Phase Drift Budget Analysis for 12 GeV 1497 MHz LLRF System John Musson 28-Sept-7 Introduction The 12 GeV upgrade effort included the creation of LLRF Requirements, directed at achieving.4% gradient regulation,.5 degrees of fast phase regulation, and 3 degrees of slow phase regulation [1]. Although the requirements are more demanding than those for the initial 6 GeV CEBAF machine, a digital LLRF Control Module has been designed which conforms to the new specifications. An abbreviated Requirements List appears in Appendix A. During the 12 GeV LLRF design, strict attention was given to phase-versus-temperature performance, in an effort to improve current operating performance by at least an order of magnitude. Original CEBAF LLRF performance was limited to ~1 degree / C (at 499 MHz), and although great attempts were made to physically temperature-stabilize the injector service building, +/- 2 C was the best achievable ambient regulation. In addition, a 5th-order polynomial correction was included in the initial CEBAF LLRF module, but attempts to apply the correction to improve stability were inconsistent and ultimately abandoned. The 12 GeV upgrade specifies a slow, diurnal RF drift of < 3 degrees over reasonable seasonal temperature ranges. A 5 C peak-to-peak change is currently considered reasonable. As part of 6 GeV accelerator hardening, the warm cavity 499 MHz LLRF components were replaced with interim-design modules, in an effort to improve the temperature stability to < 1 ps/ C. The 499 MHz injector upgrade LLRF module provided.1 degree/ C (~73 fs/ C) performance, an order of magnitude improvement. Appendix B provides supporting data for both 499 MHz LLRF systems. Overall system performance involves not only the LLRF Control Module, but also the cabling and coupling elements. Figure 1 is a system diagram, highlighting the LLRF as well as the passive components contained in the control loop. The feedback cable actually consists of 55' of LDF4 Heliax, and a 4' length of FSJ1 Heliax, serving as a jumper. Figure 1. Schematic description of LLRF control loop, highlighting major passive components.
Component Characterization Of particular concern are those elements in the feedback path, which ultimately corrupt the desired cavity field probe measurement. HP 858A Vector Voltmeters (VVM) were applied to the injector LLRF system as a slow-drift phase monitor / correction system, and also provided in-situ system performance data. HP858A specifications are provided in Appendix C. In addition to Injector phase monitoring, the VVM was used to characterize the Heliax and components. In order to verify the published specifications, of the VVM, a test was performed, whereby an identical 1427 MHz Master Oscillator signal was applied to both A and B inputs for several days. The setup and resulting baseline drift are shown in Figure 2. Figure 2. Depiction of setup used to test intrinsic phase drift of HP 858A Vector Voltmeter, in situ, and associated data. Improvements to the CEBAF Injector Driveline in 1998 led to a battery of temperature studies, in order to completely characterize the behavior of the 1427 MHz LO distribution system. Thermal data was subjected to Monte Carlo simulation, and limits were determined for best-and-worst-case ambient temperature fluctuations (with and without active driveline regulation). Subsequently, the temperature data has proven to be useful for Heliax selection, and for phase budget calculations. Figure 3 shows the experimental test setup, with an environmental chamber, used to gather the component thermal performance data. The setup was capable of single-frequency measurements, as well as IF measurements involving frequency translation. Figure 3. Test setup used to perform environmental studies of various cables and components. Ramp and soak cycles are performed, and data collected by hand or GPIB interface. Single-frequency or IF measurements are possible, as shown.
All common-mode items were co-located outside the environmental chamber, so as to limit the effects of room temperature on the the measurements. The DUT was placed inside the oven, with careful attention placed on lengths and typs of interconnect cables within the enclosure (again, to isolate the DUT from any testbed effects due to thermal and/or frequency variations). Straight-through measurements were performed, to verify that the ambient effects were at or below.1-degrees. Typically, the temperature was ramped up in several degree increments, after experiencing a 15-3 minute soak at each point, and then ramped down, but landing on different points. Residual hysteresis provided an indication as to whether soak times were adequate, and also provided information on the stability and repeatability of the DUT. It was discovered that Heliax requires a bakeout process, whereby the cable is allowed to sit for several days at 55 C. The inner dielectric appears to have a mechanical relaxation process, which stabilizes after being subjected to higher-than-expected operating temperatures. Results for several common Heliax products are shown in Figure 4, demonstrating the degree of stabilization, as well as suggesting possible optimum operating temperatures [2]. LDF4 Heliax LDF2 Heliax.1.2.8.6 2 y =.71x -.8134x +.13134 R2 =.945781.4.2 -.2 -.4 -.6 -.2 -.4 -.6 y =.172x2 -.14927x +.19876 R2 =.968127 -.8 -.1 -.12 -.14 -.8 5 1 15 2 3 4 5 1 15 4.16 1 2 -.2 3 4 5 6 7 3.18.1 y =.15219x2 -.118812x +.15811867 R2 =.97581781 -.3 -.4 -.5 -.6 -.7.14.12 y =.183x2 -.7653x +.8564 R2 =.9191.1.8.6.4.2 -.8 1 2 3 4 5 6 Conformable Cable FSJ4 Heliax, Composite, Baked 1.2.45.4 1. FSJ2 Heliax, 55' FSJ1 Heliax, Composite -.1 2.3 y =.126x2 -.111x +.368 R2 =.96574..2.15.1.5.8 y = -.2222x2 +.162947x - 2.29196 R2 =.948174.6.4.2 -.2 2 3 4 45 5 55 6 5 1 15 2 3 4 Figure 4. Measured delay response vs. temperature for various Heliax products, and associated polynomial regression data.
Table 1 is a summary of extrapolated temperature coefficients for the various cables, at a given temperature, normalized to ps/ft/ C. These slopes were derived from the resulting second-order curve fit. In application, the proposed ambient temperature is chosen, and the slope for the specific cable is used to determine the excess time delay in the neighborhood of temperatures. Large temperature ranges should refer to the entire polynomial. Table 1. Extrapolated temperature slopes for given cables, at specific temperatures. Units are normalized to ps/ft/ C. Temperature, C 5 1 15 2 3 4 45 5 55 6 65 7 LDF2 LDF4 FSJ1 FSJ2 FSJ4 -.813 -.742 -.671 -.6 -.529 -.458 -.387 -.316 -.245 -.174 -.13 -.32.386.196.186 -.1493 -.1321 -.1149 -.977 -.85 -.633 -.461 -.289 -.117.553.2273.3993.5713.7433.9153 -.1188 -.136 -.884 -.732 -.579 -.427 -.275 -.123.294.1816.3338.486.6382.794.9426 -.765 -.582 -.399 -.216 -.33.1497.3327.5157.6987.8817.1647.12477.1437.16137.17967 -.1114 -.988 -.862 -.736 -.61 -.484 -.8 -.232 -.16..1465.27.3985.5245.655 Conformable.162947.14727.11857.96287.7467.51847.29627.747 -.1481 -.373 -.59 -.8147 -.1369 -.191 -.14813 Table 2 is a tabulation of various passive components, also subjected to thermal study. The coupler slopes had much larger error bars, and were very close to the measureable limits of the HP858A. Table 2. Tabulated temperature slopes for given cables and passive components, at specific temperatures. Units are degrees/ft/ C (cable) and degrees/ C (passives), measured at 1427 Mhz. To convert to ps/ C, multiply by 1.95 ps / degree.
An in-situ measurement of a LDF4 probe cable at 499 MHz was performed by setting up a standing wave in the 55' cable at 1.6 GHz, while under full operation. Since the 499 MHz Chopper cavities fully reflect the 1.6 GHz signal, high resolution phase drift measurements were possible over a several day period. Figure 5 demonstrates the procedure, as well as providing data, confirming the ~6 fs / ft / C as given by the static oven measurements (nominal temperature = C, variation = ~2 C). Figure 5. Dynamic, in-situ measurement of 55' LDF4 probe cable for interim-upgrade 499 MHz RFCM. A 2-degree variaiton (measured at 1.6 GHz), at ~ C (+/- 2 C) confirmed the oven measurement and regression prediction of ~6 fs/ft/ C. Data courtesy of Tomasz Plawski. Finally, the a 1497 MHz LLRF receiver protoype for the 12 GeV upgrade was evaluated, and measurements were compared to the predicted performance of the system, as well as to previous 499 MHz module performance. Figure 6 demonstrates the -.213 degree / C slope, measured at 1497 MHz. The resulting ~4 fs/ C temperature coefficient is below the.5ps/ C receiver specification of the 12 GeV LLRF Upgrade Requirements. Figure 6. Relative phase drift vs. temperature for 1497 MHz prototype receiver, demonstrating a 4 fs / C tempco.
Tabulated System Performance Although the klystron gallery service buildings are temperature controlled, it is not uncommon to see several degrees of diurnal fluctuation. Presuming a +/- 3 C change, it it possible to construct a phase drift budget for the entire system, shown in Table 3: Table 3. Tabulation of phase drifts for 1497 Mhz LLRF system. Ambient temperature = C, range = 6 C (+/- 3 C). Component LLRF Module Feedback Coupler Feedback Heliax, LDF4, 55', C Probe Heliax Jumper, FSJ1, 4', C Total and RMS +/- 3 C peak / RMS @ 1497 MHz ps / C -.4 -.97 (ps/ C)2.16.94 -.348.121 -.17.1 -.862 ps / C.54 ps / C -5.17ps, peak 2.87 degrees, peak 3.24 ps, RMS 1.8 degrees, RMS Summary Although the LLRF receiver and feedback path represent only a few elements, a great deal of temperature characterization was required to quantify each component, in order to obtain an accurate perfomance prediction. The ~2 degree RMS (3 degree, peak) diurnal prediction represents a value leading to minimal impact on accelerator operations. Several degrees can actually be automatically removed using the MO Modulation system, which gently krests the Linac energy based on a precise mass-spectrometric measurement of beam energy. Operators are also able to easily remove the day-today degree variations by hand. Ultimately, preliminary engineering efforts, as well as prototype measurements, demonstrate the abillity to achieve the short and long-term stability requirements of.5 ps / C, or 15 ps over 1 C to 4 C range, thus satisfying control requirements for the future 12 GeV accelerator. Acknowledgements The author would like to thank student intern, Guy Randall, for his eager participation in painstaking data collection during the thermal testing. Also, to colleagues Tomasz Plawski for sharing his data taken while performing the in-situ probe cable measurement, and Mark Wissmann for assistance with the antique environmental chamber. References 1. H. Dong, A. Hofler, C. Hovater, J. Musson, Performance and Technical Requirements for the CEBAF Energy Upgrade Low Level RF System, M:\ees\RF\12 GeV_LLRF\LLRF Documentation and Specifications 2. Actual spreadshhet is located in the JLAB Operations Electronic Logbook, LLRFLOG, Entry #13921 http://opweb.acc.jlab.org/csueapps/elog2/elog.php
Appendix A General Requirements List Frequency 1497 MHz G*BW 1 MHz IF 7 MHz LO input 1427 MHz Phase Ctrl.5 degrees Amplitude Control.45% Dynamic Range 2 db Setpoint Resolution.1%.1 degree Diagnostics Loopback Cable Fault Thermal Stability +/-.5 db, 15 ps; 1 C < T < 4 C.2 db / C;.5 ps / C Power Use less energy than the sun
Appendix B Thermal Drift of Old 499 MHz CEBAF LLRFCM vs. 12 Gev Digital LLRFCM Bronze Age 1 degree / C > 5.6 ps per C Digital LLRF Module ~.13 degree / C
Appendix C HP 858A Vector Voltmeter Specifications