WIRELESS WORLD MAY 1981 51

Wien-bridge oscillator with low

harmonic distortion
New way of using Wien network to give 0.001 % t.h.d.
by J. L. Linsley Hood, Robins (Electronics)

The Wien-bridge network can be 1kHz of some 0.003%, which tended to

connected in a different way in an increase with frequency above this point,
R
oscillator circuit to give a sine wave as the effectiveness of the common-mode
with very low total harmonic isolation deteriorated.
distortion. An I.e.d/photocell However, it is not implicit, in the use of
amplitude control is external to the a Wien network as the frequency-control Output
circuit. method, that the configuration shown in
Fig. 1, in which the output of the network
is taken to the non-inverting input of the
The Wien-bridge network remains the amplifier and the amplitude controlling R

most popular method of construction of negative-feedback signal is taken to the

variable-frequency sine-wave oscillators, other, is the only circuit configuration
since the basic circuit can be very simple in which can be employed. In particular, con­ ov
form. It is a fairly straightforward matter sideration of the phase and transmission
to design oscillators of this type in which characteristics of such a network, shown in
the harmonic distortion is only of the order Table 1 and Fig. 2 for equal values of C
Fig. 1. Basic Wien-bridge oscillator circuit
of 0.01-0.02%, and which allow frequency
control by means of a simple 2-gang poten­ Fig. 2. Gain and phase characteristics of
tiometer. Wien network 1-0 131
The basic circuit for an oscillator of this
O· 9
form, using a single operational amplifier z
o8 0
w_
as the gain block, is shown in Fig. 1, and >Vl
_Vl
the author has shown a practical design of z
O·7 �i: 121
-IVl
oscillator, based on this, for a use as a §l
Vl
O· 6 wz
0:«
" 0:
simple, general-purpose workshop tool. 1 i: O· ....... f-
Vl
However, in the form shown in Fig. 1, a
�O4 .... ..
significant problem exists in that the trans­ 0:
f- O. 3
..... 111
w
mission of a normal Wien network, at the
O·2 ,/7 " ....�--
...z '/}:;II)' 0 Vl;!
operating frequency, is only 113, which
O·1 / '-... � CL
means that an inconveniently large propor­
0 �� I
tion of the output signal voltage appears at
the inputs of the amplifier, and will lead to 01 02 04 0·6 0 2 4�
non-linearities in the transfer characteris­ OSCILLATION FREQUENCY - 9Cf'
tics of the amplifier due to 'common mode'
defects. An oscillator design, which em­
ployed an input device operated in a cas­ Fig. 3. Rearrangement of Wien network
R
code configuration with a junction f.e.t. to between signal sources gives small in­
minimize this type of defect, was shown by phase signal at point X

the author in 1977,6 and allowed a t.h.d. at . ....'--

-....- .. ov
(al Ibl
Fig. 4. Use of arrangement of Fig. 3 in
TABLE 1. Phase and transmission charac­ oscillator circuit
teristics of simple Wien network.

FIFa phase transmission

0.1 73.14° 0.10 R
0.2 57.99° 0.18
0.3 45.32° 0.23
0.4 34.99° 0.27
0.5 26.57" 0.30 Outr>ut
0.6 19.57° 0.31 I-Eyl
0.7 13.65° 0.32
0.8 8.53° 0.33 R
0.9 4.03° 0.33
1.0 0° 0.33
1.2 -6.97" 0.33
1.5 -15.52° 0.32
2 -26.57" 0.30 --O--4�-----�---_ V
3 -41.63° 0.25
Figures in brackets
5 -57.99° 0.18
refer to diagram 3 b
8 -69.15° 0.12
10. -73.14° 0.10
52 WIRELESS WORLD MAY 1981
and R, implies that if, instead of the net­ be significantly lower, because of the very operating frequency of the oscillator. Fast
work of Fig. 3(a) being connected between small input-signal amplitude and the ab­ response-speed, high h.f. gain op.-amps.
a signal source Ein and the OV line, it was sence of any internal transfer errors be­ can therefore be used without problems.
connected between two signal sources + E x tween the inverting and non-inverting in­
and -Ey, where these are sinusoidal and puts, than is the case for an identical For these reasons, it can be expected
identical in frequency and the negative amplifying element in a series-feedback that the residual harmonic distortion of
sign implies phase opposition, as shown in configuration.3,4 this oscillator design will be exceedingly
Fig. 3(b), then a small, in-phase signal • The time-delay errors in the second am­ small, and measurements on two proto­
would exist at the point 'X', at the fre­ plifying stage (A2) no longer contribute to types have indeed shown this to be the
quency of maximum transmission, (fo), if loss of stablility in the system, but only to a case. So far as can be determined, the
+Ex was slightly greater than -2Ey• very small compensatory shift in the residual distortion - almost exclusively
This could then be used as a positive­
feedback signal in a circuit such as that
shown in Fig. 4, to sustain oscillation at R

the frequency fo. Indeed, such a circuit

will work quite well, and will sustain a Thermistor Rs

constant output magnitude of oscillation if

a thermistor is employed, as shown, to
make the gain of the second, inverting, Output
amplifier stage dependent on the ampli­
tude of the input signal. However, there is,
in practice, a small snag with such an
R
arrangement, and that is that the inverted
negative-feedback signal applied to the in­
put of Al will suffer an additional phase ____ ��--------------��------------��OV
error due to the internal time lag within Figures in brackets R p. Rs and Rin chosen to
refer to diagnom 3b SUIt individual thermistor
A2, and this will cause unwanted h.f. insta­
employed
bility if '3rd generation' high speed op.­
amps. such as the CA 3 14 0, or the 174 1 S, Fig. 5. Final form of new configuration in low-distortion oscillator
are used in the realisation of this circuit.
It is, fortunately, an easy matter to re­
solve this difficulty if the circuit is recast in 0·01
·009
the form shown in Fig. 5, in which the , (Includes
,50 Hz "hum" I
negative-feedback signal, equivalent to ·005 "
-Ey in Fig. 3(b), is derived from the am­
plifier AI> and the positive-feedback signal "-
,
is obtained from the output of the second � "-
.....
inverting amplifier A2• :z
o ..... ....
This configuration offers several signifi­ � 0 001
....
o ...
cant advantages. l­ ... �---
V)

• The input signal to Al is extremely a .0005

small, since it is only required to be
EOUl/2M, where M is the open-loop gain of
AI - typically 100dB for a good modern
op.amp. i.c. - and, as pointed out by the
2 0 0001
author in an earlier article , with semi­
100 lk 1Ok
conductor amplifiers the non-linearity of
FREQUENCY (Hz)
such devices is essentially an input charac­
teristic, dependent on the magnitude of Fig. 6. Measured total harmonic distortion of improved oscillator of Fig. 5
the input signal.
• The second-stage amplifier is operated as

a shunt-feedback element, and the non­ Fig. 7. New oscillator with external optoelectronic amplitude-control circuit. Silonex
linearities of such a stage can be shown to (formerly National Semiconductors) cell, Type NSL395, is obtainable from Cheston
Electronics Ltd., Vanguard House, 56 Oughton Street, Ormskirk, Lancs. Tel: 069572456

1/ Photoconductive
cell

10k NSL395

10}!
47k
Si

3k3 56k
Si

Si

I1
3k3
3k3

3k3

\�------�vr--�/ \�------�vr--�/ \�------,v�--�/

Oscillator Full wave rectifier Feedback control circuit

Output
Note: PC and LED are in optical contact
1 1
WIRELESS WORLD MAY 1981 53

third harmonic - is that due to the de­

pendence of the resistance of the thermis·
photo-conductive cell and the light-emit­
ting diode combination shown in Fig. 7, in
Measuring tranSient
tors used to control the amplitude of the
oscillation on the instantaneous value of
which the time constant and other dy­ intermodulation
namic characteristics of the control circuit
continued from page 47
the signal potential applied to them. This can be optimized by a suitable combina­
limiting filter.
characteristic of oscillators with averaging tion of proportional, integral and differen­
�
control s stems has been analysed by tial (p.i.d.) adjustment to the gain of the
The authors are interested in measuring
t.i.m. principally to test the effectiveness
Robinson who suggests that the distortion control circuit (A2). Needless to say, the
of anti-t.i.m. measures such as input filt­
of such a system, which is shown to be photoresistive element should be chosen to
ers, and to design low-t.i.m. monolithic
mainly third harmonic, will be have a very low voltage coefficient of resis­
amplifiers. The availability of a simple and
tance and an adequate response speed to
X3 1 . Ao-l'j 1 accurate measuring system has already
--
_

avoid the introduction of a further signifi­

Xl - 8rj l'j . 'Iitji provided useful results, exemplified by the
cant time delay into the control loop.
R.I.A.A. preamplifier shown in Fig. 12; a
where (Ao-l'j)/l'j is the fraction by which Leaving aside the question of the means
circuit designed around the TDA32 10 ste­
the low-level loop gain exceeds the gain employed to control the amplitude of the
reo preamplifier i.c. The filter on the out­
required to initiate oscillation, and T is the output signal (which imposes limitations of
put is intended to minimize t.i.m. in the
time constant of the control system (ther­ an identical kind on any oscillator system,
next stage. This circuit, in terms of tradi­
mistor or similar). In the case of a Wien­ in terms of the settling time, and the in­
tional parameters, represents the current
bridge oscillator, l'j == 3. fluence of the control time constant on the
state-of-the-art in i.c. R.I.A.A. preamps in
This equation indicates that if the feed­ harmonic distortion at any given fre­
which the total harmonic distortion is
back amplitude is very little above that quency) the improvement in performance
0.02% at 20kHz. The frequency response
required to sustain oscillation - which is given by the circuit design shown in Fig. 5
is 20Hz to 20kHz ±O.SdB and the dy­
implicit in the design - the residual over that obtainable from the more con­
namic range 100dB.
distortion will be dependent on the time ventional arrangement shown in Fig. 1,
constant of the control mechanism. By the suggests that it would be sensible to regard
Further reading
use of series and parallel resistors of appro­ the improved circuit as a general replace­
T. Roddam - Calculating transient response,
priate values with the thermistor, this can lment for the earlier system in all future Wireless World, Aug. 52,pp. 292-295.
be made to control the amplitude of the designs. M. Otala - Transient distortion in transisto­
oscillation at a resistance value which is References. rized audio power amplifiers IEEE Tram.
only a little less than its room-temperature I. Linsley Hood, J. L., Hi-Pi News and Record Audio and Electroacoustics, Sept. 70, pp. 234-
value. Under these circumstances, the Review, March 1975,pp 63-67. 239.
settling time of the amplitude is long - 2. Linsley Hood, J. L., Wireless World, Sept. J. R. Stuart - An approach to audio amplifier
design, Wireless World, August 73,pp. 387-391
perhaps 3-4 seconds at 1kHz, but the 1971,p437.
t.h.d. will be very low. The penalty in­ 3. Linsley Hood, J. L., Wireless World, Jan. - Sept. 73, pp. 439-446 - Oct. 73,pp 491-
1973,pp11-12. www.keith-snook.info 494.
curred in this type of adjustment, apart
4. Taylor, E. F., Wireless World, April 1973,p M. K. Vander Kooj - Predicting and avoiding
from the obvious inconvenience of a relati­ slew-rate limiting, Electronic Engineering, Febr.
194.
vely long settling time following any dis­ 73,pp. 18-19.
5. Robinson, F. N. H., Intl. J. Electronics,
turbance, is that the two gangs of the po­ No. 2 pp 137-148. (1980) W. M. Leach - Transient i.m. distortion in
tentiometer used to control the operating 6. Linsley Hood, J. L., Wireless World, Sept. power amplifiers, Audio, Febr. 75,pp. 34-42.
frequency of the oscillator must be B. Sundqvist - Transient intermodulation in
1977,pp 40-42 D
reasonably well matched in resistance amplifiers, Wireless World, Febr. 77,pp. 37-39
M. Otala - Non·linear distortion in audio am­
value across the adjustment range, and
also, if switched capacitors are used to The impact of new plifiers, Wireless World, Jan. 77,pp. 41-43.
E. Leinonen, M. Otala and J. Curl - A method
provide step changes in frequency sweep, technology at work for measuring transient intermodulation distor­
the ratios of their values must remain the tion, AES Journal, July 77.
More than £90,000 is being made available over
same. three years by two research councils for studies E. Leinonen and M. Otala - Correlation audio
However, this is merely a statement of into the shop-floor impact of the introduction of distortion measurements AES Journal, Jan. 78,
the obvious, that it is a pointless exercise to new technology. pp. 12-19.
try to design high-performance equipment Three studies are planned: on telephone ex­ W. G. Jung, M. L. Stephens and C. Todd -
using low-performance c o m p o n e n t s . change modernisation; on the adoption of a An overview of SID and TIM Audio, June 79,
computer-based freight information system in pp. 58-79 - July 79,pp. 38-47 - Aug. 79 pp.
Nevertheless, within the limitations im­
British Rail; and on the introduction of 42-59.
posed by the use of a thermistor as the
electronic news gathering (ENG) equipment in R. Cordell - Comments on "A method for
stabilizing element, the performance of a
television. measuring TIM" and "Correlation of audio
very simple oscillator, built around a dual distortion measurements", AES Journal, April
The Science and Social Science Research
operational amplifier (a Texas Instruments Councils are sponsoring the work which will be 79,pp. 295-300.
TL072), is very good, as is shown in Fig. carried out by the New Technology Research R. Cordell - Another view of TIM, Audio,
6. The total harmonic distortion from this Group of Southampton University. The Group Feb. 80,pp. 38-49 and March 80,pp. 39-42.
arrangment, in which the resistors asso­ has been formed by engineers and social scient­ S. Takahashi, S. Tanaka - A method of mea­
ciated with the thermistor were adjusted to ists committed to interdisciplinary research "on s u r i n g i n t e r m o d u l a t i o n d i s t o r t i o n, A E S
the introduction of new electronic and computer Convention, May 79.
give a settling time of 5 seconds at 1 kHz,
technologies at the level of the individual S. Takahashi, S. Tanaka - A new method of
and an output voltage of 2 volts r.m.s., is'
workplace". measuring TIM: a comparison with the conven­
lower than that obtainable from any other
The two main objectives of the work will be tional method, AES Convention, Nov. 79.
simple Wien-bridge oscillator (that is to P. J. Baxandall - Audio power amplifier de­
to explore the process of technological change
say with the exception of systems with low­ and to develop i n t e r d i s c i p l i n a r y research sign, Wireless World, Jan. 78, March 78 and
pass output filtering) known to the author. methods for the problems that arise. The team May 78.
This distortion is almost exclusively third will be investigating the nature of technological Yuri Miloslavskij - Audio preamplifier with no
harmonic - decreasing with frequency - innovation and engineering decision-making in TIM, Wireless World, Aug. 79,pp. 58,59,60,
which implies that the source of this wave­ the economic and social context of business or­ 86. www.keith-snook.info
ganisations; t h e b e a r i n g of organisational A. Morando, Introduzione Analitica alia distor­
form distortion is the instantaneous change
structures on the capacity of managers to gener­ sione d'Intermodulazione Dinamica, IAF.I
in gain of the system, during the excursion
ate methods and mechanisms for the introduc­ suppl. a "Suono" Aprile 77.
of each half sinusoid, due to the limited
tion and control of new technology; th� de­ P. Nuti, Misure di Intermodulazione Dinamica,
thermal inertia of the thermistor. Suono, Nov. 77 pp. 115-120.
velopment of union strategies towards new
The very high performance obtainable technology; the consequences of technological R. A. Belcher, An experimental investigation of
from such a circuit encourages the consid­ change for the nature of work and occupations; test·noise signals for the measurement of non­
eration of alternative methods of ampli­ and the effectiveness of industrial relations linear distortion of sound signal, BBC report
tude control such as that employing a procedures in handling new technology issues. RD1974/2. D