U.D.C. 621.375.

Parametric Amplifiers: Static and Dynamic Inductance and Capacitance and their Significance in the Non-Linear and Time-Varying Approaches
By F. J.HYDE, D.Sc.f
AND
This paper is sponsored by the Institution's Education Group. Summary: The analysis of circuits such as parametric amplifiers can be approached in two main ways, by regarding the variable element as having either non-linear characteristics largely controlled by a dominant pumping current or voltage, or linear characteristics subject to a specified timevariation. In the former case, "dynamic" inductance or capacitance may be used with advantage in analysis and measurement in place of the "static" parameters. In the latter case the distinction is invalid. The relation between the two cases is discussed in this paper.

Professor D. G. TUCKER, D.Sc.

1. Introduction

The analysis of circuits such as rectifier modulators and parametric amplifiers can be approached in two main ways:1'2 in specific configurations it is often convenient to regard the rectifier, inductor or capacitor as a non-linear element largely controlled by a dominant carrier or pumping current (or voltage), while in the development from general network theory it is convenient to regard the variable elements as being linear but having a specified time-variation. In the case of rectifiers and time-varying resistances there is generally no difficulty in reconciling these approaches, and there is ample literature. For the non-linear or time-varying inductance and capacitance it seems that the matter still requires some clarification. One of the authors has recently discussed3 the use of "dynamic" parameters of inductance and capacitance in place of "static" parameters in the analysis of parametric amplifiers by the non-linear approach. Whereas static inductance and capacitance are defined as where < = flux-linkage and i = current and Cs = q/v where q = charge and v = voltage the dynamic parameters are defined as: Ld = d(f>/di
and Cd = dq/dv

easily than Cs. It is often therefore advantageous and more realistic to use Ld and Cd in the analysis. The other author has been concerned with teaching the theory of circuits with time-varying parameters,4 and discussion has arisen between the authors on the significance of the static and dynamic conceptions in this approach. The conclusions reached are made clear in the following treatment of the parametric amplifier with pumped inductance. The case of pumped capacitance is the exact dual of this, and needs no separate treatment. The symbols used in this paper are, as far as possible, the same as in the paper4 on time-varying parameters.
2. Non-linear Approach

In the approach through non-linearity we assume currents in the inductance thus: / = / cos (opt due to the pumping source at angular frequency cop *o = *o c o s (coqt+0o)aX the angular frequency (coq) of the input signal, i + 1 = ?+1cos [(coq + (op)t+d+l)] at the angular frequency (coq + cop) of the output signal. It is assumed that the circuit is tuned so that no other currents exist. Now we have to ascribe a non-linear function, i.e. a current dependence, to the inductance. It is complicated to analyse the problem in general terms, but analysis is simple if we assume a second-order powerseries dependence offluxon current, namely, giving static inductance, L.-0/i-Lo+Ai and dynamic inductance,
Ld = dcf>ldi = L0+ZLt where i = / + i0 + i+1. i = L 0 (l + ai) say

The use of dynamic parameters has some advantages. In making practical measurements of the parameters, Ld is more easily measured than Ls, and Cd more
t Department of Electronic Engineering, University College of North Wales, Bangor. X Electrical Engineering Department, University of Birmingham.
Journal Brit.I.R.E., April 1963

(1)
(2)

353

F. J. HYDE and D. G. TUCKER

To obtain this kind of non-linear law clearly requires the presence of a bias current in a real inductor. The voltage v across the inductance may be expressed as either

di

di

according to whether static or dynamic inductance is used. Now eqn. (4) gives simply (5)

and eqn. (3) is readily shown to give exactly the same result. On separating out the terms of the various frequencies, and assuming that no currents other than /, i0 and /+ 1 have finite amplitude, eqn. (5) leads readily to the equations for frequencies coq and coq + cop. These may be regarded as the basic equations of the circuit of Fig. 1 on putting v = E cos coqtZ0 i0 cos (coqt + d0) at frequency coq and v = Z++1 1 t + 1 cos [ ( q p at frequency coq + cop. Replacing the cosine terms by current vectors i 0 (carrying the phase angle 60) and i + 1 (carrying the phase angle 0 + 1 ), and putting L aL0I, we obtain the equations: For frequency coq: E - Zoio = }coqL0i0 +}icoqLi i+1 (6) For frequency coq + cop:

third equation, for frequency cop, must be taken so that a solution is obtained for / as well as i0 and * +1 . It is apparent that, for the particular dependence of flux on current which has been assumed, the only difference in the analysis between the use of static inductance (Ls) and dynamic inductance (Ld) is that the variable part of the former has to be only half as great as that of the latter to give the same result. Note that if we had taken an infinite power series to represent the non-linearity in eqns. (1) and (2), then to obtain eqn. (6) and (7) we should have had to specify that the pumping current was much larger than the other currents. Furthermore, we should have had to define L1 as an infinite series involving all odd-order coefficients and powers of /, and L o as an infinite series involving all even-order coefficients and powers of /. From equations (6) and (7) the performance of the up-converter or of the negative-resistance amplifier is readily worked out. 4 (In the latter case, the frequency a>q wp has to be substituted for coq + cop in all the working.)
3. Time-varying Approach

In the approach through time-varying parameters, we put Ls or Ld as a function of time, but not of current. Thus a Fourier expansion is possible:

= n~0 LH cos ncopt

(8) (9)

~d7 = 7 or Ld = Ls

where / now comprises only i0 and i+1. Since the inductance is not a function of current,

driving current / at frequency cop through it. In the time-varying approach, L is a specified periodic function of time with fundamental frequency cop. 354

-z+ii+i = ji(o> + o)p)Lio + j(o> + p)o *+i (7) It is implied above that 1 is given; if it is not, then a and both ~ and are the same functions of time, di i There is therefore no distinction between static and TUNED TO CUq* TUNED TO Wq dynamic parameters in this approach. Thus we cannot apply eqn. (4) to this case, and the expansion must be made only from eqns. (8) and (9) taken together. Expansion leads to exactly the same equations as (6) and (7), (where Lt is now the value of Ln in (8) for n = 1), without the loss of generality implied by the limited non-linear representation of eqn. (1); on the Ecosud other hand the assumption of linear response to the signal has been necessary. This is equivalent to saying that, when pumping is done by a current driven Fig. 1. Parametric amplifier. through the inductance rather than by mechanical In the non-linear approach, L is a function of the total current variation, the pumping current must be much larger through it, and has also a third circuit, with a generator, than the signal currents. The fact that identical equations are obtained by the two approaches needs a little further explanation
Journal Brit.l.R.E.

PARAMETRIC AMPLIFIERS

in view of the fact that Lt is defined differently in the two cases. Referring to the second paragraph from the end of Section 2, and observing that Lo is obviously the same in both cases, it becomes clear that with a dominant pumping current, Lt determined from the general non-linear equation must be identical with Lx in the Fourier series of eqn. (8), since the two circuits are now operationally identical, have identical equations, and must have identical solutions.5 The fact that L1 is also the same for the simple non-linear law of eqns. (1) and (2), which does not require the assumption of a dominant pumping current, indicates that this is evidently a special case.
4. References
1. S. Bloom and K. K. N. Chang, "Theory of parametric amplification using non-linear reactances", R.C.A. Rev., 18, p. 578, December 1957.

2. H. Heffner and G. Wade, "Gain, bandwidth and noise characteristics of the variable-parameter amplifier", / . Appl. Phys., 29, p. 1321, September 1958. 3. F. J. Hyde, Discussion on paper by A. L. Cullen, "Theory of the travelling-wave parametric amplifier", Proc. Instn Elect. Engrs, 107, Part B, p. 124, 1960. 4. D. G. Tucker, "Circuits with time-varying parameters", / . Brit.I.R.E., 25, No. 3, p. 263, March 1963. 5. F. J. Hyde, "Analysis of parametric amplifiers incorporating varactor diodes", Proc. Instn Elect. Engrs, in course of publication. (This paper gives a fuller discussion of the matter.) Manuscript first received by the Institution on 20th May 1962 and in final form on 14th December 1962 {Contribution No. 63/ Ed.10).
The British Institution of Radio Engineers, 1963

April 1963

355