CONTENTS

Page

(1) ABSTRACT 2

(2) INTRODUCTION AND MOTIVATION 3

Kicker Magnets 3
Transmission Lines 6

(3) THEORY 6
Transmission Line as Pulsed Power Modulator 6
Transmission Line Equations 8
High Frequency in Coaxial Cables with Losses 10
Losses in Pulses through Transmission Line 11
Attenuation of Waveform due to Skin Effect 13

(4) MEASUREMENT 14
Experimental Apparatus 14
Measurement Method 15

(5) RESULTS AND DISCUSSIONS 19

(6) CONCLUSION 21

(7) ACKNOWLEDGEMENTS 22

(8) REFERENCES 23

1
 Measuring Coaxial Cable Impedance using Oscilloscope to
Analyze Waveform

Tolani E. Owusu

University of North Carolina At Charlotte

Supervisor: Chris C. Jensen

Fermi National Accelerator Laboratory*

P. O. Box 500, Batavia, IL 60510

July 17, 2000

ABSTRACT

The increasing use of electrical pulses in the transmission of power and data by cable has

resulted in a need for a better understanding of the electrical characteristics of a cable. Today,

the study of transmission line cable is no longer limited to idealized lengths of lines present but

also their properties as a function of materials and geometry. The finite speed of transmission

pulses and attenuation through coaxial cables plays an important role in the fast injection systems

of high-energy accelerators in the physics research programs at Fermilab. As a result, the

distortion of arbitrary pulses through coaxial transmission line becomes a problem, especially

during power transmission from the storage line pulse generators to the fast kicker magnet that

deflects the particle beam. To better understand the cable properties, the pulse response of

coaxial cable line was measured using an oscilloscope. Then analysis of different waveforms

was undertaken.

*
Operated by the Universities Research Association under contract with the U.S. Department of
Energy.
2
 First, attenuation and kick strength rise time for transmission line kicker is described.

Next, the theory of conductor and dielectric losses in high voltage pulse cable is reviewed.

Finally, the decrease in amplitude and deformation of waveform due to skin effect is examined.

The analysis of waveform shows that pulses traveling in lossless line pulse in traverses-

electromagnetic mode (TEM) without attenuation and distortions are only mathematical

abstractions. However, all physically realizable cables suffer from attenuation and distortion of

pulses, with skin effect in the conductors being the predominate losses.

I.INTRODUCTION AND MOTIVATION

In the fast injection system of high-energy accelerators, the transmission of pulses

through coaxial cables is vital. Specifications of kicker magnets for particle accelerators

typically require fast rise times and good field uniformity. Kicker magnets are used to inject and

extract beams. As a case in point, 1MHz beam chopper is use to create gaps in the beam during

which the field in the kicker magnet must rise or fall from 1% to 99% in order to minimize beam

losses. The time interval of the gap created is approximately 50ns long.

Another example is an injection kicker for the Main Injector. It must have a rise time

(1% to 99%) of 50ns, a field uniformity of 1% over 1.2us and a fall time (99% to 1% of

150ns). In general, the required rise and fall time of each kicker magnet varies from accelerator

to accelerator. Some kicker magnets consist of LC cells with each cell sandwiched between high

voltage capacitance plates. The characteristic impedance for such kicker magnet is defined in

terms of the inductance and capacitance of each LC cells. The purpose of an extraction kicker

magnet is to fully deflect (kick) bunches of particles without disturbance to the preceding or

following bunches. On the other hand, an injection kicker does the inverse. For instance, it

places a particle beam from septum magnet onto an equilibrium trajectory. Again, kicker
3
magnets are fast pulsed magnets and are pulsed from a high voltage pulse generator. All of the

injection and extraction kicker magnets at Fermilab are designed to have characteristic

impedance of sub-multiples of 50 Ohms so that standard cables may be used. To avoid

reflections in the system, the characteristic impedance of all the components are matched closely.

The output cable from the kicker magnet shown in figure 1 is connected to main switch

thyratron, which is connected to a pulse-forming network. In the thyratron switch off state, the

magnet is not connected (see figure 1) and as a result, it is possible to change the pulse forming

Figure 1. Pulse Forming Network or Pulse Forming Line

network slowly to a high voltage (kV). The thyratron switch is then closed and a large current is

quickly connected to the output of the magnet.

4
 As a fact, the electrical characteristics of the coaxial cables used in the accelerator at

Fermilab have to meet very rigid specifications. Hence, measuring accurately the properties of

transmission line parameters such as characteristic impedance, phase velocity, attenuation,

capacitance, inductance and other characteristics are vital. In general, the rise time and

attenuation of transmitted pulses must satisfy certain minimum conditions in order to inject or

extract the particle beams efficiently.

The majority of the circular accelerator is a bending magnetic field (see figure 2).

5
 Figure 2. Bending Magnetic Field

The beams need to be steered between different accelerators. Due to economical reasons, strong

electric fields cannot be produced for high-energy accelerators. Also, to maintain high interaction

rate of experiments, it is essential to accurately steer these beams (see figure3). Therefore,

Figure 3. Beam Steering

the distortion of beam diagnostics and monitoring signals needs to be known. Hence, the

theoretical methods, which allow computation of pulse response for different types of

transmission cable, need to be calculated. As it may be realized, the analytical calculation of

coaxial cables response to an input signal with an arbitrary frequency spectrum is relatively

complex and needs extensive calculation [1]. However, the frequency content of pulses applied

6
is usually concentrated with a limited frequency range. As a result, it is sometimes possible to

find simple solutions that describe the behavior of cables with good approximation. Hence, an

analysis was undertaken to study frequency and pulses response of coaxial transmission cable

using an oscilloscope to analyze waveforms.

Transmission lines play a central role in extraction and injection of beams. For instance,

they are used to transmit pulses from the pulse forming network through the switch thyratron to

the kicker magnet (see figure 1). A transmission line is a distributed capacitor and inductor.

Energy is stored in the line when the center conductor is charged to high voltage. They have a

characteristic impedance that produces a constant voltage output pulse when discharged into a

resistive load, such as at the end of kicker magnet. The output pulse has amplitude of one half of

the charging voltage when the characteristic impedance has the same value as the resistance of

the load. Again, transmission lines provide one of the most essential exceptions to the low

frequency principles that the signal source should have a source impedance small compared with

the impedance of the load being driven and that the load should present input impedance large

compared with the source impedance driving it. The equivalent rule for transmission lines is that

the load and (the source) should present impedance equal to the characteristic impedance of the

line. As a fact, this transmits the maximum power to the load.

II. THEORY

A. Transmission Line as Pulsed power Modulator

Most accelerator applications for pulse modulator line require a constant voltage pulse. A

critically damped waveform is the closest a modulator with a single capacitor and inductor can

7
approach constant voltage. Producing the desired pulse may be understood in terms of theory of

wave propagation.

Consider a transmission line with a current-fed source that produces a rectangular pulse.

At the instant when the switch is opened, the current IO from a constant current source divides

equally between the line and the resistance load if the resistance of the two are equal. A voltage

wave of amplitude IoZo traveling down the line is reflected with reversal of sign at the short-

circuited end. It travels back to the input end in a total time of

(2.1)

2 =

when the reflected wave reaches the input end, the voltage drops immediately to zero because the

line is matched by the resistance RL=Zo and this results no reflection at the input end. However,

the line is fully charged with magnetic energy since a constant current of I o is flowing through it.

The voltage pulse generated at the load during the charging period has amplitude of

(2.2)

IoRL

and duration of 2Nevertheless, if the circuit between the constant current source and the

transmission line is broken, a voltage pulse is generated by the resulting discharge of the

magnetic energy stored in the line. The total energy stored on the line is LoIo where Lo is the

total distributed inductance of the line. In general, the energy is equal to the pulse energy

dissipated in the load. In contrast, when the transmission line used voltage-fed current pulse-

generating source, the far end of the line is open-circuited instead of short-circuited. In voltage-

fed line, the line is charged to a voltage Vo, which produces as current pulse on matched load of

amplitude Vo/2RL.

8
 In summary, a discharge of a transmission line through a shorting switch such as the

thyratron into a matched resistive load produces a constant voltage pulse. The magnitude of the

pulse is Vo/2 and duration time of

(2.3)

l
t 2
v

where v is the pulse propagation velocity.

B. Transmission line Equations

Because the properties of coaxial cables such as inductance, capacitance, resistance and

conductance are so inherently distributed and intermingled, it is difficult to recognize them

individually. As a result, a brief review of transmission line theory is essential. The time it takes

a transmission line to respond to an input signal is calculated from transmission line equations

for voltage V and current I as a functions of time t and space coordinate Z.

(2.4)

V
( R L ) I
z t
I
(G C )V
z t

Where

R = resistance/meter
C = capacitance/meter
L = inductance/meter
G = conductance/meter

9
 It must be emphasized that in general, R is not the inverse of G. L, C, G, R, are per unit

length of line as shown in figure 4.

Figure 4. Differential Length of Transmission Line

For spectral amplitudes of voltage and current Vw, Iw, it follows from Eq 2.3 that

(2.5)

Vw( z )
( R jwL ).Iw( z ) Zs ( w).Iw( z )
z
Iw( z )
(G jwC ).Vw( z ) Yp ( w).Vw( z )
z

The above equations are valid if the series impedance/length Zs and the parallel

admittance/length, Yp are arbitrary of w [1]. The general solution of equation 2.5 is

(2.6)

Vw( z ) Vw exp( z ) Vw exp(z )

where Vw+ and Vw- are the voltage amplitudes of the waves moving in the +Z or Z direction

respectively. The characteristic impedance is

(2.7)

Zs ( w)
Zo
Yp ( w)
10
and the propagation velocity is

(2.8)

1
V
ZsYp

C. High Frequency in Coaxial Cables with Losses

A transmission line, whatever its form, has characteristics impedance Zo; meaning a pulse

moving along a line has a ratio of voltage to current. From equations (2.4) and (2.5) we see if

R<<wL and G<<wC then

(2.9)

L
Zo
C

where Zo is the characteristics impedance. For small losses in coaxial cables propagation

constant is given approximately by

(2.10)

1 R G
( R jwL )(G jwC ) jw LC [1 ( )]
2 jwL jwC
or j
R 1
GZo
Zo 2
LC

where is the attenuation constant in the cable conductors and is the phase constant.

11
 D. Losses in Pulses through Transmission Line

The basic structures of the two coaxial cables we analyzed are shown in figure 5. It is

experimentally found that G is negligible for most cable.

Figure 5. Cross Section of AA8740 and AA5966 Coaxial Cable

A coaxial transmission line consists of an inner conducting cylinder of radius Ri and an

outer cylinder Ro. The conductor is separated by a medium with dielectric constant and linear

magnetic permeability . The geometrical formulas for L and C are

12
 (2.11)

2
C
Ro
ln( )
Ri
and
Ro
L ln( )
2 Ri

where

dielectric
of the insulator
permeability of the dielectric

By substituting equations 2.11 into 2.8 the velocity propagation in coaxial transmission line v is

(2.12)

From the above equation it is realized that the velocity in the medium is speed of light. For all
1 c
v
LC
TEM transmission lines the geometric factors in C and L cancel, therefore, the propagation

velocity is determined only to be the properties of the medium filling the line. Further analysis

also show that at any point on the transmission line voltage is linearly proportional to current,

given by

(2.13)

V = IZo

E. Attenuation of Waveform due to Skin Effect

Attenuation is a measure of the loss of energy between the transmitting and receiving

terminal. The loss of energy depends on many factors such as frequency, conductor size and

spacing. Energy loss is high at higher frequencies and all higher frequencies are increasingly

13
attenuated. The reason is that at high frequencies changing magnetic fields does not penetrate

very deep into a good conducting material. As a result, the effective depth of the current

penetration in the conducting cable is negligibly small as compared to the cable dimensions.

Hence, the current density is non uniform a phenomenon known as skin effect. In transmission

lines, the amplitude of the wave attenuates as the wave propagates along the transmission line.

The distance through which the wave amplitude is reduced by a factor of 1/e (about 37%) is

known as skin depth or penetration depth of the line and is given by the equation

(2.15)

2


As the skin depth gets smaller with respect to the inner and outer conductors the attenuation A

and its constant can be analytically determined to be

(2.16)

However, for a matched coaxial cables analyzed, neglecting pure cable delay, the distortion of
1 1 1
A ( )
Ra Ri ln( Ra )
Ri

A
w
signal and cable rise-time is given by

(2.17)

o ( Al )

The analysis of equation 2.17 reveals two important ideas. First, rectangular pulses of

weight 1 with pulse length much smaller than to are severely attenuated. Second, rectangular

pulses with pulse length much larger than to can be approximated as a unit step input pulse with

the output as an error function given by

14
 (2.18)

1
Vout ( y ) erfc ( )
2 y
where
2
erf ( x ) 1 erfc ( x )
exp(-u)
where
t
y
o

For further information regarding equations 2.16, 2.17 and 2.18, the reader is referred to

reference 1.

III MEASUREMENTS

A. Experimental Apparatus

In order to plot the desired electrical signal against a linear time base, the most common

and widely applicable type of indicator on the screen of a Cathode Ray Tube (CRT), is the

oscilloscope. The oscilloscope makes it possible to study the pulse to pulse variations. The

digital oscilloscope LC 584 was used in our experiment. It differs from conventional

oscilloscopes principally, because it is particularly designed for the observation of fast pulses and

hence it utilizes fast sweeps that are synchronous with the signal to be observed. Again, several

adjustments are incorporated for convenience such as a means of turning off the filtering in order

to see the actual signal, varying the delay of the sweep and the signal and amplitude. This

advanced oscilloscope is used to analyze an electrical signal by displaying a replica of voltage

signal as a function of time. On the screen of a cathode ray tube (CRT) is a plot of electrical

signal against a linear time base. The digitizing oscilloscope works by discretizing the signal,

storing it in digital memory and then reconstructing it on fluorescent screen. The data can also be

stored on a floppy disk for further analysis.

15
 B. Measurement Method

In our experiment, pulse responses to several coaxial cable lines were studied; however,

we only concentrated on the AA5966 cable in MI-52 and the AA8740 spool of special cable in F-

17. Not in any particular order, we started our analysis with the AA5966 kicker cable. First,

reading the time domain reflectometer (TDR), the length of the cable was calculated to be

approximately 743ft (226.47 meters). Next, the digital low resistance Ohmmeter (DLRO)

measured the DC resistance of the center and outer conductor to be 0.112 Ohms and 0.194 Ohms

respectively. The circuit arrangement of the procedure is shown in figure 5.

. For the observation or measurement of the pulse output from the pulsar, it was necessary

to find a means by which the voltage pulse could be applied as a signal to the CRT. In respect to

this, a resistive divider was introduced. The resistive divider was a combination of a number of

resistance in series and parallel. To obtain the oscilloscopic presentations of the voltage pulse,

two different trials were performed using the same coaxial cable. The reason, to reduce the

errors associated with the resistive divider.

In the first trial A, a resistive divider output port 1 was connected to the oscilloscope

while the other output port 2 was connected to the input of the cable under test. The output of

the AA5966 the cable under test was connected to channel 2 of the scope. With the second trial

B, using the same cable, the output of port 2 of the divider was connected to the scope while the

16
 Figure 5. Circuit diagram of Cable under Test

output of port 1 of the divider was connected to the input of the cable under test. Again, the

output of the cable under test was connected to the scope. These two trials were performed in

order to measure the time and amplitude of the cable relative to the input voltage. The output

signals of the cable under test were measured and stored in the oscilloscope. Figure 7 shows the

average calculations of experimental input and output amplitudes with respect to time and the

measured output calculations of the AA5966 cable.

Similarly, the above procedure was repeated for the AA-8740 spool of special cable at F-17.

The measured cable length was approximately 793ft (225.25 meters). The DC resistance of the

center and outer conductors were measured to be 0.222 Ohms and 0.230 Ohms respectively. Again,

the result was plotted and graphed in figure 8.

17
Figure 7. Coaxial Cable AA5966

18
 Figure 8. Coaxial Cable AA8740

Figure 9. Comparison of AA5966 (smaller cable in green) and AA8740 (larger cable in red)

IV RESULTS AND DISCUSSIONS

Pulse is a departure of some electrical quantity such as voltage, current or power from

zero or equilibrium value. In general, the number of pulse shapes actually encountered in

practical circuit is legion. However, the irregularities in pulse shapes that are encountered make

it apparent that determining pulse voltage is limited to special applications where the shape is

known and the nature of the pulse response to that particular pulse shape being measured is also

known. Since the measurement of coaxial cable impedance is sensitive to pulse shape,

oscilloscopic measurements are one of the best ways to interpret pulse response. In view of this

19
fact, the LC584 oscilloscope we used in our experiment provided flexibility for average

amplitude measurements and qualitative information regarding the shape of the pulse.

In a circuit where the input voltage pulse is of an irregular shape similar to that sketched

in figures 7 and 8, how does one measure the input voltage? In this experiment, we corrected

such equivalent pulse and approximated it as a square wave. The correction was done by

determining the input pulse amplitude at a point near the center top of the pulse. It may be

observed that the two cable attenuation is not constant with frequency and that there are some

distortions of pulses in both output average pulses. The distortions are particularly noticeable in

the rounding of the leading top corner of the pulse and slowing down at rate of rise as compared

to the calculated. Again, there is a decrease in the amplitude of the output pulses. In order to

find how much the pulse had decreased, we verified the theory for both calculated and

experimental pulse shapes. The calculated pulse shape was found by using the error function in

equation 2.17 for both pulses.

Several calculations were done using high attenuation of 8E-8Np/m and low 6.5E-

8Np/m. However, the best attenuation was 7E-8Np/m for the smaller cable AA5966. Not

surprisingly, as it was noted, the larger cable (AA8740) had the best attenuation of 5E-8Np/m

while its high and low were 5E-8Np/m and 4E-8Np/m respectively. Another striking thing that

was observed in AA8740 was the ripple at the top center of the experimental pulse. This could

be due to a noise or interference. To complete our analysis, an analytical attenuation (equation

2.16) from geometry was also calculated. The calculation for AA5966 was 9E-8Np/m with an

error of 15%. And the calculations for AA8740 was 6.2E-8Np/m with an error of 15%.

When comparing these two figures, note that both calculated and experimental pulse

shapes were obtained by using average values. The significant difference between cable

AA5699 and AA8740 is that the correspondence between the calculated and experimental pulse

20
shapes is closer in AA8740 than in AA5699. This difference occurred because larger cables have

fewer losses than smaller cables due to skin effect. Summarizing the discussion on pulse

forming line, it may be said that all physical system have unavoidable losses and distributed

capacitances and inductances with skin effect in the conductors being the predominate loss.

IV CONCLUSION

High losses of attenuation and distortion of pulse shape are major concerns of

transmission of power by the coaxial cable. The cable is a power supply to the kicker and high

loss pulses means the rise and fall time of the cable is too long which could cause a particle beam

to be kicked unintentionally. During extraction and injection of beam, timing and distance are

vital therefore; longer time causes fewer beams to be loaded in longer intervals.

Our analysis indicated that for perfect pulses such as square wave, bunches of beams

would be kicked in full trajectory without disturbance to the preceeding or following bunches.

Second, the development of pulse-forming line using smaller cable relative to larger cable is

economical but has higher losses due to skin effect. As may be anticipated, no network having a

finite number of elements can exactly stimulate a transmission line. However, we found that

larger coaxial cables such as the AA8740 have lower losses, and can store more voltages and

currents that could result more kick and less unwanted deflection of the beam from one end to

another.

Since larger cables have fewer losses and can store more voltages, our analysis indicates

that the geometry properties of coaxial cables must be investigated with care to determine how

closely the loss could be approximated in order to design the most useful and economical

transmission line.

21
 ACKNOWLEDGEMENTS

This has been a great learning experience for me this summer. However, it has not have

been possible had it not be the immersed help, excellent discussions and guidance of my superb

supervisor, Chris C. Jensen. I would also like to thank my alternating supervisor Darren Qunell

for his discussions along the way. My thanks also go to Robert Hively and all the technicians in

the kicker group especially Ken Kellogg for our numerous trips to the data sites.

Finally, my sincere thanks to the SIST committee for selecting me to participate in this

program.

22
 REFERENCES

[1] H. Riege, High-Frequency and Pulse Response of Coaxial Transmission Cables, with

conductor, Dielectric and Semiconductor Losses, CERN 70-4. Proton Synchrotron Department,

February, 4 1970.

[2] F. Kirsten and R. Wallace, Pulse Response of Coaxial Cables Lawrence Radiation Laboratory,

University of California, Berkeley. Rev. July 29, 1966.

[3] Dworsky, Lawrence, Modern Transmission Line Theory and Applications. New York: J,

Wiley, 1979.

[4] Robert Grover Brown, Robert A Sharpe, William Lewis Hughes and Robert E. Post, Lines,

Waves And Antennas, The Transmission of Electric energy New York; J, Wiley 1961, 1973.

[5] Paul Horowitz and Winfield Hill, The Art of Electronics Cambridge University Press 1980.

23
24