2510

IEEE FRANSACTIONS ON ELECTRON DEVICES, VOL.

ED-32, NO.

11, NOVEMBER 1985

[I31 P. Revesz, M. Wittmer, J. Roth, and J. W. Mayer, Epitaxial regrowth of Ar-implanted amorphous silicon, J. Appl. Phys., vol. 49, p. 5199, 1978. [14] J. Garrido, E. Calleja, and J. Piqueras, Deep centers introduced by argon implantation bombardment in n-type silicon, Solid-State Electron., vol. 24, p.1121,1981.

TABLE I

PRCGRPM
SLACKLY JPNDISK SSKLY*

AUTMNS)

1
90.1
164.8

v2
105.2
160.6

3
8.4
55.5

D. R u s s e l l l T W . essel-Berg H. Yonezaua, Y. Okazaki (SLAC)

---

51.9 119.0

52.5 117.5

23.3
52.7

The Input Gap Voltage of a Klystron

NOTES:

V j , V , , V 3 a r et h e rms v o l t a g e sa tt h ec o l dr e s o n a n tf r e q u e n c y , and detuned 2% from cold resona t the hot resonant frequency, f o r t h et e s t a n c er e s p e c t i v e l y ,a l lw i t h beam on. The data case were 90 kV, 54 Amp, a=.9525 m , b=.762 o n , d=l.0312 cm, R/Q=lOO, Q E = I O O , Q,=lOOO, P i n = l W, fo=2960 MHz.

Abstract-Following the discovery that several large-signal kljstron programs did not agree on the value of the RF voltage at the inpr t gap of a klystron, the question was examined afresh. Available textbooks were found to make restrictive assumptions about matching and/or tuning of the input cavity, assumptions which are no longer accep :able. A formula was derived, in terms of cold test parameters and drive power, that correctly apportions the power between dissipation, reflection, and beam loading, thereby determining the RF gap voltage.

*This i s n o t t h e P. T a l l e r i c o a t

Same a s t h e program of t h e same name by

I. INTRODUCTION It might be thought that theproblem of calculating the input gap RF voltage of a klystron in terms of the cavity parameters, the 3eam loading, and the drive power, was solved more than thirty year:, ago. If it was, the solution appears to have been lost, at least in some influential quarters: the writerrecently discovered that he had three different computer programs, and an analysis by R. S. Symonf (unpublished), no two of which gave the same answer tothis appal ently straightforward problem. As shown in Table I, the differences were substantial. Clearly, these programs cannot all be satisfying conservation of energy, and the detuned (V, ) value for SLACKLY is obviously not showing even the same functional dependence on tuning as thr: others. Textbooks did not prove helpful; mostly they date from a time when klystrons were regarded as single-frequency or very-nwrowband devices, and make restrictive assumptions: Warneckc. and Guenard [ l ] allow for detuning, but assume that the input cavity is matched at resonance; Collin [2] assumes that the input cavity is tuned but not matched; Hamilton et al. [3] tacitly assume that the cavity is both matched and tuned. For broad-band klystrons neither of these restrictions is acceptable. An acceptable solution must include the effects of cavity mismatch on or off resonance and of beam loading and detuning, and must correctly divide the available drive power between the c avity losses, the real power transferred to the beam, and the reflscted power. We shall assume that the input line is matched at the generator end, and that the generator itself is unaffected by reflscted power. It was found that Symons solution included everything except the beam detuning, but it was expressed in terms of normal klystron design parameters that are not directly accessible to rneasurement. It was considered desirable to have a solution expr-ssed as far as possible in terms of Qs, VSWRs, and other parameters which can be measured directly with the klystron in the opeiating and nonoperating states. The beam loading conductance GB and susceptance BB d3 not meet this condition of direct measurability; they can be calct lated from formulas given by Gewartowski and Watson [4], or (witf considerably more difficulty) from the relativistic formulation by Craig [5]. This calculation is subject toconsiderable uncertainty, be :ause
Manuscript received January 22, 1985; revised June 5 , 1985. The author is with Litton Electron Devices, San Carlos, CA 940701.

the gaplength g has to be fudged to allow for the fringing of the fields in the ungridded gaps that are now almost universally used. There is no general agreement on what fudge factor to use. This problem is discussed in a companion paper [6]. Initially, we shall assume that values for GB and BB are known; but an unexpected result of this analysis was a method of finding GB and BE, and also RlQ and the effective transit angle & , from measurements made only on the completed tube. This development is given in Section I11 below.
11. CALCULATION OF
THE

GAP VOLTAGE

The known parameters are assumed to be the cavity cold resonant frequency fo, the cold unloaded andexternal Qs, Q, and QE, the shunt resistance R/Q, the beam admittance GB jBB, and at frequency f. the available drive power Pi, Then the fractional detuning is

A=-

f -fo
jo

and the detuning due to the beam is We define a beam-loading Q

and a hot unloaded Q

Then the conductance seen from the line is

Note that Q E is not involved in (5); this is thepoint at which at least some of the programs appear to have gone wrong. The corresponding conductance GE is across the cavity as seen from the interaction gap, but it is not across the cavity as seen from the line. Rather, it is behind the observer. BT has two components, the cavity and The terminal susceptance beam susceptances 2s B --+Bp T - RlQ Then the total admittance seen from the line is The equivalent conductance of the drive line is

0018-9383/85/113~0-2510$01.00 0 1985 IEEE

ELECTRON TRANSACTIONS DEVICES, lEEE ON

VOL. ED-32, NO. 11, NOVEMBER 1985

251 1

and the assumption that this line is matched at the generator end implies that this must be its characteristic admittance. In reality, a line of some, standard admittance will be used, but one connected by some form of transformer, which we assume to be ideal, so that it leaves (8) still valid. The complex reflection coefficient is then

explicitly by rationalizing and then taking the modulus, it seems much easier to solve it by iteration, using a first trial value for QB from (3) and the first estimates mentioned above for GBand RlQ. The right-hand side of (14) seems to be just about proportional to QB, and the left-hand side is known from (12), so that very few iterations are needed to find QB to 3 significant figures-the data are unlikely to justify more. Eliminating RlQ from ( 2 ) and (3), we have

But from Gewartowski andWatsons equations and the input VSWR is B, - 2 sin 0 - e(1 cos e ) GB 2(1 - cos e ) - 0 sin 0
=

We now have all the quantities required to define the r m s input gap voltage satisfying conservation of energy 2 vr =-J l + u
Pi, uQd(RJQ>.

0 cot - . 2

Hence the effective transit angle is

1)

(1

BT

2 atan (-0.5/6BQB).

(18)

For this formulation, the values corresponding to Table I are V , = 118, V2 = 121, V, = 50 V.
111. CALCULATIONOF RIQ,

GB, AND BB FROM HOT TEST MEASUREMENTS

Since this is calculated from the deviation caused by the beam, it is the naturally fudged value which we sought earlier. Given the dc beam conductance Go( =ZolVo), we can now calculate GBand BB from

Let us now. assume that we do not know the effective values of RlQ, loT, GB, and Bu, but that we can assume a trial value of 100 for RIQ, and first-cut values for the other three based on Gewartowski and Watsons analysis, with some reasonable fudge factor. We assume that we have an operable tube, and equipment for making normal cold test measurements in the drive line with the beam either on or off. With the beam off we can identify fo, and can measure Q, and QE if they are not alreadyknown. Turning on the beam, but staying at fo, we measure the new VSWR uOH,and find the reflection coefficient magnitude from
IPOHI

and RlQ is given by (3) RlQ 11GBQp (21) Equation (2) can be used as a check on RIQ, except in the case when OT is close to K, where (2) becomes indeterminate.
REFERENCES
[l] R. Warnecke and P. Guenard, Tubes a Modulation de Vitesse. Paris: Gauthier-Villars, 1951, p. 269. Foundations for MicrowaveEngineering. New York: [2]R.E.Collin, McGraw-Hill, 1972, p. 469. [3] D. R. Hamilton, J. K. Knipp, and J. B. H. Kuper, Klystrons and Microwave Triodes, Radiation Lab. Series, vol. 7. New York: McGrawHill, 1948, p. 251. [4] J. W. Gewartowskl and H. A. Watson,, Principles of Electron Tubes. Princeton, NJ: D. van Nostrand, 1965, p. 212. [ 5 ] E. J. Craig, The beam loading admittance of gridless klystron gaps, ZEEE Trans.ElectronDevices, vol. ED-14, no. 5 , pp. 273-278; and correction in vol. ED-16, no. 1, p. 139, Jan. 1969. [6] J. R. M. Vaughan, A model for the klystron cavity gap, ZEEE Pans. Electron Devices, this issue, pp. 2482-2484.

uOH
~

OOH

At this frequency, 6 = 0, BT = BE, and the terminal admittance is

GE + GT - jBB GE GT jBB

QE

1 1 +++ jBB(R/Q) Q, QB

Using (2) we have

Correction to Measurement of Diffusion Length, Lifetime, and Surface Recombination Velocity in Thin Semiconductor Layers
FRANKLIN N. GONZALEZ
AND

ARNOST NEUGROSCHEL

In the above paper, (1) should read

Equation (14) is then an equation for QBin termsof known quantities. While it would no doubt be possible to express the solution

Manuscript received August 8, 1985. F. N. Gonzalez and A. Neugroschel, ZEEE Trans. Electron Devices, vol. ED-31, no. 4, pp. 413-416, Apr. 1984.