XP3-1/XP3-2 TEST RESULTS There were several factors contributing to the poor performance of these “DFM” PPMfocused 75-MW klystrons: 1. Although the smaller “DFM” gun ceramic had been tested in a beam tester with 3.2 µs pulses at full voltage, it had been designed for only 1.5 µs (before the pulse length was doubled). The gun in XP3 –2 proved to be not sufficiently sturdy for the longer pulse length. The ceramic in that tube was damaged early at test; subsequently, XP3-2 was limited in beam voltage and peak power. 2. The output cavities in these klystrons have a potential trapped mode at approximately 11.7 MHz. This results from the mismatch presented to the output cavities by the mode converters ahead of the windows. We suppress it by coupling “resonant loss” cavities between the output cavity and the 2 windows. These cavities are designed so that, when located correctly on the waveguide, they match the window and mode converter to the output cavity at 11.7 GHz. The trapped mode is then loaded over a narrow frequency range. The potential for this oscillation, however, is exacerbated further if the 3 “penultimate” klystron cavities, which are tuned well above the operating frequency in order to improve rf beam bunching, and are therefore are tuned close to the 11. 7 GHz frequency. In XP3-1 the resonant loss cavity was placed at an incorrect location. Also, in both klystrons, the “penultimate” cavities were tuned too close to 11.7 GHz. 3. The 2 polepieces in the gun electromagnet, which is used to launch the beam into the PPM stack, were split in both klystrons, causing transverse magnetic fields to exist in the vicinity of the cathode. These fields were an order of magnitude higher that we specify for our PPM stacks (about 1% vs. 0.2% of the axial field). This may have caused “cork-screwing” in the beams of both klystrons. Evidence of that effect may have been the reduced available range of electromagnet current in adjusting the size of the klystron beam for optimum output power. 4. The same comment applies to potential transverse magnetic field in the PPM “clamshells”, which was also measured to be as high as 0.8%. Although the clamshells had been tested for such fields, the tests (which must measure a few transverse gauss in the presence of an axial field of 3000 Gauss) were questionable. They have been repeated and the results are shown in Fig. 10. Unfortunately no analytical or simulated predictions have existed on the limits for transverse fields in PPM stacks. This is being remedied by using a recently available 3D code. Item No.1 will not be a problem, since the specified pulse length has reverted to the original 1.6 µs. Items 2 and 3 have been corrected in the projected repair of XP3-2. Item 4 is being investigated further, but the rebuilt XP3-2 will employ the same clamshells. A second rebuilt XP3 will use integral (non-split) polepieces.
3
SLAC PPM KLYSTRON DEVELOPMENT 50 XP 1 1996-99
XP 1 2 1997-99
XP 3 Diode 2001
XP 3-1 (DFM) 2002
XP 3-2 (DFM) 2002
XP 4 2003
Power Output Design Attained
50 50 – 60
75 70 - 90
Beam Tester
75 50 – 70
75 30 – 40
75
Pulse Length Design Attained
1.2 2.4
1.5 3.0
1.5 3.2
3.2 3.2
3.2 2.8
3.2
60 120
60 60
120 120
120 120
120 60
120
50-2.4-120-57
79-2.8-1-62
1wk 490kV 3.2µs, 120 pps
70-0.3-120-55 50-3.2-120-39
40-0.5-120-31 30-2.8-120-24
Brazed to form the drift tube Large (11.5”) tapered
Brazed to s.s. tube
Brazed to s.s. tube
Clamp on
Clamp on
Brazed to s.s. tube
Large, tapered
Small (9”), straight
Small, straight
Small, straight
Large, tapered
Good
Good LC
Good LC
Good LC
Good Gun Breakdown4 LC
Good 99.5% but a narrow coil range
PRF Design Attained Best results (simultaneous) Po-µs-Hz-η% Polepieces Gun Seal Gun Stability
3
Beam Transport (no rf)
4
Excellent 99.9%
Fair 99%
Good 99.5%
Output Stability
Excellent
Excellent
N/A
Average Power Design Attained
3.6 kW 14.4 kW
0.1 kW 0.2 kW
DC 23 kW 49 kW
1
Good 99.5%
11.7GHz Output 4 Oscillation
Good
29 kW 19.2 kW
29 kW 10 kW
29 kW
Tube originally tested in 1997, but was retested in 1999 for longer pulse length and higher PRF. Perveance: 0.6. All subsequent klystrons built with K=0.75. Tube was rebuilt once in order to coat the polepiece IDs and eliminate multipactor. 2 This tube was built with permanent magnets instead of gun coils, and with NdFeBo, instead of SmCo magnets. Both “DFM” steps resulted in problems. It was rebuilt twice, to eliminate gun and output oscillations. It could not be operated at any significant average power since no body cooling was provided. 3 LC: Loss collar around gun stem to suppress gun oscillations. XP 1 and XP 2 were rebuilt to add this feature where XP 3 came originally with it installed. No collar was installed on the 50 XP. 4 Reason for not meeting specs.
4
Modifications incorporated to the rebuilt XP 3-2 • Cavity tuning changed for optimum efficiency and oscillation avoidance. • Resonant loss cavity placed at correct location. • Gun coil split polepiece remade without split. • Gun from XP 3-1 will be used in place of the XP 3-2 gun, which was damaged from arcing. • And, most important, no more 3.2 µsec pulses to trigger oscillations or cause gun breakdowns
5
110 tunings investigated: Saturated Gain and Pout vs. Sum of the frequencies of cavities 5, 6 and 7. Multiple solutions found which avoid high gradients and pre-determined unstable frequencies, and also give good BW, gain and power out. 80
XP3
XP3 Rebuild
75
Gain, Pwr
70
Pout
65 60
Gain
55 50 11700 X 3
45 34800
35000
35200
35400 Sum of F's (MHz)
7
35600
35800
36000
Old Loss cavity location S11 vs. new location shows better than 10dB improvement in reflection from mode converter-load assy at 11.7GHz (good match).
0
Old
-2
Meas. Load A: TS1794-4
-4
Calc. Load A
S11 (dB)
-6 -8 -10 -12 -14
New Location
-16 -18 -20 1.3
1.5
1.7 1.9 2.1 Loss Cavity Location (inches)
8
2.3
2.5
Transverse B fields at gap centers (minimum points on curve below) average > 0.2% for the lower XP3-1 clamp-on
1.0%
Bt/Bz(RMS)
0.8%
0.6%
0.4%
0.2%
0.0% -500
-450
-400
-350
-300
-250
mm 10
-200
-150
-100
-50
0
11