Sie sind auf Seite 1von 3

Note on Helix Impedance Measurements Using an Electron Beam

D. A. Watkins and A. E. Siegman

Citation: Journal of Applied Physics 25, 133 (1954); doi: 10.1063/1.1721506


View online: http://dx.doi.org/10.1063/1.1721506
View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/25/1?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in


Some Notes on the Measurement of Acoustic Impedance
J. Acoust. Soc. Am. 28, 351 (1956); 10.1121/1.1908326

Helix Impedance Measurements Using an Electron Beam


J. Appl. Phys. 24, 917 (1953); 10.1063/1.1721402

Some Notes on the Measurement of Acoustic Impedance


J. Acoust. Soc. Am. 19, 420 (1947); 10.1121/1.1916499

Notes on Acoustic Impedance Measurement


J. Acoust. Soc. Am. 14, 143 (1942); 10.1121/1.1916210

Notes on Acoustic Impedance Measurement


J. Acoust. Soc. Am. 14, 127 (1942); 10.1121/1.1902318

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
141.52.11.150 On: Tue, 19 Aug 2014 08:55:24
LETTERS TO THE EDITOR 133

Further Comments on Summation of Slowly have that liIlla ... (Dmf.) =0 for m a non-negative integer," he
Converging Series supposes that f. is the general term of the series to be summed.
But for a series which violates the validity conditions of Euler's
IGOR GUMOWSKI
formula, his subsequent relation
U nivtTsit~ Laval. Quebec. Canada
(Received October 24. 1953)
Jo" f(x)dx= Jo" an+zdx= i" a.dx
I ADMIT readily that once the result of my paper is known,
it is possible to show that a formally equivalent, but not a
fundamentally identical, expression can be deduced from Euler's
becomes meaningless, because the integral on the left side diverges.

summation formula. However, this result was not established


rigorously before the publication of my paper, as can easily be Note on Helix Impedance Measurements
seen by checking the published literature.
By consulting K. Knopp [Theory and Application of Infinite Using an Electron Beam
Series (Blackie and Son, Ltd., Glasgow, 1951)J, who is an author- D. A. WATKINS. Stanford Uni.ersity. Stanford. California
ity on series, it can be noted that my result differs essentially from AND
Euler's formula as the latter is rigorously stated, both in form and A. E. SIEGMAN. Hughes Research and Development Laboratories.
in the type of series to which it applies. In fact, Euler's formula Culver City. California
cannot be used for summing series whose general term is of the (Received October 28. 1953)
typefn= link, k>O, because at least one of the following integrals, N connection with our recent paper in this journal,1 two items
Jon f(x)dx or Jon P2k+1 (X)PHI (x)dx, Ifollows:
of correction or addition have come to our attention as

diverges in the Riemann sense. No transformation of Euler's (1) Dr. R. Kompfne~ of the Bell Telephone Laboratories has
formula can change this fact. suggested to us that the mysterious backward wave labeled "?"
On the contrary, the result of my paper is especially adapted in Fig. 5 of the paper could be the result of our signal generator
to series of the type fn= link, k>O, because only the existence of having third-harmonic output. The experiment has been repeated
.r,... f(x)dx is required. with a similar signal generator, and Dr. Kompfner's postulate
However, if a series S not summable by Euler's formula is verified. Thus Fig. 5 would correctly represent our experimental
split up, there results findings if the component labeled "?" were not shown.
(2) Although Fig. 10 is substantially correct, it is somewhat
00 noon 00
misleading. The figure shows two curves-one is the computed
S= ~ ak= ~ ak+ ~ ak= ~ ak+ ~ a p+n ' impedance of the (-1) space-harmonic component at the helix
k-I k-I k~n+l k~1 p~1

radius versus ka tan"" and the other is the measured impedance


It usually happens that the series converted to the mean helix radius versus the same variable. The
large discrepancy between the two is a result of the finite thickness
of the tape used in the experimental helix. If the measured im-
pedance is converted to the inside radius of the helix rather than
having for general termfp=ap+n, is summable by means of Euler's the mean radius, the discrepancy between the computed and
formula. Thus, the sum of the series S can be expressed in terms measured curves is less than 17 percent over the entire range, and
of the sum of the essentially different series S', which amounts to most of the remaining discrepancy can be attributed to effects of
changing the series to which Euler's formula is applied. This space charge on the measured curve at the low-frequency end.
result is, however, a new and original contribution. This experimental result is of some practical importance in
To illustrate the above point on a concrete example, let us connection with the design of helix-type backward-wave oscil-
consider the series lators. Thus, if we are interested in the impedance at some radius
within a helix wound of tape with finite thickness, the theoretical
~ 1 impedance given by Eq. (8) or the dashed line of Fig. 10 of the
S=~ '2
n-ln paper should be considered to exist at the inside radius of the
which obviously is not summable by Euler's formula. Splitting helix rather than at its mean radius. A similar rule applies to
up S at the nth term, there results the region outside. This must of course be regarded as an approxi-
mation for tapes of finite thickness since the additional energy
n 1 ~ 1
S=~-+~--.
stored in the fields between the wires must actually lower all of
2
k-I k k-I (k+n)2 the space-harmonic impedances somewhat. This reduction should
be very small for relatively thin tapes.
Now, the series
1 D. A. Watkins and A. E. Siegman. J. Appl. Phys. 24. 917 (1953).
s,=i __1_
k-l (k+n)2
is summable by Euler's formula; therefore,

S= i; .!..+S'.
2
Comment on "The Summation of Series Involving
k-l k Roots of Transcendental Equations and
Of course, this result has to be formally equivalent to the result Related Applications"*
of my paper, because it represents the sum of the same series,
G. KUERTI
but it is by no means identical in principle. Case Institute of Technology. Cleveland. Ohio
As far as I know, the suggestion leading to the above reasoning (Received October 12. 1953)
was first made by C. E. Sunborn and R. A. Mugele of the Shell
ROFESSOR Spiegel's aim, to present in this paper a "quite
Development Company, Emeryville, California, and it is to them
that should be given the credit for making possible the proof of
the formal equivalence (but not the identity in principle) of
P general method by which the summation of such series may
be obtained," has not been achieved, in my opinion. Four well-
Euler's formula and the result of my paper. known examples are given, and the impression is created that the
Mr. J. H. Braun does not seem to bring out the really significant method is safe in all cases where "the transcendental function
aspects of the problem. In the sentence "For convergent series we has real zeros" only. Actually, the applicability of the procedure,

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
141.52.11.150 On: Tue, 19 Aug 2014 08:55:24
134 LETTERS TO THE EDITOR

which was extensively used by Euler and Jacobi, depends in no Then the "usual algebraic relations" will hold in case (a) between
way on the reality of the zeros, but it is a simple matter to give the coefficients a" and the roots, and in case (b) between the
sufficient conditions which cover the four examples in the paper coefficients a. and the kth powers of the roots; but in this case
and may hold in many other practically important cases. To each group of k roots
arrive at these conditions, we have to quote two theorems. Zn+).=Z" exp (2n)./k) , h=O, 1,2, "',k-l,
J.l If i c" is absolutely convergen t, then the product IT (1 +c"z)
1 1
contributes only once.
is an integral function, and the coefficients a" of its Taylor * M. R. Spiegel, J. ApJlI. Phys. 24, 1103 (1953).
1 See e.g., K. Knopp, Theory and App/icalwn of Infinite Series, translated

expansion about z=O, viz., 1+2:" a"Z", are related to the zeros from the second German Edition (London & Glasgow, 1928), Sec. 254.
• Hadamard's canonical representation. See e.g., E. C. Titchmarsh, The
1 Theory of Functions (London, 1939), Second Edition, Sec. 8.24-8.26.
of TI, -c,,-1, in the same way that holds for finite products, that • This idea is implied in a remark by E. Borel: Le~ns sur les fonelions
entih-es (Paris, 1921), Second Edition, p. 83.
is, a" = 2:cvICV2' •. C''''
II.2 Every integral function fez) of finite order, p, can be repre-
sented in the following form:

f(z)==zkexpQm(z).~
..
(l-u,,) exp(u,,2
u.+ 2 +·· '+p' u"P) (1)
u,.=z/z", Metal Whiskers
S. ELOISE KOONCE AND S. M. ARNOLD
where the Zn are the zeros of fez) j they contribute to (1) according Bell Telephone Laboratories, Murray Hill, New Jersey
to their multiplicity, and zo=O has been put in evidence as a (Received September 30, 1953)
k-fold zero. Q.. (z) is a polynomial of degree m, and p is the smallest
'" will converge everywhere. [p may be zero j
integer for which TI
I
IN the earlier work with filamentary growths on metal sur-
faces,t-a the whiskers quite generally were observed to be
approximately 2 microns in diameter. Since then many more
in this case we re-define (1), omitting the factors exp(u,,+" . ).J
The order, p, introduced above is the smallest number for which specimens have been examined and we have found that a greater
variation in whisker size is possible, not only on different but even
If(z) I <A exprP+', r= 1211 (2) on the same specimens.
In Fig. 1, two tin whiskers may be seen with approximate
is true for any positive E, A being a constant which in general diameters of 5.8 and 0.05 microns, respectively. A satisfactory
depends on E. We finally need the exponent of convergence, PI, being
formulation of the growth mechanism must take into account a
the smallest number for which variation of this order and at the same time explain the fact
that a large part of the filaments are approximately 2 microns in
~ Is" I-(Pl+'>' (3) diameter. Two published articles regarding possible growth
I
mechanisms··· have not been concerned with actual limits as to
is convergent for any positive E. Only the following results of the whisker size. Frank,S proposing that an oxidation process occurring
factorization theory are important for our purpose: at a dislocation is responsible for growth, infers whisker diameters
(4) of the order of 2 microns.
and Further, during the earlier work referred to, the whiskers were
PI=P, if m=O. (5) found to be generally straight. Recently we have observed a few
which show an offset, apparently at some stage developing at an
Even if p==O, it is often quite difficult to determine Qm(Z) for angle to the axis of the whisker for a short distance and then
a given f(z) , and it is the presence of the factor expQm(z), which reverting to the original growth direction. The change in direction
in this case makes (I) in general inapplicable. Consider e.g.,
f(z) = expz2z-1 sinz. The zeros are, of course, not affected by
the factor exps2, but the Taylor expansion starts now with
1+5z2/6+···, and, according to Professor Spiegel's argumenta-
tion, this would imply 2:1/n = -511" /6.
In the case p<1, however, we can definitely state m==O, p=O
as consequences of (4), and therefore Pl=P by (5). This immedi-
ately implies the applicability of Theorem I.
But in practical problems one often meets p= 1. All four ex-
amples given by Professor Spiegel are of this nature. This would
seem to make the search for the factor exp(az+b) indispensable,
were it not for the fact' that the Taylor series in the first three
+ +
examples has the form 1 2:a",z2n and the form 1 2:a,.z<" in the
fourth. Thus fez) is an integral function of t=Z2 with order i in
the first three cases, and of t = z4 with order l in the fourth case,
and we have m= P=O for the canonical representation in the
variable t. The exponent of convergence PI equals i (or i, resRec-
tively). With t,,=Z,,2 (or z""
respectively), the sums 2:1/,,1- ttt)
(or 2:t,,-(t+.), respectively) converge, and the sums 2:lt"l-l do so
a fortiori. Thus Theorem I applies.
This result may be formulated in the following way. Let f(z) = 0
be a transcendental equation and let fez) be regular everywhere
and have the order
(a) (b)
(a) 2250 X
or

p~l
..
and the Taylor expansion l+2:a"zk" with p/k<1. (b)
I", ..

FIG. 1. Electron micrographs of silhouettes of tin whiskers


1 illustrating size range.

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
141.52.11.150 On: Tue, 19 Aug 2014 08:55:24

Das könnte Ihnen auch gefallen