AUDIO engineering society

Containing the Activities and Papers of the Society, and pub-

lished monthly as a part of AUDIO ENGINEERING Magazine
OFFICERS
Theodore Lindenberg, President
Audio Engineering Society, J. D. Colvin. . .Executive Vice-Pres. John G. Frayne. .Western V~ce-Pres.
Box F, Oceanside, N. Y. Norman C. Plckerlng. . . . .Secretary Ralph A. Schiegel . . . . . . .Treasurer

Longitudinal Noise in
Audio Circuits- Part 2
H. W. AU6U3 I A U I * and W. t. K A N N t N B t K b "

A discussion of the general effect of the presence of longitudinal noise on a transmission circuit, with
a description of the differences bdween metallic circuit noise and longitudinal noise. Test circuits and
representative conditions are illustrated and discussed.

fixperience snows ~ntrx, In general, ule 1ecel v 111stqulp~ilc~llr

IS, 111 SCALCA UL, CI UIIUIVIIIICI, I C ~ L C D C L A ~ ~ C U111 uuv U L U ~ ----A

neither a longitudinal noise voltage the greatest single

. cause
. of trouble from
-. as lumped
. . parasitic
-. - capacitors
. C1 .and.

nor a current can be impressed on the longitudinal induced noise, especially C2. A longitudinal nolse current z t / % ,
input circuit of an amplifier without when the center point of the input cir- whose magnitude is determined by the>
degrading the signal-to-noise ratio of cuit is not grounded. Difficulties, in coupling impedance Zc, flows along the
the system. I t is, therefore, of interest this case, generally will be experienced conductors of the circuit to ground
to investigate by what means the longi- from longitudinal noise currents. The through the capacitors CI and Cz. The
tudinal induced noise is converted into manner i n which the translation from flow of this current through C2 causes
a metallic-circuit voltage in order that longitudinal to metallic-circuit noise little difficulty. However, the flow of
.I . 1
16 may ue ampllnea ana appear in tne
1 . 7 1 . .1 4.-1--?.- I - - - 4.-2 ?,,.L-A-&:--ll-
Lanes yrzroe 21-s 1epLesellLeu aulleuaLlually
4.L- 1---:4.--.3:--1
~llelullg~~uulllzrl
4. 4.L .
I /-l 2
v~
uullall~~ l l u u g ~
-----
r allu
--
rn~tnitt - - "the ~mnlifier
- - - - - - nf
- - - r - - -circitit I--------
in Fia. 8. It is assumed. for the ~ u r ~ o s e sthe .
A L ,
grid-to-ground -
imvedance of the
of illustration, that the impedance to amplifier sets up a metallic-circuit volt-
Shielding ground of the input equipment and in- age on the grid side of the coil which
The omission of an electrostatic terconnecting circuit is large compared is amplified and degrades the signal-
shield from the input transformer of to the impedance to ground of the re- to-noise ratio of the system.
ceiving equipment. The impedance to Figure 8 and its discussion show in
*Bell Telephone Laboratories ground of the amplifier results from the fairly simple manner how a longitudinal
Murray Hill, N. I . interwinding capacitances of the input noise current is converted into a me-

Fig. 8 (left). Conversion of longitudinal current to a metallic circuit voltage by interwinding capacitances in input trans-
former. Fig. 9 (right). Conversion of longitudinal current to metallic circuit voltage by capacitance unbalance.

18 AUDIO ENGINEERING FEBRUARY, 1950

tallic-circuit noise voltage and thus verted into a rnetallic circuit voltage grid to gr&nd voltage er and i t also
becomes a source of noise$in an audio is depicted in F i g . 9. The conditions produces the small primary metallic
system. Whether this noise source is assumed here are the same as those circuit current i m . This metallic circuit
troublesome or not in a particular sys- discussed in connection with F i g . 8. current is indicated by a dashed arrow
tem depends solely on the magnitude An electrostatic shield has, however, in F i g . 9.
of the longitudinal noise current. The been introduced into the input trans- Effect of Circuit Balancing
magnitude of the longitudinal noise former in order to eliminate the capac- At. this point, the question may be
current is, i n turn, determined in part itance between the primary and sec- raised, "Why not drain the longitudinal
by the size of the coupling impedance ondary windings. Assume that, due to current to ground by shorting the ten-
20. This example shows the desirability the physical construction of the coil, ter tap of the input coil to ground and
of obtaining a large physical separation the capacitances between the ends of thereby eliminate the necessity for an
between power circuits and the input the primary winding and the shield are electrostatic shield and also avoid the
circuits of audio equipment in order not identical. This condition will al- difficulties due to capacitance unbal-
to minimize noise troubles from longi- ways occur, of 'course, when the primary a n c e s i n the input transformer?"
tudinal induction. windings are layer wound over the Grounding the center point of the input
One method of mitigating difficulties electrostatic shield so that one end of circuit, either at the source or the re-
of the type described above is to employ the winding is at greater physical dis- ceiving equipment does, i t is true, elim-
an electrostatic shield i n the i n p u t tance from the shield than the other. inate most of the troubles resulting
traqsformer of the receiving equipment. For purposes of illustration and analy- from longitudinal noise currents but,
Then the disturbing currents will flow sis, these distributed capacitances have under certain conditions, it greatly in-
down the two conductors to ground been assumed to be lumped at the ter- creases the wossibilitv of noise troubles
through the capacitance between pri- minals of the coil and C2 is assumed from longitudinal noise voltages.
mary winding and shield and will not to be larger than 01. The reason for this may be learned
be able to reach the secondary winding. Under the assumption that the cir- from a consideration of Pig. 10, in
Note, however, that in order to be effec- cuit is subject to a longitudinal noise which i t will be assumed that the
tive, the shield must reduce the inter- current, equal longitudinal currents i l l 2 input circuit of the amplifier is subject
winding capacitances to values so small flow down the conductors of the circuit to a longitudinal noise voltage. The
that only a negligible amount of the to ground through the capacitors C1 effect of such a voltage on the circuit
longitudinal current flows from the pri- and C2. Since C1 does not equal C2, is simulated by means of the zero-im-
mary to the secondary winding of the and by supposition equal currents flow pedance generator e l . I t is further as-
transformer. through them to ground, the potential sumed that the source of excitation is
The introduction of an electrostatic drops across the capacitors will not be connected to the input circuit by means
shield i n the input transformer of the equal. Since the two terminals of the of an ideal repeating coil, between
receiving equipment may greatly re- input transformer have different poten- whose center point and ground the
duce the troubles resulting from longi- tials to ground, a potential difference longitudinal voltage is introduced, and
tudinal noise, but it doe4 not entirely must exist across the primary of the that the center tap of the input coil
eliminate them. Another manner in coil. This voltage appears on the sec- on the receiving equipment is strapped
which the longitudinal noise is con- ondary side of the transformer as the to ground.

INDUCTANCE CANCELLATION I N COlL ONLY FLUX PRODUCED

A R

Resistance Unbalance in Loyer-wound Coil BALANCE

CENTER OF COlL
I'

@'MEAN TURN DIA.OF INNER HALF

Do* " " ' " OUTER HALF
EACH HALF CONTAINS N/2 TURNS
WINDING CROSS-SECTION

Fig. 10 (left). Conversion of a longitudinal voltage to a metallic circuit current by metallic circuit unbalances. Fig. I I
(right). Circuit elements (above) which may require adjustment to achieve satisfactory equipment performance in
the presence of longitudinal noise. Equivalent bridge circuit (below) for analytical purposes.

AUDIO ENGINEERING FEBRUARY, 1950 19

 The longitudinal current'produced by in the two potential drops will cause
the applied voltage is il = 4et/(R+Ro) a metallic-circuit current i m to flow
in which Re is the resistance of the in the input circuit of the equipment
conductors of the input circuit and R of the correct magnitude to make the
is the resistance of the primary winding potential drops along the two paths to
of the input transformer. Note that ground equal. However, the flow of the
neither the primary inductance of the metallic circuit current im, indicated
input transformer or of the repeating by the dashed arrow in the figure,
coil, nor the internal input impedance through the primary winding of the CHASSISOF AMP.
of the amplifier, nor the output im- input transformer sets up a voltage IS CONNECTED----
pedance of the input equipment appears Vtn across the terminals of the receiving
in this =pression. These latter ,factors amplifier. This then is another means
cancel out because the longitudinal cir- by which longitudinal noise is converted
cuit currents flow in opposing direc- into metallic-circuit noise.
tions to ground through the primary Impedenee Unbalance
windings of the tcoils, and hence the The actual means for converting the
associated magnetic fluxes set UP by longitudinal noise voltage into metallic-
them cancel out, as indicated in a, b,
end c of Fig. 10, The impedance, due
,to the residual leakage flux will, in Fig. 14. Multiple grounding method.
general, be negligible with respect to -
the winding resistance in the frequency noise. The very act of,putting center
band of interest, i.e. power frequencies tap grounds on a circuit, while render-
and their important harmonics. The ing the circuit relatively insensitive to
repeating wil has, of course, no leak- longitudinal noise currents, greatly in-
age by the assumption that it is an creases its susceptibility to longitudinal
ideal transformer. noise voltages. Likewise, operating a
Consider first the consequences of - circuit without center point grounds
resistance unbalances only on this cir- makes it relatively insensitive to longi-
tudinal noise voltages and markedly
increases its sensitivity to lon,@tudinal
EQUIPMENTUNDER TEST-. noise currents.
I n actual conditions of operation, the
input circuits of audio s y s t m may be
subject simultaneously to both types of
longitudinal noise, and the problem is
C3. Analysis of source of noise in therefore to uncover a general solution
Z
73 an a.c.-d.c. amplifier. that will protect the circuit under both
P
I9 kinds of exposure. The solution, as may
u, circuit noise, in the general case, is
W impedance unbalance in the two halves have been anticipated, involves the
I
of the input circuit. This impedance simultaneous adjustment of all the fac-
o
unbalance may exist in the input cir- tors discussed so far. The problem which
must be solved in a given design may
-
977
3
cuit of the receiving equipment, in the
connecting pair, or it may originate in be grasped by a consideration of Fig. 11.
the input equipment. Unbalance any- The metallic circuit voltage V m must
Fig. 12. Test circuit for laboratory where in the two halves of the circuit be reduced to a negligible quantity for
evaluation of performance of equip- two conditions of operation: with 21.
will, under the conditions 'assumed in
ment in the presence of longitudinal Fig. lo, comert part of the longitudinal equal to zero, representing the case of
voltage or current. a longitudinal noise voltage; and with
current into a metallic-circuit current.
cuit. I t is assumed in the illustration The remedy which this analysis sug- 21. large compared with the circuit
that the input transformer is a layer- gests is to reduce the circuit unbal- impedance to ground, representing the
wound coil, and hence the resistance ances and so protect the circuit against case of a longitudinal noise current.
of its inner winding is less than that longitudinal noise voltages. \ The elements of the circuit which
of its outer winding. This resistance The magnitude of the metallic-cir- must be adjusted to achieve the desired
unbalance in the coil is designated AR. cuit voltage introduced in the input objective are shown as circuit variables
I t is also assumed that the input con- circuit, under the conditions assumed (it ..iq assumed that an electrostatic
ductors are slightly unbalanced, and in Fig. 10, is a function of the magqi- shield is incorporated in the input
this conductor resistance unbalance is tude of the longitudinal current. This, transformer in order to eliminate the
designated ARC.On the assumption that therefore, suggests that an alternate transformer interwindingcapacitances):
the unbalances are a small part of their remedy in tNis case is to remove a cen- The circuit of Fig. 11is also presented
respective resistances, their effect on ter tap ground froni the circuit in order schematically in the alternate Wheat-
the circuit may be determined by as- to reduce the longitudinal current to n stone bridge form. The problem is then
suming that equal longitudinal cur- negligible quantity and thereby reduce to adjust the circuit variables so as to
rents i t / l flow down the two conductors the metallic-current resulting from reduce the bridge output to zero in the
to ground. The flow of these equal cur- circuit unbalances. I t is thus apparent presence of the longitudinal voltage e l .
rents through resistances which differ that the expedient of operating the The variables which require adjustment
slightly in magnitude will produce input circuit of the equipment with a are the series circuit impedance un-
slightly different potential drops along center-point ground is not a general balances represented in the figure as
the two paths to ground. This difference protective measure against longitudinal conductor and coil resistance unbal-

20 . AUDIO ENGINEERING e FEBRUARY, 1950

ances, the impedance unbalancee to zero-longitudinal current it. The mea-
ground of the circuit represented in the surement is therefore made with the
figure as capacitances, and the imp* center point of the input circuit of the
dance to ground i n the center tap cir- equipment open circuit to ground. The
cuit of the input transformer. longitudinal voltage suppression of the
The entire input circuit of an audio equipment, under these conditions of
system must be so designed that it will operation, is the ratio of Vra to Vm. I n POSITIONS OPEN and AVAILABLE
perform i n an acceptable manner i n the decibels, it is 20 loglo Via/Vm and i t PERSONNEL may be listed here at no
presence of longitudinal noise. I n gen- should be determined over the appro- charge to industry or to members of
eral, this means that appropriate pro- priate frequency band. the Society. For insertion in this col-
tective measures against longitudinal The performance of the equipment in umn, brief announcements should be in
induction must be incorporated in all the presence of a longitudinal current the hands of the Secretary, Audio En-
the elements of the input circuit, in- is determined by measuring the ratio gineering Society, Box F, Oceanside,
cluding input equipment and the con- of the metallic circuit voltage V m to N. Y., before the fifth of the month pre-
necting pair, as well as the receiving the, longitudinal current iz under the ceding the date of issue.
equipment. This is true because both condition of minimum-preferably zero Wanted: E. E., electronics major,
the input equipment and the connect- -longitudinal voltage Vr.. This meas- with extensive bkgnd in magnetic and
ing pair can equally well convert the urement is therefore made with the audio cct design a n d acoustical theory
longitudinal noise into metallic-circuit center point of the input circuit of the and practice. Must have at least 5 yrs
noise. The problem is not restricted to equipment shorted to ground. The lon- exp. in product design on products now
being sold nationally. Must be capable of
amplifiers alone as longitudinal conver- gitudinal current suppression of the following product from experimental
sion sources, but is likewise true for equipment, under these conditions of through production, and be responsible
microphones, phonograph pickups and operations, is the ratio of V m to ir. It is for specifications, quality control, field
tests, operation and service manuals on
all other types of input equipment. generally expressed as so many micro- product developed. Must be creative and
volts per ampere and should be deter- have an exceptionally high degree of
Equipment Limitations mined over the appropriate frequency mechanical aptitude. State age, education,
band. and qualifications when answering. Loca-
The limitations of a piece of equip tion: Minnesota. Box 102.
ment must be understood in order to Field vs Test Performance e Communications Engineer (MIT), 22,
use it effectively, and these limitations The correlation between the perform- single, with good theoretical background
are generally established by suitable and some experience; interested in re-
ance of a piece of equipment when in search, development, or teaching in audio,
test procedures. From the preceding a test circuit and when installed in the acoustics, electric circuits, and vacuum-
discussion, it is apparent that data is field is often difficult to establish be- tube circuits. Box 111.
desired, i n appropriate form, on the cause of the wide range of field operat- Graduate Student of radio and tele-
equipment i n the presence of both types vision desires Junior Engineering posi-
ing conditions. This situation is par- tion in audio or recording industry. Age
of longitudinal induction. This means ticularly true, when i t comes to predict- 23, married, child. Willing to travel oc-
that the test circuit should provide a ing with accuracy, on the basis of lab- casionally. Prefer midwest or south. Box
measure of the effect of the series im- oratory test data, the longitudinal sup- 113.
pedance unbalances of the two halves pression performance of equipment. Recording Engineer. Experienced
of the metallic circuit and of their mixer; knowledge of classical music.
Hence, in the remainder of this article, Knows motion picture and radio mixing
impedance unbalances to ground, and some of the limitations and special con- techniques, 30, married, presently em-
should also reflect the effect of the ditions encountered will be indicated ployed. Graduate of Hollywood Sound
other factors such as interwinding ca- briefly as a guide to the wide range of Institute. Box 121.
pacitances that contribute to poor per- problems encountered in the practical Audio Engineer. Graduate Hollywood
formance i n the presence of longitu- Sound Institute, 25, married. Now attend-
application of this information. A ing television production school. Desires
dinal noise. An appropriate test circuit source of possible discrepancy between position with broadcast or recording
for this purpose is shown i n Fig. 12. predicted and actual performance re- studio, evenings, N.Y.C. Free to travel for
The test circuit of Pig.12 is arranged full-time position. Box 1 2 2 .
sides in the fact that lumped noise Graduate: Advanced Technology
to impress the longitudinal voltage on sources are employed in both the analy- course at RCA Institutes, 26, married,
the equipment under test via the mid- sis and test circuits, whereas the noise with knowledge of music, desires position
point of the source impedance for which experienced in the field is usually that in audio field; salary secondary. Box 101.
the equipment was designed. The longi- due to a distributed source. I n the case Audio Engineer: M.S. Physics, Electri-
t u d i n a 1 voltage may be impressed cal Engr; ten years research, develop-
of the interconnecting pair between ment, design experience with sound re-
through an impedance 2 1 .whose value the sending and receiving equipment, cording, acoustic measurements, and
is dictated either by the conditions of the metallic-circuit impedances and the transducers. Thorough bkgnd in magnetic
the test or the sensitivity of the instru- impedances to ground are, in addition, and mechanical recording including mag-
ments used in making the test. Appro- netic recording systems for computer ap-
distributed rather than lumped de- plications. Presently employed, prefer
priate precautions should be taken to ments. It generally will be found, how- firm in which could invest capital, small
insure that the test circuit is itself not ever, that a satisfactory correlation be- city or suburban location. Box 201.
a source of error. tween the longitudinal suppression per- * Graduate VTI: employed at present.
formance of a piece of equipment in a Desires position in audio field. Some ex-
The test circuit is employed to evalu- perience. Age 20, single, and in good
ate the performance of the equipment test circuit and in the field can be estab- health. Willing to learn, and will go any-
in the presence of a longitudinal volt- lished when the effect of these factors where. Box 202.
age and in the presence of a longi- are correctly evaluated. Audio Engineer: experienced man
tudinal current. The performance of the Another factor of importance is that with family desires position in recording,
broadcasting, film,or TV sound. Excellent
equipment i n the presence of a longitu- in this discussion it has been assumed operator. troubleshooter, and mainte-
dinal voltage is determined by measur- that the longitudinal noise is introduced nance man. Superior knowledge of classi-
ing the metallic-circuit voltage V m and into the input circuit between the send- cal music. Studio and equipment design
the longitudinal voltage Vta under the ing and receiving equipment. However, and construction experience. Fluent Eng-
lish, French, and German. Willing to re-
condition of minimum - preferably [Continued 0% page 341 locate. Box 203.

AUDIO ENGINEERING FEBRUARY, 1950

LONGITUDINAL NOISE
it often happens that the noise i s intro- been connected to a quiet audio ground,
duced, by means so far not considered, but that the panel has been connected
at other ~ o i n t sof the circuit. A well to the conduit of the power circuit. I t
known illustration and one which p r e is assumed, i n addition, that the secon-
sents many design difficulties is that dary winding of the input transformer
of an a.c.-d.c. amplifier operated on a.c. has appreciable capitance to its case
I n this case, the amplifier itself is the and core which are electrically con-
source of the noise. This is true because nected t o the amplifier panel. Substan-
the secondary side of the input trans- tial currents originating from external
former is directly connected to the a.c. sources are presumed to be returning to
power circuit. Poling of the power plug ground via the conduit. This condition
on the amplifier cannot be relied upon sets up a potential difference between
to reduce its noise, because i n many the amplifier panel and the audio cir-
areas the so-called low side of the cuit ground and causes noise currents
power circuit has a substantial voltage to flow from the conduit to the audio
to ground. I n addition, this voltage ground via the secondary winding of
often includes substantial amounts of the transformer and its associated capa-
the higher harmonic and therefore the citance to core and case. This noise
more disturbing components of thg ourrent introduces a voltage into the
power supply frequency. equipment on the grid side of the input
The problem i n designing a universal transformer. Installations in which
a:c.-d.c. amplifier is thus to make a unit transients on the power circuit appear
with acceptable performance for either i n the output of the system may be sub-
polarity of the a.c. power supply. This ject to noise trouble of this type. This
means that the signal-to-noise
- ratio difficulty may be eliminated by employ-
of the amplifier must be acceptable ing the audio circuit ground for both
when the full power circuit voltage is panel and circuit ground purposes.
impressed between the windings of the Summarizing, then, there are two
transformer i n the manner shown in general means by which longitudinal
Fig. 13. One method of solving this noise is introduced in a circuit, one
problem is by the use of two separate of -high internal impedance so that the
electrostatic shields. One shield encloses noise has the characteristics of a con-
the secondary winding and is connected stant current introduced into the cir-
to the low side of this winding. The cuit, the other of substantially zero in-
sther shield encloses the primary wind- ternal impedance so that the noise has
ing and is connected to the audio cir- the characteristics of a constant volt-
cuit ground associated with the input age introduced into the circuit. It has
circuit. This arrangement virtually also been shown that these two types of
eliminates the parasitic coupling capa- induced noise affect a circuit in differ-
citances between the primary and s e e ent manners and therefore require wide-
ondary windings of the transformer ly different treatment to avoid their
and thus substantially eliminates the unwanted effect on a circuit. Super-
flow of the longitudinal current through ficial remedies to render a circuit in-
the transformer windings from this sensitive to longitudinal noise are as
cause. It also eliminates the flow of apt to increase the difficulty as to miti-
the longitudinal current from winding gate i t because of the diverse character
to its associated shield since each wind- of its two types. However, as outlined
ing is at its shield potential, from a in this article, once the nature of lon-
longitudinal circuit point of view, by gitudinal noise induction is under-
virtue of the connection between them. stood, it i s as amenable to reduction of
The longitudinal current flow is thus its disturbing effects as many of the
from one shield to the other, but this other sources of noise with which the
current flow will not degrade the signal- audio engineer must contend.
to-noise ratio of the amplifier.
Multiple Grounds
Another manner i n which longitudi-
nal noise may be introduced i n a system
at a point other than the interconnect-
ing pair i s depicted i n Fig. 14. In this
case, it i s assumed that the panel and
circuit grounds on the amplifier have
been separated for utmost flexibility in
application. It is further assumed that
on installation the circuit ground has

AUDIO ENGINEERING FEBRUARY, 1950