KEK Accelerator Test Facility Low-Level RF and Timing Systems

This paper presents facility-wide measurements of the KEK Accelerator Test Facility's Low-Level RF clock phase-noise power spectral density and discusses the resulting synchronization floor imposed by the stability of its Linac and Damping Ring signal generators, which are critical for achieving the ~100 fs-level synchronization required for nanobeam technology testing.

The Gist

Imagine a massive, high-speed train station where the "trains" are actually beams of electrons moving at nearly the speed of light. At the KEK Accelerator Test Facility (ATF) in Japan, scientists are testing technologies to build the ultimate version of this station: the International Linear Collider.

To make this work, everything has to happen with perfect timing. If the lights, the doors, and the engines don't sync up within a fraction of a trillionth of a second, the whole system fails. This paper is essentially a "health check" report on the facility's internal clock system.

Here is a breakdown of what they found, using simple analogies:

The Master Clock and the Orchestra

Think of the facility's timing system as a giant orchestra.

• The Conductor (The Master Clock): The system uses two main signal generators (like high-end metronomes). One controls the main "Linac" (the long track), and the other controls the "Damping Ring" (a circular track where particles are prepared). The Linac generator is the "Grandmaster," meaning it sets the rhythm for the whole facility.
• The Musicians (The Subsystems): These are the lasers, magnets, and cameras that need to fire or move at the exact right moment.
• The Score (The Clock Signals): The facility sends out a steady "tick-tock" signal (a clock) to every musician so they stay in sync.

The Problem: Noise in the Signal

In a perfect world, the "tick-tock" would be perfectly steady. But in reality, there is always a little bit of "jitter" or "wobble" in the signal.

• The Analogy: Imagine trying to walk in a straight line while someone is gently pushing you left and right. If the pushes are tiny, you stay on track. If the pushes are big, you stumble.
• The Measurement: The scientists measured how much this "wobble" (called phase noise) happens. They looked at the "wobble" over different speeds of change (frequencies) to calculate the total "stumble" (time jitter) in femtoseconds (one femtosecond is one quadrillionth of a second).

The Findings: Two Different Worlds

1. The Main Track (Linac): The Smooth Ride
When the main track is running in its normal mode, the system is incredibly precise.

• The Result: The "wobble" is tiny—about 70 to 120 femtoseconds.
• The Analogy: This is like a tightrope walker who barely sways at all. Even after the signal travels through long cables and gets converted from electricity to light and back again (like a message being translated from English to French and back to English), the timing remains incredibly sharp. This proves the system works well for its intended purpose.

2. The Circular Track (Damping Ring): The Bumpy Ride
Things get messy when they try to speed up the particles in the circular ring. To do this, they have to constantly change the frequency of the clock signal (a process called "frequency ramping").

• The Result: When they engage this speed-up mode, the "wobble" explodes. It jumps from tiny femtoseconds to several picoseconds (which is 1,000 times larger).
• The Analogy: Imagine the tightrope walker suddenly starts dancing wildly while trying to cross. The "feedback loop" used to control the speed-up is introducing a lot of noise, like a microphone picking up too much static and screeching.
• The Culprit: The scientists found that the specific electronics used to manage this speed-up are the main source of the problem. They are the "noisy neighbor" ruining the party.

The Conclusion: What Needs Fixing?

The paper concludes that the main track (Linac) is doing a fantastic job and is ready for the future. However, the circular track (Damping Ring) has a "bottleneck."

To get the entire facility to the level of precision needed for the next generation of particle accelerators, they don't need to fix the main clock or the cables. Instead, they need to quiet down the speed-up mechanism in the circular ring. If they can smooth out that specific "dance," the whole facility can achieve the ultra-stable, sub-100-femtosecond synchronization required for cutting-edge physics experiments.

In short: The facility's clock is mostly perfect, but one specific part gets "jittery" when it tries to speed things up. Fixing that specific part is the key to the next level of performance.

Technical Summary

Technical Summary: KEK Accelerator Test Facility Low-Level RF and Timing Systems

Problem Statement
The KEK Accelerator Test Facility (ATF) serves as a critical testbed for nanobeam technologies supporting the International Linear Collider (ILC). The facility relies on stable pulsed operation, which necessitates precise synchronization between the facility timing system and the Low-Level RF (LLRF) system. This synchronization is vital for maintaining the sub-picosecond stability of electron bunch arrival times and accelerating-field phases. Key subsystems requiring this synchronization include the RF-gun laser, Linac and Damping Ring (DR) high-power RF, Final Focus (FF) cavity beam position monitors (cBPMs), klystrons, and data acquisition (DAQ) systems. The primary challenge addressed is quantifying the phase-noise power spectral density (PN-PSD) of the distributed clock signals to determine the synchronization floor imposed by the stability of the signal generators and distribution networks, particularly under operational conditions involving frequency modulation (FM) and energy ramping.

Methodology
To evaluate the facility-wide performance, the authors conducted measurements of the PN-PSD across the KEK ATF LLRF clock-distribution branches. The measurements were performed using an Agilent E5052B Signal Source Analyzer (SSA) operating in PLL carrier-tracking/direct-homodyne mode.

• Measurement Parameters: The SSA IF gain was set to 50 dB. Data was averaged 20 times with 5 correlations. The offset frequency range was measured from 1 Hz to 10 MHz, with an upper bound of 5 MHz for the 10 MHz reference signal cases.
• System Architecture: The LLRF system is based on two Agilent E8663B Signal Generators (SGs). The Linac SG acts as the "grandmaster," generating a 1428 MHz continuous wave (CW) signal which is frequency-multiplied to 2856 MHz for klystrons and divided for sub-harmonics. The DR SG is synchronized to the Linac SG via a low-noise 10 MHz reference.
• Distribution: Clock signals are distributed via Phase Stabilized Optical Fiber (PSOF) without active delay compensation. The study analyzed various branches, including those feeding klystrons, the RF-gun laser, frequency-ramp electronics, and the Final Focus station.
• Analysis: The PN-PSD data was integrated from 1 Hz to 10 MHz to calculate the integrated root-mean-square (rms) time jitter, providing a quantitative metric for synchronization performance. Measurements were taken in two modes: with FM disabled (nominal) and with FM enabled (operational, specifically for DR frequency ramping).

Key Results
The study provides a baseline characterization of the jitter performance across different facility branches:

• Linac Branches:

  • The Linac SG output exhibits an integrated rms time jitter of 70 fs (FM disabled) and 1.64 ps (FM enabled).
  • After E/O and O/E conversion and distribution to klystron branches, the jitter degrades slightly. At the klystron endpoints, the jitter ranges from 90 fs to 190 fs (FM disabled) and 1.45 ps to 1.72 ps (FM enabled).
  • The RF-gun laser clock shows an rms jitter of 370 fs (FM disabled) and 1.71 ps (FM enabled).
  • The branch serving frequency-ramp electronics shows significant degradation, reaching 6.02 ps (FM disabled) and 6.15 ps (FM enabled).
  • The 357 MHz clock signal shows jitter of 750 fs (FM disabled) and 1.6 ps (FM enabled).

• Damping Ring (DR) Branches:

  • The DR SG output (714 MHz) has a baseline jitter of 90 fs (FM disabled).
  • When the frequency ramp electronics and associated PLL are active with FM enabled, the jitter increases to 1.99 ps.
  • After transfer via heliax cable and amplification, the jitter remains at 90 fs (FM disabled) but degrades slightly to 2.02 ps (FM enabled).
  • The DR analog phase/amplitude feedback further degrades the jitter to 1.09 ps (FM disabled).
  • Sub-harmonics (357 MHz and 178.5 MHz) show increased jitter, reaching up to 2.92 ps with FM enabled.
  • The DR SG 10 MHz reference output shows 520 fs jitter, higher than the Linac SG due to daisy-chain processing within the SG electronics.

• Final Focus (FF) Branches:

  • The signal transferred to the FF station via optical fiber degrades from 90 fs to 5.67 ps (FM disabled) due to the E/O and O/E conversion chain.
  • With FM enabled, the noise increases from 2.02 ps to 5.67 ps.
  • The up-converted 6.453 GHz C-band LO signal inherits this jitter, measuring 5.67 ps (FM disabled) and 6.02 ps (FM enabled).

Significance and Conclusions
The paper establishes that the RF distribution architecture, utilizing E/O and O/E conversion, successfully preserves low-jitter performance (on the order of 100 fs) for the Linac klystron references under nominal settings (FM disabled). This confirms the viability of the current distribution method for maintaining sub-picosecond stability.

However, the study identifies a significant performance limitation when the Damping Ring reference operates with frequency modulation and the ramp-generation loop engaged. In this operational mode, the PN-PSD level increases by several tens of dB, resulting in integrated jitter in the range of a few picoseconds. The authors conclude that the DR ramp-generation path, specifically the feedback loop and associated electronics, is the primary contributor to phase noise and the dominant limitation for further facility-wide jitter reduction.

The significance of this work lies in identifying the specific architectural bottlenecks preventing sub-100 fs stability across all ATF sections under operational conditions. The authors assert that future upgrades must focus on the ramp-generation feedback architecture, including controller design, noise shaping, loop bandwidth, and the location of critical electronics, to extend high-stability performance to the entire facility.