0028349
2 log i?
R^
{zi)
This
is
mean
(8<^),
already found by
means
of
formula
by
less
than one
be a square of length 2 units, divided into four unit be the g. m. d. of the large square from itself squares. Let = .894098; R^ be the g. m. d. of each small square from itself
Let
ABCD
any small square from an adjacent one= 1.006549 (see above) and R^ the g. m. d. of two squares corner i?3 is to be determined. to corner, as i from 4 or 2 from 3. By the
i?2
d. of
= .447049;
the g. m.
mean
distance,
{l^)
^1=1-6503551
2 logio i?2
log 2
0.1 505 1 50
R
Fig. 22
.*.
1.9990005
^^V2
i?3
= 0.997701 XV^
24
Bjilletin
is,
[ \ 'ol.
J,
No.
I.
the g. m. d. of two squares, corner to corner, is 0.997701 times the distance between their centers. This is the result previously found from equation (17).
12.
That
"SLa^
Let the rectangle of Fig. 23 consist of three unit squares. Let %. m. d. of the rec-
tangle from
self
Fis?.
= 0.447049, i?2 = g. m.
23
d. of
g.
m.
d. of
square
on square
3,
which
is
be determined.
By
mean
distance
^=3
i?3
log ^1
+4
log
^2+2
log
i?3
(36)
and
R' Ri'R.
= 2.001022 = 1. 0005 1 1 X
2a
R.
1.
R.'
p. 9.
= .0005II
have thus derived from the comparatively simple formulae (31 and 32) the same vahies of a.^ and a.^ that were obtained from equations (8) and (8).
13.
We
Circle of square
The
section.
As
first
mean
distance
a single circular
Rosa.]
Geometrical
Mean
Distances
and Self-Inductance.
25
ratio of the radius to the side of the cross section is greater; hence,
we may
take a
2^ cm
The
and
/^
= <r=o.i cm
as a case favorable to
high accuracy.
stein's, Stefan's,
Wein-
mean
The
Weinstein's
b
0.1
em
Z = 47r
Stefan's
\,
Sa /
log
y.
d'
-^--^
^ l+.o3657p-i.i949i3|
1
d'
^'
{37)
4jiog
where
ji
f.
^- ['^j^j-y^^Y6^^y-^\
,
,
I?'
d'
(38)
^1
il
Hi
= 0.848340
and
^.^
= 0.8162
for a square.
Maxwell's
L = 4^{log
J.
(i
+ g)_ 2 _ ^,[
(39)
where 7? is the geometric mean distance of the square secYis. 21 tion from itself =.0447049 cm in this case. = 0.1 cm we find the following Substituting a = 2^ cm and
(^
By
Weinstein's
= 2000.
1
log^
l?'
2000 =
8a
7.
600902
03657
p=
.000001
7.600908
-I.I949I3
6.405995
4<2=ioo
.'.
= 640.5995
TT
cm.
26
\voi.3,no.i.
By
Stefan's:
log^
2000=
log
7.600902
-Vi
b^
.,
2= -0.346573
7-254329
=
8a
log:
.000005
.000001
24^'
b^2
b^
16^^-ri
.
7-254335
==
0.848340
6.405995 640.5995
L
IT
Bv
Sa
:Maxwell's:
200
~~
2000
.447049
logg
2000 =
.805 087
.0447049
log
.447049^+
1
8.405989
^ V
no-
0. 00000 S
^(-i->
TT
8.405994 2.000000
6.405994
= 640.5994
These three values of L are practically identical. Stefan's formula is derived from Weinstein's, but Maxwell's is an independent one. This confirms the value of A^ for the square.
Problem
2.
As
calculated on
which is equivalent to a circle of infinite radius. We see by the above example that it must be substantially the same for a circle where the radius is 250 times the side of the square as it is for a circle of infinite radius. For a square of
the basis of a straight conductor,
the radius of the circle being 25 cm, and hence the radius only 25 times the side of the square, the value of would not be as exact, and hence the agreement among the three formulae would
i.o
cm side,
Rosa.}
Geo7netrical
Mean
Distances
and
Self-hiductance.
27
for this
we
By By
g.
The
m.
is
d. is
little
too great
when
applied to a circle
considerable.
ProbleiJi J.
cm
round wire by the method of summation, and also by Stefan's formula, we can obtain the correction for mutual
distance.
is
mean
turns
The
Z=2A+2Afi2
Suppose
<^
a ^25 cm
= 99.85 cm
of the section
Z^ = 791.694990
25
The
self-inductance of a coil of
of sq7ia7^e section
by
1x2
mm
and the
radius the
same
as before is
Z,
47r
2949-55944
is
The
na (.1380605), ;/^=
.-.
199.7,
^^ ^
4'7r
27.57068
Difference
28
If
Bulleti7i
Vol. J,
No.
I.
should agree.
V O.J
Their difference
is
That
is
0. -'-->}
^c>L=naa^=
?/= 199.7
.-.
Jk a
a^
=.0065278
a^
199.7
061
^1
by the method of geometrical mean distances. If we had used .0065285 and applied the correction 7taa^ based upon it we should have found J^L\ir =1.30374 and L^d^ir This differs from the value Zo-/47rby the 2978.43386. method of summation given above by less than one in
This
is
20,000,000.
This
is
mean
and shows that when the radius large in comparison with the dimensions
of the coil is
of the cross
Problem
^.
We
will
now
find
yS^,
(40)
L_2-?7)}-J
and Zg are calculated b}- Wien's formula and the Ms by Maxwell's formula for mutual
inductances.
The
2A= 2A=
2il/,2=
2^1/34=
1583.3900 1585-1758
1395.0438
1
Oil
396.6410
4-^13=^
2791.6843
2653.1941
1
4J/,,=
.-.
Lfj\/\.ir
1405. 1 290
Kiy. 27
i?osa]
Geometrical
Mean
Distances
and
Self- Inductance.
29
By Stefan's and
coil.
Weinstein's formula
Lyj^^ir is
7^=11345-6632
Correction to reduce from square to round
wire
.*.
^___o^
47r
47r
= 4.2Q68 = the ^
naiza^^^^
,
ry
.-.
2a,4-ft
^ ^ = 4.2968 =
399.6
22
.-.
.010753
-013057
^1
=.002304
distances
By
mean
we found
/S^
to
be
This slight difference amounts to only one part in 15,000,000 of the whole value of the self-inductance of the coil. Thus we see that the two largest correction terms a^ and ^^ derived by means of formulae of self and mutual inductance agree with the values found by the method of geometric mean distances.
It is
.002302.
this
way.
As we have
4a2+4;5i
Ofa
^,
a-2
and
E (page
4),
which
/3i
^2
,3,
results
from the correction given by Stefan. This Fig. 28 neglects all the other wires and assumes the effect This is indeed a very good equal on all the wires of the section. first approximation to the correction, and amply accurate for most
purposes.
30
[Voi,3,no.i.
Problem
5.
The
the
by
method
O.ltcm
summation
is
l<
I
Z = 4A + 6il/i,+4J^,3+2J/,,
0. 1
(41)
last article
k ^k ^k
I
cm
The Ls and
are as follows:
^14=
^-a-
47io.8697r
=44239-i747r
By
coil of
section
Tig. 29
for the
above dimensions:
Z^/
4400 1
The
7 2 97r
We
must add
latter involves
it
sum
of these
Correction
^+^-=.148415
.*.
237.4647r
.-.
= 44001.72977 /, = 44239.i937r
.01977
I^osa.]
Self-Inditctance.
3T
This difference
the correction
less
E for
Problem
Ivct
6.
Coil of Ten
Tiltjis.
being 25 cm, diameter of bare wire being 0.8 mm, of covered wire i.o mm, the whole length of coil being i cm. The self-inductance of this coil has been calculated in a previous paper" to be 47385. 827r
cm
h
1
cm-
cm
^^
Eio:.
30
(substituting a
= 2^^
cm.
7i=zio l?i.o,c=:0.i)
Z,,
= 47oii.937r
This assumes a uniform distribution of current over the rectangular section, as though the coil were wound with square wire having insulation of infinitesimal thickness. The correction to apply to reduce it to the actual winding of round insulated wire is as follows:
JL /^iran
Standards,
2, p. 165;
logg
+ -1380605+-
2a
22261
07
32
Bulletin of tJie
Bureau of Standards.
n turns
is
voi.3,no.i.
2a
as follows:
2a=2
For a
+ (;/-3)a, + (;/-4)a5+(;2-5)ae+...J(42)
we have
.006528
.000102
9 = 058756
.
^=- 2a =
n
(7
012764
1
38060
^= log. ^ = -223144
Sum
=.37397
= 47oii.937r cm
cm cm cm
summation, and the
(See
p. 31.)
==47385-907^
.o87r
That
is,
method
of
amounts
to
one part
Problcfn
7.
Coil of
Twenty Turns.
a 2^ h2 cm ;/ = 20 r=o.i cm Diameter of bare wire 0.6 mm, of covered wire i.o mm.
In the last case
we obtained
first
j?osa.]
Geometrical
Mean
Distances
and
Self-Inditctance.
ZZ
ond by assuming the current uniformly distributed over the section and then applying the three corrections C, F^ E. In this problem we may calculate Z, first, by use of the current sheet formula and apply the corrections A and B to reduce to round wires, and then, second, by Stefan's formula for uniform distribution and apply the and reduce to round wires. three corrections C, F^
6 %
cm-
4-l4-l-f-l-l-h-N-H-^-l-H-H--Fn Q.2
Fii?.
31
is
as follows:
8a 4 L = 47ranmog^-o.5
Here
Sa
= lOO
log,
100 = 4.605170
lo.^^+j)
20,000\
'
.000971 4.606I4I
o. 500000
4.I06I4I
47r^;/2
= 4o,ooo7r
.'.
Z^,= i64,245.647r
cm
the self -inductance of a winding of 20 turns of infinitely thin tape, each turn being i wide, with edges touching without
is
This
mm
making
electrical contact,
which arrangement
fulfills
the conditions
34
Bulletin of the
Bureau of Standards.
ivoLs^no.i.
of a current sheet.
To
we
and
B for self
By Table By Table
VII, for
VIII, for
l = o.6, = 20,
;/
A=
B=
A^B.^
200077
.3434
47r<2;/=
r.
JL--=\iran(^A^B)^
.-.
cm Z = Az/Z= 163,558.8477 cm
686.877
By
;/,
l)^
Stefan's formula
<r,
we
find,
<7,
and taking
j'j==. 548990
Z.,, 162,234.6077
cm
The
correction
E ~ 2a
is
found
b}'
The
^=.13806
i^=. 51082
.66245
.*.
^Z=:477<7;/(C+^+^)=
1324.9077
cm
.-.
This value
1))'
of
only
foiir
parts in a million.
method
of
by Stefan's or Weinstein's formula and applying the corrections C, /^, Ogives practicalh' identical results with the method of summation and also with the current sheet method y?;r^//<?;'/ eoils. When, however, the coils are longer the agreement is not so good,
calculating
Z,,
from
it)
if
when
is
greater.
Thus
the coil in the above problem had been 5 cm long and 2.5 deep, and wound with 20 turns of heavier wire the difference would
mm
have been
part in 25,000
(still
if it
were
''''Rosa,
Rosa.i
Geometrical
long and 0.5
Mean
Distances
and
Self-Inductance.
35
locm
cm
deep (the radins being 25 cm) it would have For most experimental work, therefore, the
Problem
8.
Coil of Sixteen
b
Ttcrns.
<^=ioo.
c^^\rvvci.
i
The
turns
coil
is
mm
wire.
It is a
somewhat
test
by the method
work
is
dimin-
o..kcm-
The
result is
Zo-= 656,954.2877 cm
By Weinstein's
b-
0.4,
16,
A. = 655,973.8777 cm
To
The
bered
numeach
Fior.
to
32
^lius
(2x4+ 3x84-4X4)--i6 = 3
That
is, ag
In the same
average and
way we
a^
on the
once.
Considering
all
way we
find
2(a, ^,7,8)
= ^=3a,+ 2a3 + a, +
%+
-:}
?^,
+ 3^, + 7, + 7, + ^ 4
(43)
36
[ l-^ol.
J,
No,
I.
The sum
u
^3
10
= ;/!
33=
2a3
15
.019586 .001021
.000102
13
<^4
16
72
11
a,=
i2
9 3
5 2
P2
6
7
72
8
^^3
i*-
.000035
.020744
^2
2 "2
3
4
'^4
Mg. 33
'1
i5'=+.oi5i24
the diameter of
The
correction
F
i
is
zero, since
we have assumed
d.
^=.13806
^^'=.01512
Sum = .i53i8
/^irna 64007r
JL /\'Tr7ta{C-\-E)
From
above.
.-.
980.3577
A/ = 655,973.87 77
cm cm
This vahie
of of
of
L agrees
summation within one part in 10,000,000. In the calculations La and L the values were carried out two more decimal places
than are g-iven above, so that the close agreement is not merely accidental. With the large radius and small section chosen the formulce
and this gives an excellent test of the corrections, a, yS, 7, 3, for mutual induction. The relative importance of the corrections (7, F^ and E decreases with the number of turns, for JL is proportional to // whereas L is
are very exact,
number of turns it is quite unnecessary to know E accurately, 0 far as knowing the selfinductance for experimental purposes. If we take (7=. 138, E= .oi<S, or ^^'-^A'^o. 156, we shall be amply accurate, when ;/ is large,
proportional to
n^\
Thus
for the
For a
coil of
few turns
Rosa.]
Geometrical
Mean
Distances
and Self-Inductance.
2>7
is
from the above list that will be nearly right, or it can be calculated with great precision from the values of the corrections, a, yS, 7, 8, etc., given in Fig. 16. These corrections of course are carried out much further than is necessary- for any requirements of experimental purposes. Measurements can not be made of the section of the coil and of the diameter of the wire accuratel)' enough to justify such refinements. It is better, however, that the formulae should be more accurate than necessary rather than less so, and for the purpose of testing other formulse it is desirable to be able to calculate self-inductance by this method with the highest precision. If further justification for these corrections is needed it must be found in
the interest for
distance.
its
own
mean
Summary of the values ofE fou7td for the various cases considered:
2 turns
E=^
(one layer)
.006528
.009045 .01691
01035
(Problem
^=
3)
4
8
(two layers)
(one layer)
(
E^ E=
E^=^
4)
5)
10
6)
7)
E^ (four layers) ^= 100 (ten layers) E E= (20x20) 400 1000 E (50x20) Infinite turns E^
20 16
"
"
8)
14.
We
7, 3, e
which the
ratio of the
is is
is
very great.
When
the section
reduced; that
is,
mean
distances
is
can find the magnitude of this difference by computing the mutual inductance of two circular coils of square section by formulse
We
38
BtiUetiii
I
\_Vol.3,No.i.
have given elsewhere/^ and finding the distance apart of This distance is greater than the mean disthe equivalent circles. tance between centers of the coils when they are near and less when they are further apart, and it is always less than the geometrical
which
mean
2
cm
cm
ductors
are equal
(/
and the section considerable. The two straight conductors whose square sections are shown in Fig. 34 may be replaced by
two single conductors Oj O2 near their centers, the slight displacement from the center being such as to be equivalent to the correction The distance apart of Oj Og would of course be the geofor section.
1
140
120 100 =0.1
canoi
80
60
\
V\ \ v\
\, \
40 20
V v;^ "^ \
20
--. _A
a-1(:oo
m
cm
L:_5o
^^
-.^
.
-r
-"~
40
60 80
100 =
0.1
a^
25^ Tl
mm
.,
,.
(}
CE NTl^4ETE RS
Fig: 35.
metrical
mean
distance of
and
P>.
\\'e
and
P>,
Fig. 34.
This is a little greater than have seen for parallel squares The curve A (Fig. 35) shows
Rosa.}
Geometrical
Mean
Distances
and
Self-Inducta7ice
39
O3 exceeds the distance between centers for distances between 2 cm (when A and B are in contact) and 20 verticm. Each square represents i cm horizontally and .02 cally. Curve A shows how rapidly the displacement Jd falls off at d^^ as the conductors are further apart, Jd being only .001 10 cm.
miicli the distance O^
how
mm
mm
If,
however,
is
AB
corrections
by the method
I
mean
is
distances as
we have
seen
only reliable
when
the section
have calculated the correction for section three circular coils, all of square section 2x2 cm, I have 50, and 1,000 cm; this correction is JM}^
^d which
J-,
[
^M by
i^d^
10 s^^ \
256^*
64.128^7
1 2 8)
Vy/
^^^^
These displacements are plotted in the curves A, B, C, Fig. 35. Curve A represents the coil of 1,000 cm radius, which is 500 times the side of the section, and approximates closely to the case of a straight conductor, having sensibly the same correction for section The (represented by the displacement) as the straight conductors. correction for section is negative (represented by a positive value for
zld) in
all
cases
when
the conductors
are
quickly becomes positive for the circular conductors of radius 25 and 50 cm as the distance increases. In the case of the coils of
smaller radius the negative displacement Zld
the larger distances, although
it is
is
relatively great at
value for
d20
than o.i
is
mm.
the correction
con-
that the
or,
in other w^ords,
for
straight
ductors
can be applied
circular
is
conductors
only
when
the
40
[j^oi.j,a'o.i.
APPENDIX.
Table of Inverse Tangents
to
tan
i2
1
0.083
141
232
tan
3
1
0.321
750
554
tan
11
1
.090
659
887
1
tan
2 2
.463
647
609
tan
10
1
.099
668
652
!
tan
11
.179
853
506
tan
.110
9
1
657
221
1
tan
.380
506
377
tan
1
.124
:54
994
tan
.588
3 3
002
603
tan
7
1
.141
897
055
j
tan"'
4
3
.643
501
109
tan" tan"
.165
6
1
148
681
tan"'
.540
5
419
500
.197
5
1
395
560
tan"
tan
4
.674
5 5
740
942
tan
.244
978
663
.694
738
276
tan
TT
tan -II
,
3
-1 I
tan
"^
= 2 tan
"
TT
etc.
's
Equation
f^icor
'-,6
.Vl
y-^
i
f>:c
or
cii,
"
0.80815
.81823
>'i
0.50000
.54899
.59243
.63102
0.1250
.1269
.1325
0.55
0.3437
.3839 .4274 .4739
0.60
0.65
.82648
.83311
0.70
0.75
0.20
0.25
.66520
.69532
.72172
.83831
.5234 .5760
.6317
.6902
0.80
0.85
.84225 .84509
.84697
.84801
0.30
0.35
0.90
0.95 1.00
0.40
0.45
.7518
.8162
.2728
.84834
0.50
.79600
.3066
j?osa.]
Geometrical
Mean
to
Distances
and
Self-Inductance.
Numbers from
1 to
41
100.
2 3
4
5
000
147 612
000
181
294 437
759 910 441 224 585
51 52 53
54
55
6
7
8 9 10
11 12 13
56 57 58 59 60
61 62 63 64 65
14 15
16 17 18 19
4.110 4.127 4.143 4.158 4.174 4.189 4.204 4.219 4.234 4.248
4.262 4.276 4.290 4.304 4.317
873 134 134 883 387 654 692 507 106 495 679 666 459 065 488
733 805 708 447 026
20
21 22 23
66 67 68 69
70
71 72 73 74 75
24 25 26 27 28 29 30
31 32 33
76 77 78 79 80
81 82 83
4.330 4.343 4.356 4.369 4.382 4.394 4.406 4.418 4.430 4.442
204 903
561 525 061
34 35
36 37 38 39 40
41 42 43
84 85 86 87 88 89 90
91 92 93
938 913 160 646 454 067 618 116 634 490 396 602 Oil 298 005
4.454 4.465 4.477 4.488 4.499 4.510 4.521 4.532 4.543 4.553
4.564 4.574 4.584 4.595 4.605
44 45 46 47 48 49 50
94 95
96 97 98 99 100
820 023
W'"^
i^rtr
\h n'