Comparative Overview of

Inductive Output Tubes

Rebecca Seviour

June 2011


!
!
!
!
! ! ! !!!!!!!!!!
!
!!!Accel erator Di vi si on
!
!
!
!
!
!
!
!
!
!







Rihua Zeng, Anders J. Johansson,
Karin Rathsman and Stephen Molloy

Influence of the Droop and Ripple of Modulator
on Klystron Output




23 February 2012
ESS AD Technical Note
ESS/AD/0033
!


I. Introduction

An IOT is a beam driven vacuum electronic RF amplifier. This document represents a
comparative overview of the Inductive Output Tube (IOT). Starting with an overview of the IOT, we
progress to a comparative discussion of the IOT relative to other RF amplifiers, discussing the
advantages and limitations within the frame work of the RF amplifier requirements for the ESS. A
discussion on the current state of the art in IOTs is presented along with the status of research
programmes to develop 352MHz and 704MHz IOTs.



II. Background

The Inductive Output Tube (IOT) RF amplifier was first proposed by Haeff in 1938, but not
really developed into a working technology until the 1980s. Although primarily developed for the
television transmitters, IOTs have been, and currently are, used on a number of international high-
powered particle accelerators, such as; Diamond, LANSCE, and CERN. This has created a precedence
and expertise in their use for accelerator applications.

IOTs are a modified form of conventional coaxial gridded tubes, similar to the tetrode,
although modified towards a linear beam structure device, similar to a Klystron. This hybrid construct
is sometimes described as a cross between a klystron and a triode, hence Eimacs trade name for IOTs,
the Klystrode. A schematic of an IOT, taken from [1], is shown in Figure 1. Similar to a tetrode the
initial part of an IOT consists of a cathode with a control grid in front, the drive RF causes the beam
passing through the grid to undergo modulation, operating the grid in this fashion can enable the IOT
to act as a Class C amplifier. The beam is then accelerated by passing through a high voltage DC
region, where magnetic lens focus the modulated high-energy electron beam through a small drift tube,
similar to a klystron. This drift tube prevents the backflow of electromagnetic radiation. The bunched
electron beam passes through a resonant cavity, equivalent to the output cavity of a klystron. The
electron bunches excite the cavity, and the electromagnetic energy of the beam is extracted by a
coaxial transmission line [1].

The highest frequency of operation achievable in an IOT is limited by the grid-to-cathode
spacing. The electrons must be accelerated off the cathode and pass the grid before the RF electric field
reverses direction, the limit in voltage stand-off to frequency dependence limits the upper frequency
attainable of IOTs to approximately 1.3 GHz.


Figure 1 schematic of an IOT

One key issue, common to all gridded tubes, is the effect of heat from the cathode evaporating
cathode material which then condenses on the grid, as material accretes on the grid the gap between
grid and cathode narrows eventually electrically shorting the cathode and grid. In addition, the
emissive cathode material on the grid causes a negative grid current, although using modern
coatings/materials on the cathode/grid can mitigate these effects.

IOTs by virtue of their design can operate as Class C amplifiers, with achievable efficiencies
between 70-80% obtained in operation. IOTs also offer a monotonic dependence of P
in
to P
out
, as
shown in Figure 3, where the maximal efficiency is obtainable over the power range of operation. Due
to the simplicity of design the IOT represents a very compact (electrically short) RF amplifier with
gains usually between 20-23 dB. Modern IOT designs are able to achieve an average lifetime of
approximately 40 K hrs, with the most common failure arising from damage to the cathode.

III. ESS Requirements

The ESS requires a number of RF sources and amplifiers at 352MHz and 704 MHz, operating
over a range of RF powers. Current thoughts are to use 1MW 704MHz Klystrons to supply the
medium beta and high beta cavities. For the spoke cavities the current proposal is to use 28 tetrodes at
352 MHz with a forward power of 360KW. These tetrodes will be driven, by 20KW solid-state RF
amplifiers/sources. For the DTL tanks and the RFQ 1MW 352 MHz Klystrons will be used. These
requirements also have to be balanced against the other key deliverables of the ESS, namely achieving
at least 95% availability and as a green carbon neutral facility. This necessitates that all technologies
used in the ESS must be highly efficient, with long life times and high reliability. This last requirement
is critical for the RF sources, where the majority of failures and energy losses occur in current
accelerator facilities.

Currently no manufacture has a working 704 MHz klystron, Thales is working on a 704MHZ
prototype for CERN. Thales does market a 1MW 352MHz klystron suitable for application in the
ESS, and this model has already been tested with parameters similar to those required for operation in
the ESS. Existing tetrodes have been identified (TH 781) that can supply the 360KW power at 352
MHz, although the current identified tube is required to operate outside the manufactures specified
operational range.

IV. Comparison with other RF amplifiers

At this point it is salient to compare IOTs with other RF amplifiers, such as the; Klystron,
Tetrode, Triode, Traveling Wave Tubes, and solid-state. Solid-state devices can act both as
oscillators and amplifiers, with efficiencies of over 85%, although this technology does have a number
of issues, including current technologies readiness state, and will be discussed in greater detail in a
separate document. Other vacuum electronic amplifiers such as the Gyrotron and Bakertron only
operate at frequencies well beyond the range of interest for the ESS and hence are not discussed here.
Magnetrons do offer high power, high efficiency, at frequencies ideal for the ESS, and are cheap to
manufacture, but the innate phase drift inherent from the principle of operation means multiply sources
need to be phased locked [2]. Complex schemes for locking the phase of magnetrons do exist [2], but
the associated cost, physical footprint, and technology readiness state renders magnetrons currently
unsuitable for deployment in the ESS. The Traveling Wave Tube (TWT) is a beam driven, vacuum
electronics, amplifier, the principle of operation is the synchronization between wave dispersion curve
and electron beam, at periodic points to initiate energy transfer [3]. The principle is to effectively slow
the wave down to match the velocities of wave and beam, this is achieved in a variety of ways (folded
wave guide, Helix, dielectrically loaded), but the result is the same, a slow wave structure. Although
simple in concept TWTs present the most challenging vacuum electronic amplifier to manufacture,
resulting in the largest cost per W to produce. TWTs operate with varying degrees of efficiency,
ranging from Class A-F, over frequency ranges from 300MHz to 100 GHz. The maximum power
output that has been achieved to date from a TWT is 15KW[3], with very low efficiency. The low
power and high cost make TWT unsuitable for accelerator applications compared to alternative
amplifiers.


Tetrode/Triode
Both Tetrodes and Triodes belong to the same class of vacuum electronics as the IOT, gridded
vacuum tubes. One key advantage of the IOT over both the tetrode and triode is gain, tetrodes and
triodes offer a gain of between 13-15dB [3], with efficiencies around 60% (c.f. TH391 [4]). The gain
of a tetrode can be increased but at the expense of both efficiency and the maximum sustainable drive
power. In comparison the IOT offers a gain of between 20-23 dB, with an efficiency of >70% [3].
Although the most important discriminator is average life-time, tetrodes and triodes have expected life-
times <20K hrs [5], compared to the IOT with lifetimes >40K hrs. For example in 2010 CERN had to
replace 25% of their Tetrodes on the LHC [6]. The cost of an individual Tetrode is the cheapest HPA
(compared to IOT/Klystron), the cost due to replacements due to failure (2-3 Tetrodes replacements
per IOT/klystron replacement) balances over all cost.

The anode of a tetrode is part of the output circuit and therefore restricted in size, and operated
at high-voltage and hence more suited for forced air-cooling than for water-cooling. This limits the
tetrodes power capability and shortens its lifetime expectancy. The collector of an IOT is entirely
separated electrically and grounded, enabling water cooling, and hence can be designed for almost any
power capability, thus contributing to high lifetime figures [8].

Other options considered for the ESS include using the TH781 tube to provide RF power for
the spoke cavities at 352 MHz with 360KW power. Experience in operation of the TH781 has been
gain from LANSCE [7]. Figure 2 shows the operation of the TH781 tetrode at 200MHz, not that for
360KW power the gain is 13.5 dB and the efficiency is 50%. Rolling the tube outside of its operational
range to meet an ESS requirement of 352MHz is possible, but the efficiency will be reduced to
approximately 40%. This difference in gain between tetrode and IOT has a marked effect on the cost of
the drive RF source and substantially reduces the cost difference between the IOT and tetrode.


Figure 2 showing efficiency and gain for the TH781 tetrode, from reference [6].
The output cavity of the IOT is an evacuated integral part of the tube, entirely on ground
potential, whereas the tetrode requires an external coaxial cavity under atmospheric pressure,
complicated by thin high-voltage insulation foils embedded in its walls. The tetrode cavity assembly is
further complicated by a coaxial cavity for the RF neutralization of the screen-grid and a coaxial cavity
for the RF input. The IOT, on the other hand, does not require a screen-grid circuit, and its RF input
adaptor is fully separated from the output circuit, thus reducing reliability hazards [8].

In an IOT the RF power is extracted from the modulated beam via a re-entrant cavity in the
DC-free space between anode and collector. Uninfluenced by DC potentials, the peak RF amplitude in
the cavity can grow as high as the beam voltage, causing the efficiency to reach its highest possible
value. In the output cavity of a tetrode the anode potential is superimposed on the RF voltage. To
prevent excessive screen-grid current, the peak RF amplitude must not exceed the voltage difference
between anode and screen-grid potentials, thus limiting both output power capability and efficiency
[8].


Klystron
Klystrons are Class A amplifiers where the DC input is independent of the RF drive, for zero
RF drive there is zero RF output although full DC beam input. To prevent possible destructive heat
dissipation at the collector it is necessary to the interlock the DC input with RF drive. Whereas for the
Class C IOT amplifier, there is no current at the collector unless the RF drive is non-zero, this occurs
because in the IOT the control grid is DC-biased to cathode-current cut-off.

Klystrons do offer the advantage of high-gain, typically between 40 and 50 dB. Although high
gain can make klystrons susceptible to RF feedback from a DC-isolated collector to RF input, resulting
in phase instability and self-oscillation. Although the IOT is less sensitive to these issues, as a bunched
beam is produced directly from the cathode region, without the need for gain cavities, the IOT does
suffer from a lower gain, typically 23-24 dB.

Variations in the output signal phase from a klystron are dominated by small variations in the
supplied beam voltage, where the phase deviation is related to electrical length, and is typically 10
degrees per 1% change in beam voltage. Whereas the IOT is electrically short, resulting in a phase-
pushing factor approximately 10% of a klystron, making the IOT 20 dB less sensitive to power supply
voltage variations.

Amplitude sensitivity to beam supply variation is also different between the klystron and IOT.
The fundamental- frequency value of cathode current is directly determined by the RF input, and is
relatively independent of beam voltage. For a given value of RF output there will be an optimum beam
voltage. Less than optimum voltage saturates the output, leading to excessive body current (an effect
Efficiency and Gain
37.8 W Output Transformer

RF Pout
O
v
e
r
a
l
l

E
f
f
i
c
i
e
n
c
y

(
%
)

P
o
w
e
r

G
a
i
n

(
d
B
)
Ea=16.65 kVDC
Eg2=1.4 kVDC
Eg1=-250 VDC
Power Transfer Characteristic
37.8 Output Transformer

Pin (kW)
P
o
u
t

(
k
W
)
Ea=16.65 kVDC
Eg2=1.4 kVDC
Eg1=-250 VDC
DC, and anode current was 7 Amps with no RF, and 44
Amps peak at 407 kW pulsed RF output. The maximum
DF was 10% for this class AB
2
operating point.
Figures 4a (top) and 4b (bottom): FPA Test
Operating with 19 kV DC anode voltage raised the
power gain from 13.5 dB to 14.8 dB, an advantage for the
selection of a driver stage. However, this reduced the
designed safety margin in the anode DC blocking
capacitor, composed of wraps of 0.13 mm Kapton

film
between the anode connector sleeve and the cavity wall.
Considerable trouble was initially encountered with
surface flashovers at the blocking capacitor, caused by an
under-damped response, ringing at 400 KHz during the
HV turn-on transient. The peak overshoot was then
reduced from 33 to 22 kV by installing a series RC
network at the cavity B
+
terminal, and by a parallel RL
network in series at the capacitor bank output terminal,
impedance-matching the DC HV cable at both ends for
high frequencies.
A novel method was used to adjust the amplifier for
optimum linearity in real time, using the XY mode on a
TDS784D digitizing oscilloscope. RF drive came from a
signal generator, which was modulated with a pulsed
linear ramp function. The input channels were connected
directly to RF samples from input and output directional
couplers, without using detectors. Both polarities of the
RF cycle are displayed, resulting in symmetry seen in the
display in figure 5. The more linear trace (red) results
from operating with -250 volts control grid bias and the
second (blue) trace with -350 volts bias. Screen voltage
was 1.4 kV DC and anode voltage was 16.7 kV for both
measurements. Note that the peak power (509 kW) is
slightly higher for the second (blue) trace, but the gain is
lower (12.5 dB). The bend in the curve at low power is
typical for a tetrode with reduced quiescent anode current.

Figure 5: Linearity Test for Two Operating Points
RF DRIVER STAGE
There are two variations of driver/preamplifier required.
The existing solid-state drivers at LANSCE are capable of
5.5 kW peak at 15% DF; these will be reused for the new
IPA for DTL tanks 2 through 4. At DTL tank 1, however,
a new solid state amplifier will be required, with a
maximum output rating of 24 kW. This class AB
1
amplifier will be obtained commercially, using either
NMR/MRI pulsed amplifiers or a modified version of a
band IV television transmitter.
PROJECT SUMMARY
The Thales cavity amplifier and TH781 tetrode have
been tested for two upcoming requirements at LANSCE.
As an IPA to drive a future TH628 Diacrode

FPA with
120-150 kW, the amplifier provided over 16 dB power
gain with 50% efficiency or better. As a stand-alone FPA
to drive a 5 MeV Alvarez DTL tank, the amplifier
provided 13.5 dB power gain with 50% efficiency or
better. It can also be used to drive a 200 MHz RFQ in the
future. Power supplies, driver amplifier and coaxial
circulators are being specified for the complete
installation.
REFERENCES
[1] J. Lyles,C. Friedrichs, 'LANSCE 201.25 MHz Drift
Tube Linac RF Power Status, Proc. International Linac
Conference, Geneva, 1996.
[2] J. Lyles, S. Archuletta, G. Bolme, D. Clark, J. Davis,
D. Keffeler, R. McCrady, 'A New 201.25 MHz High
Power RF System for the LANSCE DTL, Pr oc.
European Particle Acclerator Conference, Paris, 2002,
pp. 2329-2331.
[3] W. Bennett, 'New Beam Power Tube for UHF
Service, IRE Transactions on Electron Devices, Jan.,
1956, pp. 57-61.
[4] M. Hoover, 'Advances in the Techniques and
Applications of Very High Power Grid-Controlled
Tubes, Proc. IEE, v. 105, Part B, Suppl. 10, Nov. 1958,
pp. 550-558.
[5] R. Faulkner, T. Boyd, 'LAMPF 200 MHz Power
Sources, Proc. Proton Linear Accelerator Conference,
Upton, NY, 1968, pp. 87-90.
Proceedings of EPAC 2004, Lucerne, Switzerland
1080
also seen in klystrons). Excessive beam voltage produces higher collector dissipation. Small voltage
variations around the optimum value have much less effect on RF power output than comparable
variations in a klystron.

The efficiency of a Klystron is highly dependent upon drive power, with maximum efficiency
only available at saturation. Although for effective control over the phase/frequency the Klystron must
be operated off saturation, i.e. below maximal efficiency. This is not true for the IOT where the
monotonic dependence of P
in
vs P
out
means an IOT operates at maximal efficiency at points below
maximal power, as shown below in Figure 3.




Figure 3 Klystron and IOT P
in
vs. P
out
. Taken from presentation made to ESS by CPI.



V. State of the art

As the IOT was primarily developed in the 1980s as a klystron replacement for TV
transmitters, availability of current off-the-shelf IOTs is limited in frequency to 470-800 MHz, with
power ranges up to 500 KW CW, and 1 MW pulsed. To the best information available, currently, no
manufacturer provides IOTs which would meet the ESS specifications, of 352MHz at 360KW/1MW or
704MHz at 1MW. Although recently interest in the use of IOTs for particle accelerator, security and
high-energy directed weaponry is stimulating interest in the development of IOTs suitable for high-
power, high/low-frequency applications, which would also meet the ESS requirements. The current
status of all know (to the author) companies, research projects to establish high-power IOTs are listed
below;

a) Calabazas Creek Research,
Calabazas Creek Research Incorporated (CCR) is a research organization with extensive expertise in
R&D of vacuum electronics. CCR has recently completed fabrication and assembly of a 350 MHz, 200
kW CW, Multiple Beam IOT (MBIOT) prototype. The target efficiency is 70% with a predicted gain
exceeding 22 dB. All parts have been received and fabrication is in progress with an expected
completion date of September 2012, with testing scheduled for fall 2012 [9]. The specifications of this
tube would be ideal for application within the ESS.

b) CPI,
CPI are one of the largest manufactures of vacuum electronics, with extensive expertise in IOT
research and manufacture. CPI does not currently have an IOT that would meet the ESS requirements
the company does have tubes similar in specification, such as, the 2KDW250PA (250 kW CW at 267
MHz), and the SDI IOT (500 kW peak, 50 kW avg. at 425 MHz..

CPI \ previously had an intensive research programme into developing a high-powered (1MW) coaxial
IOT. CPI is considering reinitiating this development programme for the ESS, with the right
motivation. To this end they have provided the following schedule for development;

352 MHz, 400KW, >70% efficiency, 23 dB gain.
Prototype 15-18 MARO
Production 2-4/Month at 6 After Prototype

704 MHz 1MW, >65% efficiency, 23 dB gain (5KW drive power).
Prototype 24-30 MARO
Pre-production 2/month 30 MARO
Production 10/month at 42 MARO


c) E2V
E2V are an established manufacture of vacuum electronics, recently their focus has shifted more to
CCD manufacture, although they still offer a range of high-frequency, low-power IOTs. Also E2V in
conjunction with Lancaster University had a 3-year research programme to consider a 350 MHz
1MW Radial Multiple Beam IOT. The programme ended in 2007, with no conclusive results.


Further developments are taking place in cathode technology, to produce new materials and structures
to form cathodes with much higher life-times, see below in Figure 4 taken from reference [10]. As seen
from reference [10] these new cathode designs offer to increase cathode life-time quite considerable, to
approximately 9 years of continuous use.


Figure 4 showing developments in cathode life time [10].


VI. Discussion

As discussed above the IOT offers a number of advantages over both the Klystron and the tetrode.
Table 1 shows the key parameters for each; klystron, tetrode and IOT, for the example presented, in
800 850 900 950 1000 1050 1100 1150 1200 1250 1300
0.1
1
10
100
1
.
10
3
M-Type Cathode Life vs Op. Temp
Degrees Centigrade true
k
i
l
o
H
o
u
r
s
'
1 2 4 8 16 32
table 1, the assumed output power from each tube is 360KW, cost does not include the cost of the
power driver. Another major acquisition cost driver for any RF system based on the klystron or tetrode
is the collector cooling system. As seen in table 1 the difference in cooling overhead between the IOT
and other tubes is rather marked, which implies the cost differential, for a system using 30 devices
each, is substantial. The size of the IOT, comparable to the tetrode, is much smaller than the klystron,
this creates an easy of both maintenance are replacement, which could have a major impact on
recovery time in the event of failure. The increased gain of the IOT compared to the tetrode has a
dramatic impact on the cost, maintenance and footprint of the RF driver. The low life-time of the
tetrode creates a major impact on the cost effectiveness of the tetrode, for example we would expect a
failure rate of 3 tetrodes for each klystron/IOT that would fail. The cost figure represents the base cost
of the tube, the number in parentheses is the cost for the RF driver (costings from quotes, except for
the klystron).

Table 1: based on producing 360KW output RF.
Gain Efficiency Drive Power Cooling Life time Size Cost
(Drive P in) overhead K euros
Klystron 43dB 30% 1 8 W (40W) 840KW 50 KHrs 3m X 1m 280 (1000)
Tetrode 12dB 40% 20 KW (30KW) 549KW 15 KHrs 1m x 1m 70 (97)
IOT 23dB 70% 2 KW (3KW) 154KW 45 KHrs 1m x 1m 300 (10)


In summary IOTs are very useful for the following performance characteristics:
Very high conversion efficiencies, above 70% in class C operation
High efficiency even at reduced drive power
Monotone increase in output power even at the point of maximal efficiency
Linear phase characteristics ideal for phase control
Substantial higher Gain than tetrodes (lower than a klystron)
Much smaller and lighter than a klystron (comparable to a tetrode)
Lifetimes comparable to klystron, much larger than for the tetrode
Substantially lower power requirements, than a Klystron or tetrode
Substantially smaller, cheaper, drive power than the tetrode
Substantially lower cooling requirements, compare to Klystron and tetrode
Reduced sensitivity to variations in supply power

Currently no IOTs are available off the shelf that meet the requirements of the ESS, although
as discussed above several companies have close to market research programmes for an IOT that will
meet the specifications of the spoke cavities. These development programmes could within a 4 year
period produce a multi beam IOT at 704MHz with 1MW pulsed power, ideal for the elliptical cavities
of the ESS. For the 352MHz range it is conceivable to use lower power IOTs and combine then using
the same technique as Diamond. Development of a suitable IOT could be achieved by adapting
existing 400MHZ 200KW IOTs, altering the electron-gun and output cavity to function at 352 MHz.
As an estimate this could be achieved in less than a year for the production of a prototype 352 MHz
200KW IOT.

The potential benefits of the IOT amplifier would have a huge impact on the ability of the ESS
in reaching, and sustaining, its over arching principle of efficiency, reliability and cost-effectiveness.
Especially when compared to the current chosen technological options of the tetrode and the klystron.



[1] Eichmeier and Thumm Editors, Vacuum Electronics, Springer 2008, ISBN 978-3-540-71938

[2] I. Tahir, A.C. Dexter, and R.G. Carter, IEEE Trans. Electron Devices, vol. 52, No.9, (2005).

[3] Gilmour, A. S. (1994). Principles of traveling wave tubes. Artech House Radar Library. Boston: Artech
House.

[4] Note the TH391 operates up to 450MHz, the TH781 up to 200MHz, over 200MHz the TH781 with effects
such as decreased effeciency

[5] Heppinstall and Clayworth, EBU Technical Review (1997) ; and private communication with CPI

[6] https://espace.cern.ch/acc-tec-sector/Chamonix/Chamx2011/talks/OB_23_11_talk.pdf

[7] Lyles et. al., EPAC 2004

[8] H. Bohlen, CPI report 2012

[9] Ives, et.al. , IEEE. Proc. IVEC 2010, page 214; http://www.calcreek.com/hardware.html#inductiveoutputtube;

[10] Grant et. al., IVEC 2002