betaFeedback

WorksPhysics and Astronomy

Nonlinear transmission line: shock waves and the simple wave approximation

Abstract

The transmission lines we consider are constructed from the nonlinear inductors and the nonlinear capacitors. In the first part of the paper we additionally include linear ohmic resistors. Thus, the dissipation being taken into account leads to the existence of \mbox{shocks – the} travelling waves with different asymptotically constant values of the voltage (the capacitor charge) and the current before and after the front of the wave. For the particular values of ohmic resistances (corresponding to strong dissipation) we obtain the analytic solution for the profile of a shock wave. Both the charge on a capacitor and current through the inductor are obtained as the functions of the time and space coordinate. In the case of weak dissipation, we obtain the stationary dispersive shock waves. In the second part of the paper we consider the nonlinear lossless transmission line. We formulate a simple wave approximation for such transmission line, which decouples left/right-going waves. The approximation can also be used for the lossy transmission line, which is considered in the first part of the paper, to describe the formation of the shock wave (but, of course, not the shock wave itself).

Introduction

The concept that in a nonlinear wave propagation system the various parts of the wave travel with different velocities, and that wave fronts (or tails) can sharpen into shock waves, is deeply imbedded in the classical theory of fluid dynamics[1]. The nonlinear electrical transmission lines (where the nonlinearity can be either due to nonlinear capacitors or nonlinear inductors forming the line, or both) are of much interest both due to their applications, and as the laboratories to study nonlinear waves[2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13]. Nonlinear Transmission Line technology has historically been used for pulse shaping applications and in digitizing oscilloscopes. Over the years it has proven itself to be a highly credible, robust technology.

A very interesting particular type of signals which can propagate along such lines - the shock waves - is attracting interest since long ago[14, 15]. We published a series of papers on the travelling waves in nonlinear transmission lines: the kinks, the solitons and the shocks (see the most recent publication of that series and the references therein[16]).

In the present short note we would like to add to our previous publications on the subject. In the first half of the paper we start from reproducing in a concentrated form the analytic results for the profile of the shock wave in the transmission line with the nonlinear capacitors obtained in Ref.[16]. Those results were obtained by reducing the second order ordinary differential equation describing the travelling waves in the transmission line and factorising the thus obtained equation for the definite values of the parameters. While in the latter publication only the charge of the capacitors was analysed as the function of the time and the coordinate (which is equivalent to studying the voltage in the line as the function of the time and the coordinate), in the present note the time and coordinate dependence of the current is also calculated.

The second half of the paper is dedicated to the presentation of the simple wave approximation to the wave equation describing general lossless transmission line, which allows to decouple the nonlinear wave equation into two separate equations for the right- and left-going waves[17, 18, 19]. While we used this approximation previously for the Josephson transmission line (JTL)[20], here we formulate the approximation for the case when both the inductors and the capacitors are nonlinear[16].

The circuit equations

The transmission line constructed from the identical nonlinear inductors and the identical nonlinear capacitors is shown on Figure 1. We take the capacitors charges

Lossy nonlinear transmission line circuit with nonlinear inductors, nonlinear capacitors, and resistors

Figure 1. Lossy nonlinear transmission line

QnQ_n and the integrated voltages on the inductors

Φn≡∫Vn dt(1)\Phi_n \equiv\int V_n\,\mathrm{d}t \tag*{(1)}

as the dynamical variables. The circuit equations (Kirchhoff laws) are

dQndt=In−In+1+1RLddt(Φn−Φn+1),(2)\frac{\mathrm{d}Q_n}{\mathrm{d}t}=I_n-I_{n+1}+\frac{1}{R_L}\frac{\mathrm{d}}{\mathrm{d}t}\left(\Phi_n-\Phi_{n+1}\right), \tag*{(2)}
dΦn+1dt=Vn−Vn+1+RCddt(Qn−Qn+1).(3)\frac{\mathrm{d}\Phi_{n+1}}{\mathrm{d}t}=V_n-V_{n+1}+R_C\frac{\mathrm{d}}{\mathrm{d}t}\left(Q_n-Q_{n+1}\right). \tag*{(3)}

To close the system (2) we should specify the connection between the voltages on the capacitors VnV_n and the charges and also between the currents through the inductors InI_n and the integrated voltages

Vn=V(Qn),(4)V_n=V(Q_n), \tag*{(4)}
In=I(Φn).(5)I_n=I(\Phi_n). \tag*{(5)}

In the continuum approximation we treat nn as the continuous variable xx (we measure distance in the units of the transmission line period) and approximate the finite differences in the r.h.s. of the equations by the first derivatives with respect to xx, after which the equations take the form

∂Q∂t=−∂I(Φ)∂x−1RL∂2Φ∂x∂t,(6)\frac{\partial Q}{\partial t}=-\frac{\partial I(\Phi)}{\partial x}-\frac{1}{R_L}\frac{\partial^2\Phi}{\partial x\partial t}, \tag*{(6)}
∂Φ∂t=−∂V(Q)∂x−RC∂2Q∂x∂t.(7)\frac{\partial\Phi}{\partial t}=-\frac{\partial V(Q)}{\partial x}-R_C\frac{\partial^2Q}{\partial x\partial t}. \tag*{(7)}

Further on we’ll consider the travelling waves, for which all the dependent variables depend upon the single independent variable τ=t−x/U\tau=t-x/U, where UU is the speed of the wave. For such waves (4) turns into the system of ODE which after integration takes the form

1RLdΦdτ=UQ−I(Φ)+I~,(8)\frac{1}{R_L}\frac{\mathrm{d}\Phi}{\mathrm{d}\tau}=UQ-I(\Phi)+\widetilde{I}, \tag*{(8)}
RCdQdτ=UΦ−V(Q)+V~,(9)R_C\frac{\mathrm{d}Q}{\mathrm{d}\tau}=U\Phi-V(Q)+\widetilde{V}, \tag*{(9)}

where I~\widetilde{I} and V~\widetilde{V} are the constants of integration. eq:5 contains 3 arbitrary constants - UU, I~\widetilde{I} and V~\widetilde{V}. Since we are considering localized travelling waves, it would be convenient to take as the constants Q1Q_1, Q2Q_2, Φ1\Phi_1, Φ2\Phi_2, defined by the equations

lim⁡τ→−∞Q(τ)=Q1;lim⁡τ→+∞Q(τ)=Q2;(10)\lim_{\tau\to-\infty}Q(\tau)=Q_1;\qquad\lim_{\tau\to+\infty}Q(\tau)=Q_2; \tag*{(10)}
lim⁡τ→−∞Φ(τ)=Φ1;lim⁡τ→+∞Φ(τ)=Φ2,(11)\lim_{\tau\to-\infty}\Phi(\tau)=\Phi_1;\qquad\lim_{\tau\to+\infty}\Phi(\tau)=\Phi_2, \tag*{(11)}

and connected by the relation

Q2−Q1Φ2−Φ1=I2−I1V2−V1.(12)\frac{Q_2-Q_1}{\Phi_2-\Phi_1}=\frac{I_2-I_1}{V_2-V_1}. \tag*{(12)}

Thus eq:5 can be presented as

1RLdΦdτ=U(Q−Q2)−I(Φ)+I2,(13)\frac{1}{R_L}\frac{\mathrm{d}\Phi}{\mathrm{d}\tau}=U(Q-Q_2)-I(\Phi)+I_2, \tag*{(13)}
RCdQdτ=U(Φ−Φ2)−V(Q)+V2,(14)R_C\frac{\mathrm{d}Q}{\mathrm{d}\tau}=U(\Phi-\Phi_2)-V(Q)+V_2, \tag*{(14)}

where I2=I(Φ2)I_2=I(\Phi_2) and V2=V(Q2)V_2=V(Q_2). The speed of the travelling wave is

U=I2−I1Q2−Q1=V2−V1Φ2−Φ1.(15)U=\frac{I_2-I_1}{Q_2-Q_1}=\frac{V_2-V_1}{\Phi_2-\Phi_1}. \tag*{(15)}

Further on we consider transmission line with the linear inductors (Φ=LI\Phi=LI). and nonlinear capacitors. We can exclude the current from (8) and after a bit of algebra obtain closed equation for Q(τ)Q(\tau)

LRCRLd2Qdτ2+RCdQdτ+LRLdV(Q)dτ=U2L(Q−Q2)−V(Q)+V2,(16)\frac{LR_C}{R_L}\frac{\mathrm{d}^2Q}{\mathrm{d}\tau^2}+R_C\frac{\mathrm{d}Q}{\mathrm{d}\tau}+\frac{L}{R_L}\frac{\mathrm{d}V(Q)}{\mathrm{d}\tau} =U^2L(Q-Q_2)-V(Q)+V_2, \tag*{(16)}

and the travelling speed is given by the equation

U2=1LV2−V1Q2−Q1.(17)U^2=\frac{1}{L}\frac{V_2-V_1}{Q_2-Q_1}. \tag*{(17)}

The solution of (10) should satisfy the boundary conditions (6a). Let us approximate V(Q)V(Q) for QQ between Q1Q_1 and Q2Q_2 as

V(Q)=const+Q+βQ2Cs,(18)V(Q)=\mathrm{const}+\frac{Q+\beta Q^2}{C_s}, \tag*{(18)}

where β\beta and CsC_s are constants. Let us also take V(Q)V(Q) in the last term in the l.h.s. of (10) in the linear approximation. In this cases, (10) becomes

T2d2Qdτ2+τRdQdτ=F(Q),(19)T^2\frac{\mathrm{d}^2Q}{\mathrm{d}\tau^2}+\tau_R\frac{\mathrm{d}Q}{\mathrm{d}\tau}=F(Q), \tag*{(19)}

where

T2≡LCsRCRL,(20)T^2\equiv\frac{LC_sR_C}{R_L}, \tag*{(20)}
τR=RCCs+LRL.(21)\tau_R=R_CC_s+\frac{L}{R_L}. \tag*{(21)}

and

F(Q)=β(Q−Q1)(Q−Q2),(22)F(Q)=\beta(Q-Q_1)(Q-Q_2), \tag*{(22)}

We can also improve the approximation (12) by modifying it to

V(Q)=const+Q+βQ2+γQ3Cs,(23)V(Q)=\mathrm{const}+\frac{Q+\beta Q^2+\gamma Q^3}{C_s}, \tag*{(23)}

where γ\gamma is also assumed to be constant. This modification can be important when Q1,Q2Q_1,Q_2 are close to zero and, as it normally happens, the charge is an odd function of the voltage. In this cases (13) is still valid but (15) becomes

F(Q)=γ(Q−Q1)(Q−Q2)(Q+Q3),(24)F(Q)=\gamma(Q-Q_1)(Q-Q_2)(Q+Q_3), \tag*{(24)}

where

Q3≡βγ+Q1+Q2.(25)Q_3\equiv\frac{\beta}{\gamma}+Q_1+Q_2. \tag*{(25)}

Strong dissipation

The ODE which doesn’t contain explicitly the independent variable

Consider the generalized damped Helmholtz-Duffing equation1

xττ+kxτ=γx(xn−x1)(xn+x3)(26)x_{\tau\tau}+kx_\tau=\gamma x\left(x^n-x_1\right)\left(x^n+x_3\right) \tag*{(26)}

where nn, kk, γ\gamma, x1x_1, x3x_3 are constants, with the boundary conditions

lim⁡τ→−∞x(τ)=x11/n,lim⁡τ→+∞x(τ)=0.(27)\lim_{\tau\to-\infty}x(\tau)=x_1^{1/n},\qquad\lim_{\tau\to+\infty}x(\tau)=0. \tag*{(27)}

Equation (19) doesn’t contain explicitly the independent variable τ\tau. This prompts the idea to consider xx as the new independent variable and

p=dxdτ(28)p=\frac{dx}{d\tau} \tag*{(28)}

as the new dependent variable. In the new variables (19) takes the form of Abel equation of the second kind2.

ppx+kp=γx(xn−x1)(xn+x3).(29)pp_x+kp=\gamma x\left(x^n-x_1\right)\left(x^n+x_3\right). \tag*{(29)}

The boundary conditions in the new variables are

p(x11/n)=p(0)=0.(30)p\left(x_1^{1/n}\right)=p(0)=0. \tag*{(30)}

One can easily check up that for γ\gamma and kk connected by the formula

k2=γ[x1+(n+1)x3]2n+1,(31)k^2=\frac{\gamma\left[x_1+(n+1)x_3\right]^2}{n+1}, \tag*{(31)}

the solution of (22) satisfying the boundary conditions (23) is

p=mx(xn−x1),m=γn+1.(32)p=mx\left(x^n-x_1\right),\qquad m=\sqrt{\frac{\gamma}{n+1}}. \tag*{(32)}

Substituting p(x)p(x) into (21) and integrating we obtain the solution of (19) as

x(τ)=x11/n[exp⁡(nmx1τ)+1]1/n.(33)x(\tau)=\frac{x_1^{1/n}}{\left[\exp\left(nmx_1\tau\right)+1\right]^{1/n}}. \tag*{(33)}

Now consider the equation

xττ+kxτ=βx(xn−x1)(34)x_{\tau\tau}+kx_\tau=\beta x\left(x^n-x_1\right) \tag*{(34)}

with the boundary conditions (20). We can present (27) as

xττ+kxτ=βx(xn/2−x11/2)(xn/2+x11/2).(35)x_{\tau\tau}+kx_\tau=\beta x\left(x^{n/2}-x_1^{1/2}\right)\left(x^{n/2}+x_1^{1/2}\right). \tag*{(35)}

Hence we realize that for kk and β\beta connected by the formula

k=(n+4)βx12(n+2),(36)k=(n+4)\sqrt{\frac{\beta x_1}{2(n+2)}}, \tag*{(36)}

the solution of (27) satisfying the boundary conditions (23) is

x(τ)=x11/n{exp⁡[nβx12(n+2)τ]+1}2/n.(37)x(\tau)=\frac{x_1^{1/n}}{\left\{\exp\left[n\sqrt{\frac{\beta x_1}{2(n+2)}}\tau\right]+1\right\}^{2/n}}. \tag*{(37)}

Let us return to (19) and modify it to

xττ+k(1+2βxn)xτ=γx(xn−x1)(xn+x3).(38)x_{\tau\tau}+k\left(1+2\beta x^n\right)x_\tau=\gamma x\left(x^n-x_1\right)\left(x^n+x_3\right). \tag*{(38)}

Thus we take into account possible nonlinearity of friction. Thus instead of (22) we obtain

ppx+k(1+2βxn)p=γx(xn−x1)(xn+x3).(39)pp_x+k\left(1+2\beta x^n\right)p=\gamma x\left(x^n-x_1\right)\left(x^n+x_3\right). \tag*{(39)}

Acting as above we obtain that for kk and γ\gamma connected by the formula

k2=γ[x1+(n+1)x3]2(1−2βx3)(n+1+2βx1),(40)k^2=\frac{\gamma\left[x_1+(n+1)x_3\right]^2}{\left(1-2\beta x_3\right)\left(n+1+2\beta x_1\right)}, \tag*{(40)}

the solution of (32) satisfying the boundary conditions (23) is (25) (same as it was for β=0\beta=0), only this time

m=γ(1−2βx3)n+1+2βx1.(41)m=\sqrt{\frac{\gamma\left(1-2\beta x_3\right)}{n+1+2\beta x_1}}. \tag*{(41)}

Hence the solution of (31) with β≠0\beta\ne0 is of the same form as for β=0\beta=0 ((33)), only with the modified mm.

Back to the transmission line

Now let us return to the transmission line. For (19) with F(Q)F(Q) given by (17), using the results of the previous Section, we claim that when the parameters of the equation satisfy the relation

τR2T2=γ(Q1+Q2+2Q3)22,(42)\frac{\tau_R^2}{T^2}=\frac{\gamma\left(Q_1+Q_2+2Q_3\right)^2}{2}, \tag*{(42)}

the solution of the equation can be expressed in terms of elementary functions:

Q=Q2+Q1−Q2exp⁡(ψτ)+1,(43)Q=Q_2+\frac{Q_1-Q_2}{\exp\left(\psi\tau\right)+1}, \tag*{(43)}

where

ψ=γ2Q1−Q2T.(44)\psi=\sqrt{\frac{\gamma}{2}}\frac{Q_1-Q_2}{T}. \tag*{(44)}

For (13) with F(Q)F(Q) given by (15), we claim that when the boundary conditions satisfy the relation

τRT=5β(Q1−Q2)6,(45)\frac{\tau_R}{T}=5\sqrt{\frac{\beta\left(Q_1-Q_2\right)}{6}}, \tag*{(45)}

the solution of the equation can be expressed in terms of elementary functions:

Q=Q2+Q1−Q2[exp⁡(χτ)+1]2.(46)Q=Q_2+\frac{Q_1-Q_2}{\left[\exp\left(\chi\tau\right)+1\right]^2}. \tag*{(46)}

where

χ=β(Q1−Q2)6⋅1T(47)\chi= \sqrt{\frac{\beta(Q_1-Q_2)}{6}}\cdot\frac{1}{T} \tag*{(47)}

We used previously the linear approximation for V(Q)V(Q) in the last term in the l.h.s. of Eq. (16). Strictly speaking, since we are considering nonlinear V(Q)V(Q) in the r.h.s. of Eq. (16) more consistent would be to treat the same way the l.h.s. Equations. (33), (34) allow us to go one step in this direction, that is to consider V(Q)V(Q) in the last term in the l.h.s. of Eq. (16) in quadratic approximation (for cubic nonlinearity of V(Q)V(Q)). As the result, Eqs. (35) and (36) slightly change and Eq. (43) doesn’t change at all.

The analytic results for the profile of the shock wave were obtained by reducing the second order differential equation (10) to the first order one

T2pdpdQ+TRp=F(Q)(48)T^2p\frac{dp}{dQ}+T_Rp=F(Q) \tag*{(48)}

where p=dQ/dτp=dQ/d\tau. However, looking back to Eq. (8) we see the opportunity to reduce the system of two first order differential equations to the single first order one

1RLRCdΦdQ=U(Q−Q2)−I(Φ)+I2U(Φ−Φ2)−V(Q)+V2(49)\frac{1}{R_LR_C}\frac{d\Phi}{dQ}=\frac{U(Q-Q_2)-I(\Phi)+I_2}{U(\Phi-\Phi_2)-V(Q)+V_2} \tag*{(49)}

already at this stage. It would be interesting to try to obtain exact analytic solutions of Eq. (49) for some values of the parameters.

Postponing such attempt until later time, note here that Eq. (49) allows us to obtain the τ\tau dependence of the current II which was absent in our previous publications. To achieve this aim let us rewrite Eq. (49) in the form

I=I2+U(Q−Q2)−LRLRC[UL(I−I2)−V(Q)+V2]dIdQ.(50)I=I_2+U(Q-Q_2)-\frac{L}{R_LR_C}\left[UL(I-I_2)-V(Q)+V_2\right]\frac{dI}{dQ}. \tag*{(50)}

The dependence Q(τ)Q(\tau) being found earlier, Eq. (50) gives infinite series for I(τ)I(\tau), and the first two terms of the series are

I=I2+U(Q−Q2)−LCsRLRCI2−I1Q2−Q1F(Q).(51)I=I_2+U(Q-Q_2)-\frac{L}{C_sR_LR_C}\frac{I_2-I_1}{Q_2-Q_1}F(Q). \tag*{(51)}

Thus from (36) follows

I=I2+I1−I2exp⁡(ψτ)+1{1+L(Q1−Q2)CsRLRCexp⁡(ψτ)exp⁡(ψτ)+1[Q1−Q2exp⁡(ψτ)+1+βγ+Q1+2Q2]}.(52)I=I_2+\frac{I_1-I_2}{\exp(\psi\tau)+1}\left\{1+\frac{L(Q_1-Q_2)}{C_sR_LR_C}\frac{\exp(\psi\tau)}{\exp(\psi\tau)+1}\left[\frac{Q_1-Q_2}{\exp(\psi\tau)+1}+\frac{\beta}{\gamma}+Q_1+2Q_2\right]\right\}. \tag*{(52)}

and from (39) follows

I=I2+I1−I2[exp⁡(χτ)+1]2{1+L(Q1−Q2)CsRLRC[1−1[exp⁡(χτ)+1]2]}.(53)I=I_2+\frac{I_1-I_2}{[\exp(\chi\tau)+1]^2}\left\{1+\frac{L(Q_1-Q_2)}{C_sR_LR_C}\left[1-\frac{1}{[\exp(\chi\tau)+1]^2}\right]\right\}. \tag*{(53)}

Weak dissipation

Let us return to Eq. (19) and consider the case of weak dissipation. In this case it is convenient to rewrite the equation as

d2Qdτ2+dΠ(Q)dQ=−κdQdτ,(54)\frac{d^2Q}{d\tau^2}+\frac{d\Pi(Q)}{dQ}=-\kappa\frac{dQ}{d\tau}, \tag*{(54)}

where τ~≡τ/T\tilde{\tau}\equiv\tau/T, κ≡τR/T\kappa\equiv\tau_R/T, and Π(Q)\Pi(Q) is defined by the equation

dΠ(Q)dQ≡F(Q).(55)\frac{d\Pi(Q)}{dQ}\equiv F(Q). \tag*{(55)}

We will use the method of time averaging[22]. We assume that we know the undamped solution Qud(τ~;E)Q_{ud}(\tilde{\tau};\mathcal{E}), satisfying equation

12(dQuddτ~)2+Π(Qud)=E,(56)\frac{1}{2}\left(\frac{dQ_{ud}}{d\tilde{\tau}}\right)^2+\Pi(Q_{ud})=\mathcal{E}, \tag*{(56)}

and express damped oscillations as

Q(τ~)=Qud(τ~;E(τ~)),(57)Q(\tilde{\tau})=Q_{ud}(\tilde{\tau};\mathcal{E}(\tilde{\tau})), \tag*{(57)}

where E(τ~)\mathcal{E}(\tilde{\tau}) satisfies equation

dEdτ~=−κ⟨(dQuddτ~)2⟩E;(58)\frac{d\mathcal{E}}{d\tilde{\tau}}=-\kappa\left\langle\left(\frac{dQ_{ud}}{d\tilde{\tau}}\right)^2\right\rangle_{\mathcal{E}}; \tag*{(58)}

the averaging is with respect to the period of the undamped oscillation with the energy E\mathcal{E}, or more explicitly,

dEdτ~=−2κ∫dQE−Π(Q)∫dQ/E−Π(Q);(59)\frac{d\mathcal{E}}{d\tilde{\tau}}=-\frac{2\kappa\int dQ\sqrt{\mathcal{E}-\Pi(Q)}}{\int dQ/\sqrt{\mathcal{E}-\Pi(Q)}}; \tag*{(59)}

the limits of integration in both integrals are found from the equation E−Π(Q)=0\mathcal{E}-\Pi(Q)=0. Notice that the shock in the considered case can be called the stationary dispersive shock wave[23].

We want to show how the method of time averaging works for F(Q)F(Q) given by Eq. (22). After the change of variable

Q~=β6(Q−Q1+Q22),(60)\widetilde{Q}=\frac{\beta}{6}\left(Q-\frac{Q_1+Q_2}{2}\right), \tag*{(60)}

Eq. (56) takes the form

(dQ~uddτ~)2=4Q~ud3−g2Q~ud−g3,(61)\left(\frac{d\widetilde{Q}_{ud}}{d\tilde{\tau}}\right)^2=4\widetilde{Q}_{ud}^{3}-g_2\widetilde{Q}_{ud}-g_3, \tag*{(61)}

where

g2=β2(Q1−Q2)212(62)g_2=\frac{\beta^2(Q_1-Q_2)^2}{12} \tag*{(62)}
g3=−β218E.(63)g_3=-\frac{\beta^2}{18}\mathcal{E}. \tag*{(63)}

(61) defines Weierstrass elliptic function24^{24}.

Q~(τ~;E)=P(τ~;g2,g3).(64)\widetilde{Q}(\widetilde{\tau};\mathcal{E})=\mathcal{P}(\widetilde{\tau};g_2,g_3). \tag*{(64)}

Thus the solution of (47) is

Q(τ~)=P(τ~;g2,g3(τ~)).(65)Q(\widetilde{\tau})=\mathcal{P}(\widetilde{\tau};g_2,g_3(\widetilde{\tau})). \tag*{(65)}

(59) can be rewritten as

dg3dτ~=4κN(g3)Y0(g3),(66)\frac{d g_3}{d\widetilde{\tau}}=4\kappa\frac{\mathcal{N}(g_3)}{Y_0(g_3)}, \tag*{(66)}

where

Y0(g3)=∫cbdψP(ψ),N(g3)=∫cbdψP(ψ),(67)Y_0(g_3)=\int_c^b\frac{d\psi}{\sqrt{P(\psi)}},\qquad\mathcal{N}(g_3)=\int_c^b d\psi\sqrt{P(\psi)}, \tag*{(67)}
P(ψ)=4ψ3−g2ψ−g3;(68)P(\psi)=4\psi^3-g_2\psi-g_3; \tag*{(68)}

aa, bb, cc (a>b>ca>b>c) are the roots of cubic equation P(ψ)=0P(\psi)=0.

All integrals

Ym=∫ψmdψP(ψ),(69)Y_m=\int\frac{\psi^m d\psi}{\sqrt{P(\psi)}}, \tag*{(69)}

where mm is an arbitrary natural number and P(ψ)P(\psi) is some polynomial of power pp, are expressed through the p−1p-1 first integrals Y0,Y1,…,Yp−2Y_0,Y_1,\ldots,Y_{p-2} and algebraic quantities25^{25}. Y0Y_0 and Y1Y_1 are table integrals26^{26}:

Y0=2a−cK(k),(70)Y_0=\frac{2}{\sqrt{a-c}}K(k), \tag*{(70)}
Y1=2aa−cK(k)−2a−cE(k),(71)Y_1=\frac{2a}{\sqrt{a-c}}K(k)-2\sqrt{a-c}E(k), \tag*{(71)}

where KK and EE are complete elliptic integrals of the first and second kind respectively, and k=(b−c)/(a−c)k=\sqrt{(b-c)/(a-c)}. To calculate N(g3)\mathcal{N}(g_3) let us integrate the identity

P(ψ)=4ψ3−g2ψ−g3P(ψ).(72)\sqrt{P(\psi)}=\frac{4\psi^3-g_2\psi-g_3}{\sqrt{P(\psi)}}. \tag*{(72)}

We obtain

N=4Y3−g2Y1−g3Y0.(73)\mathcal{N}=4Y_3-g_2Y_1-g_3Y_0. \tag*{(73)}

Integrating the identity

ddψ[ψP(ψ)]=P(ψ)+12ψ3−g2ψ2P(ψ),(74)\frac{d}{d\psi}\left[\psi\sqrt{P(\psi)}\right]=\sqrt{P(\psi)}+\frac{12\psi^3-g_2\psi}{2\sqrt{P(\psi)}}, \tag*{(74)}

we obtain

2N+12Y3−g2Y1=0.(75)2\mathcal{N}+12Y_3-g_2Y_1=0. \tag*{(75)}

Combining (63) and (65) we obtain

N=−15(2g2Y1+3g3Y0)(76)\mathcal{N}=-\frac{1}{5}(2g_2Y_1+3g_3Y_0) \tag*{(76)}

and, finally,

dg3dτ~=4κ5[ag2+3g3−(a−c)g2E(k)K(k)].(77)\frac{d g_3}{d\widetilde{\tau}}=\frac{4\kappa}{5}\left[ag_2+3g_3-(a-c)g_2\frac{E(k)}{K(k)}\right]. \tag*{(77)}

(One should keep in mind that aa, cc, kk are functions of g3g_3.) Numerical analysis of the formula obtained in this Section one may find in our previous publication22^{22}.

The simple waves in a lossless nonlinear transmission line

As it is well known17^{17}, the shock waves exist only in the presence of dissipation. (In the absence of the dissipation, the localized travelling waves in the systems we consider are kinks and solitons20^{20}.) This corresponds to the case of a lossy nonlinear transmission line, considered in the previous part of the paper.

On the other hand, many properties of the shocks in fluids were studied beginning from the XIX century without taking the dissipation explicitly into account, but just postulating the existence of discontinuous solutions and the jump conditions27^{27}. Similarly to that, the initial phase of the formation of a shock wave can be considered ignoring the dissipation. The present section is dedicated to a lossless transmission line, constructed from the identical nonlinear inductors and the identical nonlinear capacitors as shown on Figure 2. We take the capacitors voltages VnV_n and the currents through the inductors InI_n as the dynamical variables. The circuit equations (Kirchhoff laws) are

Lossless nonlinear transmission line

Figure 2. Lossless nonlinear transmission line

C(Vn)dVndt=In−In+1,(78)C(V_n)\frac{dV_n}{dt}=I_n-I_{n+1}, \tag*{(78)}
L(In+1)dIn+1dt=Vn−Vn+1.(79)L(I_{n+1})\frac{dI_{n+1}}{dt}=V_n-V_{n+1}. \tag*{(79)}

Further on we’ll consider C(V)C(V) and L(I)L(I) as known functions.

In the continuum approximation we treat nn as the continuous variable xx (we measure distance in the units of the transmission line period) and approximate the finite differences in the r.h.s. of the equations by the first derivatives with respect to xx, after which the equations take the form

C(V)∂V∂t=−∂I∂x,(80)C(V)\frac{\partial V}{\partial t}=-\frac{\partial I}{\partial x}, \tag*{(80)}
L(I)∂I∂t=−∂V∂x.(81)L(I)\frac{\partial I}{\partial t}=-\frac{\partial V}{\partial x}. \tag*{(81)}

The simple wave approximation allows to decouple the wave equations into two separate equations for the right- and left-going waves17–19^{17\text{--}19}. In our previous publications we introduced such approximation for eq:69 for the case of half-nonlinear transmission line (more specifically for the case of JTL[20]. In the present Section we formulate the simple wave approximation for the general lossless nonlinear transmission line including both the nonlinear inductors and the nonlinear capacitors.

The simple wave approximation, well known in the theory of nonlinear waves, refers to a wave where the relationship between the wave’s properties (like displacement, pressure, or density) and its spatial and temporal coordinates can be described by a single, independent function. In essence, it’s a wave where the relationship between its amplitude and other characteristics is straightforward, and not influenced by other waves or complex interactions. More specifically, it allows to decouple the waves moving in the opposite directions (in a 1D case).

To formulate such approximation in our case, let us start from the small amplitude waves on a homogeneous background V0V_0, I0I_0.

V=V0+v,(82)V = V_0 + v, \tag*{(82)}
I=I0+i.(83)I = I_0 + i. \tag*{(83)}

For such waves Eq. (69) is simplified to

C(Q)dvdt=−didx,(84)C(Q)\frac{dv}{dt}=-\frac{di}{dx}, \tag*{(84)}
L(I)didt=−dvdx.(85)L(I)\frac{di}{dt}=-\frac{dv}{dx}. \tag*{(85)}

(for brevity we have discarded lower index 0 in Eq. (71)). The solutions of Eq. (71) are right- and left-propagating travelling waves, each depending upon the single variable τ±=t∓x/u0\tau_{\pm}=t\mp x/u_0, propagating with the speed

U(I,V)=1L(I)C(V).(86)U(I,V)=\frac{1}{\sqrt{L(I)C(V)}}. \tag*{(86)}

The voltage and current in the ”sound” wave are connected by the equation

v=±Z(I,V)i,(87)v=\pm Z(I,V)i, \tag*{(87)}

where

Z(I,V)=L(I)C(Q).(88)Z(I,V)=\sqrt{\frac{L(I)}{C(Q)}}. \tag*{(88)}

The simple wave approximation, that is decoupling of Eq. (69) into two separate equations for the right- and left-going waves, is achieve by considering VV as a function of II (or vice-verse). We emphasize that our approximation (see below) is based on previously known approach but the application of the approximation to the fully nonlinear transmission line was (to the best of our knowledge) was never done before.

Then from Eq. (69) we obtain

C(V)L(I)dVdI=dIdV,(89)\frac{C(V)}{L(I)}\frac{dV}{dI}=\frac{dI}{dV}, \tag*{(89)}

or equivalently

dVdI=±Z(I,V).(90)\frac{dV}{dI}=\pm Z(I,V). \tag*{(90)}

Substituting into (69) we obtain a system of two coupled equations for each of the simple waves

∂I∂t=±U(I,V)∂I∂x,(91)\frac{\partial I}{\partial t}=\pm U(I,V)\frac{\partial I}{\partial x}, \tag*{(91)}
∂V∂t=±U(I,V)∂V∂x.(92)\frac{\partial V}{\partial t}=\pm U(I,V)\frac{\partial V}{\partial x}. \tag*{(92)}

Further on for the sake of definiteness we’ll talk only about the right-going wave which corresponds to taking the sign minus in the r.h.s. of both equations in (77). The system (77) simplifies in half-nonlinear cases, that is when either the capacitor or the inductor is linear. In the first case (C(V)=C(V)=const)

U=U(I),(93)U=U(I), \tag*{(93)}

and (77a) becomes closed equation for the current

∂I∂t+U(I)∂I∂x=0.(94)\frac{\partial I}{\partial t}+U(I)\frac{\partial I}{\partial x}=0. \tag*{(94)}

Instead of (77b) we can use equation

U=Z(I)I.(95)U=Z(I)I. \tag*{(95)}

Talking about this case we have in mind first and foremost the Josephson transmission line. Both Josephson laws can be presented as

L(I)=ℏ2eIc2−I2,(96)L(I)=\frac{\hbar}{2e\sqrt{I_c^2-I^2}}, \tag*{(96)}

thus we obtain

U(I)=2eℏCIc2−I24.(97)U(I)=\sqrt{\frac{2e}{\hbar C}}\sqrt[4]{I_c^2-I^2}. \tag*{(97)}

However, the fully nonlinear case can also be treated easily. If we consider the initial value problem

I(x,0)=I0(x),V(x,0)=V0(x),(98)I(x,0)=I_0(x),\qquad V(x,0)=V_0(x), \tag*{(98)}

the solution of Eq. (77) can be obtained by inspection[28]

I(x,t)=I0(ξ)(99)I(x,t)=I_0(\xi) \tag*{(99)}
V(x,t)=V0(ξ),(100)V(x,t)=V_0(\xi), \tag*{(100)}

where

x−ξ=U(V0(ξ),I0(ξ))t.(101)x-\xi=U(V_0(\xi),I_0(\xi))t. \tag*{(101)}

The simple wave approximation (77) can be applied also to Eq. (4). In fact, the dissipative terms in the equation determine the profile of the shock. On the other hand, if we want to study the formation of the shocks, then assuming these terms to be in some sense small, we can ignore the influence of the dissipation on the process of the formation (until we don’t approach to close to the singularity of the dissipationless equation). Thus ignoring the dissipation in (4) we may rewrite the equation in the form

dQdV∂V∂t=−∂I∂x,(102)\frac{dQ}{dV}\frac{\partial V}{\partial t}=-\frac{\partial I}{\partial x}, \tag*{(102)}
dΦdI∂I∂t=−∂V∂x,(103)\frac{d\Phi}{dI}\frac{\partial I}{\partial t}=-\frac{\partial V}{\partial x}, \tag*{(103)}

which coincides with eq:69 if we put

C(V)=dQdV,L(I)=dΦdI.(104)C(V)=\frac{dQ}{dV},\qquad L(I)=\frac{d\Phi}{dI}. \tag*{(104)}

After that we can apply the procedure presented above in this Section.

Now let us forget about the dissipation and consider the strictly dissipationless case. If we want to study the profile of the travelling wave in such case, additional complication (with respect to the lossy case) arises. In the latter case we started from considering the discrete transmission line, but the presence of the dissipation introduced the space scale into the system, and this scale was implicitly assumed to be much larger than the period of the line. This allowed us to use the continuum approximation, actually ignoring the discrete nature of the system. For the lossless case the scale of the localized travelling wave is determined by the period of the transmission line[20]. This makes the continuum approximation inadequate and we introduced the quasi-continuum approximation[20], which corresponds to approximating the finite differences in the r.h.s. of the equations (68) by the two first terms in the Taylor expansion[20]. Thus instead of eq:69 we obtain

C(Q)∂V∂t=−∂I∂x−124∂3I∂x3,(105)C(Q)\frac{\partial V}{\partial t}=-\frac{\partial I}{\partial x}-\frac{1}{24}\frac{\partial^3 I}{\partial x^3}, \tag*{(105)}
L(I)∂I∂t=−∂V∂x−124∂3V∂x3.(106)L(I)\frac{\partial I}{\partial t}=-\frac{\partial V}{\partial x}-\frac{1}{24}\frac{\partial^3 V}{\partial x^3}. \tag*{(106)}

We studied in details the travelling waves described by these equations for the case of half-nonlinear transmission line (more specifically for the case of JTL[20]). In distinction to the lossy case, where the travelling waves turn out to be the shocks, in the lossless case the travelling waves turn out to be the kinks and the solitons.

We also studied (with much less details) the formation of the solitons and the kinks via introducing the simple wave approximation for the JTL. Now we want to formulate the simple wave approximation for the general lossless nonlinear transmission line including both the nonlinear inductors and the nonlinear capacitors on top of the quasi-linear approximation.

One must understand that eq:69 (and hence eq:77 can describe the formation of the kinks and the solitons (until we don’t approach to close to the singularities of the equations). Our present aim is to formulate the approximation which will describe both the formation of the kinks and the solitons and their profiles.

Starting from eq:88 and repeating the process which led from (69) to (77) we obtain instead of the latter

∂I∂t=±U(I,V)(∂I∂x+124∂3I∂x3),(107)\frac{\partial I}{\partial t}=\pm U(I,V)\left(\frac{\partial I}{\partial x}+\frac{1}{24}\frac{\partial^3 I}{\partial x^3}\right), \tag*{(107)}
∂V∂t=±U(I,V)(∂V∂x+124∂3V∂x3).(108)\frac{\partial V}{\partial t}=\pm U(I,V)\left(\frac{\partial V}{\partial x}+\frac{1}{24}\frac{\partial^3 V}{\partial x^3}\right). \tag*{(108)}

Let us apply thus improved simple wave approximation to the JTL. In this case instead of (94) we obtain

∂I∂t+2eℏCIc2−I24(∂I∂x+124∂3I∂x3)=0.(109)\frac{\partial I}{\partial t}+\sqrt{\frac{2e}{\hbar C}}\sqrt[4]{I_c^2-I^2}\left(\frac{\partial I}{\partial x}+\frac{1}{24}\frac{\partial^3 I}{\partial x^3}\right)=0. \tag*{(109)}

If we make an additional assumption ∣I∣≪Ic|I|\ll I_c, (109) can be written down as[20]

∂I∂t+2eIcℏC(∂I∂x−112Ic2∂I3∂x+124∂3I∂x3)=0,(110)\frac{\partial I}{\partial t}+\sqrt{\frac{2eI_c}{\hbar C}}\left(\frac{\partial I}{\partial x}-\frac{1}{12I_c^2}\frac{\partial I^3}{\partial x}+\frac{1}{24}\frac{\partial^3 I}{\partial x^3}\right)=0, \tag*{(110)}

where we have ignored the term proportional to I2∂3I/∂x3I^2\partial^3 I/\partial x^3. Looking at (110) we recognize the modified Korteweg-de Vries (mKdV) equation[29].

On the other hand, considering small variations of the current on the constant background presented by (83), from (91) we obtain[20]

∂i∂t+2eℏCIc2−I024(∂i∂x−I04(Ic2−I02)∂i2∂x+124∂3i∂x3)=0,(111)\frac{\partial i}{\partial t}+\sqrt{\frac{2e}{\hbar C}}\sqrt[4]{I_c^2-I_0^2}\left(\frac{\partial i}{\partial x}-\frac{I_0}{4(I_c^2-I_0^2)}\frac{\partial i^2}{\partial x}+\frac{1}{24}\frac{\partial^3 i}{\partial x^3}\right)=0, \tag*{(111)}

where we have ignored the term proportional to i∂3i/∂x3i\partial^3 i/\partial x^3. Looking at (111) we recognize the Korteweg-de Vries (KdV) equation[29].

To conclude we state that we obtained the exact analytical expressions for the profile of the shock waves (both the current and the voltage) in half-nonlinear transmission lines for the appropriate values of the parameters. We also formulated the simple wave approximation for the lossless discrete nonlinear transmission line.

References

  1. [1]G. B. Whitham, Linear and Nonlinear Waves, John Wiley & Sons Inc., New York (1999).
  2. [2]E. Kenge, W. M. Liu, L. Q. English, and B. A. Malomed, Phys. Rep. 982, 1 (2022).
  3. [3]T. Congy, G. A. El, and M. A. Hoefer, J. Fluid Mech. 875, 1145 (2019).
  4. [4]D. M. French and B. W. Hoff, IEEE Trans. Plasma Sci. 42, 3387 (2014).
  5. [5]B. Nouri, M. S. Nakhla, and R. Achar, IEEE Trans. Microw. Theory Techn. 65, 673 (2017).
  6. [6]L. P. S. Neto, J. O. Rossi, J. J. Barroso, and E. Schamiloglu, IEEE Trans. Plasma Sci. 46, 3648 (2018).
  7. [7]M. S. Nikoo, S. M.-A. Hashemi, and F. Farzaneh, IEEE Trans. Microw. Theory Techn. 66, 3234 (2018); 66, 4757 (2018).
  8. [8]L. C. Silva, J. O. Rossi, E. G. L. Rangel, L. R. Raimundi, and E. Schamiloglu, Int. J. Adv. Eng. Res. Sci. 5, 121 (2018).
  9. [9]Y. Wang, L.-J. Lang, C. H. Lee, B. Zhang, and Y. D. Chong, Nat. Comm. 10, 1102 (2019).
  10. [10]E. G. L. Range, J. O. Rossi, J. J. Barroso, F. S. Yamasaki, and E. Schamiloglu, IEEE Trans. Plasma Sci. 47, 1000 (2019).
  11. [11]A. S. Kyrygeyan, Semiconductors 53, 511 (2019).
  12. [12]N. A. Akem, A. M. Dikande, and B. Z. Essimbi, SN Applied Science 2, 21 (2020).
  13. [13]A. J. Fairbanks, A. M. Darr, A. L. Garner, IEEE Access 8, 148606 (2020).
  14. [14]R. Landauer, IBM J. Res. Dev. 4, 391 (1960).
  15. [15]S. T. Peng and R. Landauer, IBM J. Res. Dev. 17, 299 (1973).
  16. [16]E. Kogan, Phys. Stat. Sol. (b), 262, 2400335 (2025).
  17. [17]L. D. Landau and E. M. Lifshitz, Fluid Mechanics: Landau and Lifshitz: Course of Theoretical Physics, Volume 6 (Vol. 6), Elsevier (2013).
  18. [18]M. I. Rabinovich and D. I. Trubetskov, Oscillations and Waves, Kluwer Academic Publishers, Dordrecht / Boston / London (1989).DOI
  19. [19]M. B. Vinogradova, O. V. Rudenko and A. P. Sukhоrukov, The Wave Theory, Nauka Publishers, Moscow (1990).
  20. [20]E. Kogan, Phys. Stat. Sol. (b) 259, 22000160 (2022).
  21. [22]E. Kogan, Journal of Applied Physics 130, 013903 (2021).
  22. [23]A. M. Kamchatnov, Theory of Nonlinear Waaves [in Russian], Printing House of School of Economics, 2024.
  23. [28]J. D. Logan, An introduction to nonlinear partial differential equations (Second Edition), A John Wiley & Sons, Inc. Publication, Hoboken, New Jersey, 2008.DOI
  24. [29]P. G. Drazin and R. S. Johnson, Solitons: an introduction. Vol. 2., Cambridge university press, 1989.

Paper details

vlnc:W_ww3w3gtk
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents