You are on page 1of 7

Supporting Online Material

Observation of the Inverse Doppler Effect


N. Seddon and T. Bearpark

Materials and Methods.

Transmission line and pump pulse.

A transmission line with 118 LC sections was used in the experiment. The transmission

line was an LC ladder network to which a series of cross-link capacitors C* were added

and nonlinear material included in the inductors. The LC ladder network and the cross-

link capacitors defined the dispersion characteristics of the line and a saturable magnetic

material (soft ferrite) provided the nonlinear characteristics. Component values used in

this experiment were C(n)=2.83pF, C*=3.18pF and saturated value of L=5.5nH. The

550mm long transmission line was arranged in a linear configuration and was housed in a

600mm long aluminium trough. The transmission line electrical length (118 sections) was

sufficient to ensure that reflections from the output end of the line were transit time

isolated while experimental data was recorded close to the input of the line. Commercial

electrostatic field plotting software (Coulomb©) was used as a design aid to ensure that

the circuit parameters for the experimental hardware were close to the design

requirement.

A pump pulse, with a 4ns rise time and a 20ns long flat top, was generated by a Blumlein

pulse forming circuit and a thyratron switch. The rise time of this pulse was too long to

be used directly and the pulse was passed through a sharpening circuit (S1) to reduce the

pulse rise time to 1ns while maintaining a 20ns flat top. The 1ns rise time, 20ns long
pump pulse was injected into the input of the transmission line via a 3m long 50W coaxial

cable.

Formation of the moving boundary and propagation of the RF wave.

On injection of the pump pulse into the line two competing processes occur; the nonlinear

magnetic material is rapidly driven from an unsaturated to a saturated state which forms

an electromagnetic shock discontinuity at the leading edge of the pump pulse and energy

is dispersed from the shock front. Nonlinearity tends to shorten the pump pulse rise time,

dispersion tends to lengthen the rise time and the steady state rise time for the

discontinuity is reached when dispersive and nonlinear effects are balanced. We measure

Doppler signals close to the input of the transmission line and require that the

propagating shock discontinuity has reached a steady state close to the input of the line.

This is achieved by injecting a pump pulse with a rise time that is close to the steady state

value i.e. 1ns.

Fig. S1 illustrates the formation and propagation of waves in this transmission line. The

convention used in Fig. S1 and the following text is that the pump pulse propagates in the

positive x direction. The RF wave that is produced by the propagating discontinuity,

wave (a), has a fixed phase relationship with the discontinuity, a pictorial analogy is the

bow wave that is formed at the front of a ship. In this experiment the RF wave maintains

a phase peak immediately behind the propagating discontinuity, Fig. S1, and the phase

velocity, np, is equal to the velocity of the discontinuity, ns.


Fig. S1. (A) Pump pulse and RF wave propagation in a transmission line with normal
dispersion. A pump pulse is injected at input interface. The pump pulse propagates from
left to right and produces a wave (blue trace) that propagates from left to right. (B) Pump
pulse and RF wave propagation in a transmission line with anomalous dispersion. The
pump pulse produces a wave (blue trace) that propagates from right to left, and is referred
to as wave (a) in the text. Wave (a) reflects at the input terminals to produce wave (b)
(red trace) with the same oscillation frequency but reversed directions for np and ng.

In Fig S1 panel A, the pump pulse excites an RF wave in a medium with normal

dispersion, (npng>0), and the phase velocity is greater than the group velocity (np>ng).

The discontinuity velocity, phase velocity and the group velocity in the positive x

direction so wave (a) streams out behind the discontinuity and propagates in the positive

x direction. At time t1 three RF cycles have been produced and at time t2>t1 the pump

pulse has propagated further into the line and seven RF cycles have been produced. In

Fig. S1 panel B, wave (a) is produced in a region of anomalous dispersion (npng<0). The

discontinuity velocity and phase velocity are in the positive x direction and the group

velocity is in the negative x direction, so wave (a) streams out from the discontinuity and

propagates towards the input terminal. Wave (a) is produced immediately on injection of

the pump pulse into the line and continues to be produced as the pump pulse propagates
further into the line, so the expanding cavity between the discontinuity and the input

terminal is always filled with wave (a). Wave (a) is continually reflected from the input

terminals to produce wave (b) which has group velocity in the positive x direction and

phase velocity in the negative x direction. If the magnitude of the group velocity for wave

(b) is greater than the velocity of the discontinuity then wave (b) is incident on a receding

boundary and an inverse Doppler shift is expected, (main text, Fig. 2B). The Doppler

shifted wave is referred to as wave (c).

Wave Reflections and Doppler Shift Calculation.

Figure S2 shows the relationships of waves (a), (b) and (c) on an extended dispersion

diagram.

Fig. S2. Extended dispersion diagram showing the relationship of waves (a), (b) and (c).
The wave that is produced by dispersion of energy from the moving discontinuity, wave

(a), is generated at point a. The phase velocity for wave (a), np= w/y, is positive and the

group velocity, ng= dw/dy, is negative. Reflection of wave (a) at the input terminal

produces wave (b), which has the same frequency, and the same magnitudes for np and

ng. However the directions of np and ng are reversed by the reflection and the reflected

wave is represented by point b on the dispersion curve. Wave (b) is reflected from the

moving shock discontinuity according to the non-relativistic Doppler equation, which can

be represented on the dispersion curve as a straight line with a slope of ns/2. The

intersection of the Doppler shift line and the dispersion curve define the allowed Doppler

shifts and the characteristics of the shifted wave. In Fig. S2 the wave (b) is Doppler up-

shifted to produce a wave at point c on the dispersion curve.

For each experimental condition we measure ns and the frequency of wave (b) and use a

graphical solution, as shown in Fig S2, to predict the frequency of any Doppler shifted

signal. It is possible that the Doppler shifted wave can undergo further reflections

between the input terminal and the shock, in which case additional Doppler signals may

be produced. The signals that are produced in practice are determined by the requirement

that the dispersion characteristic and the Doppler shift line provide a solution to carry

energy away from the boundary.

Experimental measurement and data processing

In the experiment the frequency of RF signals, w, and the normalised velocity of the

shock discontinuity, ns, were calculated from data produced by two sensors that measured
the differential of electric field at positions close to the input of the transmission line.

Raw experimental data was displayed and recorded using a 3GHz bandwidth

(Tektronix© 694) oscilloscope. The RF signal was monitored at the 10th LC section from

the input of the transmission line and the commercial data processing software,

FAMOS©, was used to produce Fast Fourier Transforms (FFTs) of the data.

Normalised shock propagation velocity, ns, was calculated from measurements of the

shock transit time between section 10 and section 29 of the line. The shock transit time

for each experimental condition was compared with the transit time measured when the

nonlinear components were pre-saturated by 1ms long, 10A current pulse. The transit

time for the fully saturated line corresponds to propagation at the cut off velocity for the

system, no, and the transit time measured in the experiment corresponds to the shock

velocity, ns.

Equivalent conditions were obtained between the experiment and the numerical model by

setting the initial condition of the nonlinear inductor in the model so that the modelled

shock velocities were equal to the experimental shock velocities. While it is possible to

calibrate the experimental system, by measuring the relationship between shock velocity

and bias current, we do not include a DC bias current supply in the numerical model. This

was considered to add an unnecessary level of complexity to the model and was avoided

by simply setting the initial condition of the magnetic material as a parameter in the

numerical model.
References

S1. I. G. Kataev, Electromagnetic Shock Waves. (Illife, London 1963).

You might also like