Design and Fabrication of a Q-BPM Calibration Signal for the ATF2

Ian Bullock

July 22, 2008

Intro + Methods Draft

Design and Fabrication of a Q-BPM Calibration Signal for the ATF2

Introduction

Accurate measurement of beam position is important in the operation of any linear accelerator. The ATF2 (Accelerator Test Facility) in Japan has as its goal nanometer control of beam position, as well as to produce a 37nm beam size [1]. To accomplish this, when the electrons are removed from the damping ring, they pass through a series of focusing magnets, some of which have cavity beam position monitors (BPMs) attached to them [1]. Through precise measurements of beam position, the equipment can be adjusted to appropriately focus the beam.

The BPMs in the ATF2 which are attached to the quadrupole magnets are known as Q-BPMs. Each BPM has a pill-box shaped cavity with waveguides and antennas to connect coaxial cables [1]. As the electron bunch passes by the Q-BPM, the waveguides pick up signals from the electric field of the pulse. Different signals are produced depending on the position of the electron bunch relative to the Q-BPM. These 6426 MHz signals are downmixed to 20 MHz, digitized, and then analyzed by computer to determine beam position.

Since the Q-BPMs are required to have 100nm single pass resolution and 1 m accuracy [1], careful calibration of the downmix circuitry is needed. Some calibration can be done using the magnet movers for the quadrupole magnets that the Q-BPMs are attached to. However, it is also necessary to calibrate the gain of the electronics, which has been seen to vary significantly even over the course of a few days.

To calibrate the electronics, a calibration signal source has been produced. The source uses the 714 MHz master oscillator as an input, and produces a 6426 MHz output, nine times the input frequency. This signal must then be switched to provide calibration pulses only when the electron bunches are not passing through the accelerator cavity, to ensure that the calibration does not interfere with normal Q-BPM operation.

In general, the goal of the signal source is to imitate the signal that would be generated from a real electron bunch as much as possible while keeping the signal very consistent from pulse to pulse so that it is useful for calibration.

Methods

The project can be divided into three different systems. The first system is the logic level pulse input which determines when the calibration signal is on. The second system is a switching mechanism that receives the logic pulse and turns on the third system, the 6426MHz calibration signal. The first step in satisfying the requirements of the three systems is to come up with designs for each system, some of which may be physically combined in the final product.

The design space was initially restricted in a few ways. Doug McCormick recommended the use of mostly Mini-Circuits microwave components because of their quality, price, availability for small orders, and fast shipping. Two layer ExpressPCB manufactured boards were selected for any printed circuit boards due to the cheap price of the boards and the simple software. It was suggested to produce two versions of the calibration tone device, one using modular shielded parts with SMA connectors, and one using surface mount parts on a printed circuit board.

The Mini-Circuits staff suggested a mixer to use to mix the 714MHz input with a doubled 1428MHz signal to give a 2142 MHz signal [2]. This 2142 MHz signal would then be multiplied with a 3x multiplier to give the desired 6426MHz output. This suggestion was due to the unavailability of a single part to triple 714 MHz.

It was decided that a high isolation solid state switch should be used to turn on and off the calibration signal. A solid state switch was chosen due to the potentially short lifetime of a frequently switched mechanical switch. High isolation is required so that the calibration signal will not be present when the Q-BPMs are picking up the signals from the actual electron bunch.

Theory

A simple set of MATLAB functions was produced to predict circuit performance, to make the choice of amplifiers, filters, and multipliers for the third 6426 MHz system easier. Each function takes as its input a matrix of frequencies and their respective amplitudes in dBm, along with any other relevant parameters for the modeled device, and returns a matrix of output frequencies and amplitudes. The filter model is the most accurate of the functions, as it interpolates using a table of insertion losses at various frequencies obtained from the manufacturer. Perhaps the least accurate of the functions is the mixer function, which behaves as an ideal mixer to output only the sum and difference of the two input frequencies rather than the many other frequencies that a physical mixer produces.

Although Mini-Circuits gave a general outline for the circuit, the addition of amplifiers and filters is important to get as close as possible to a pure 6426MHz output signal at sufficient output power. In general, the amplifiers were chosen first for having the appropriate gain, and then for having a low noise figure (NF). The linearity of the amplifiers was thought to be less important, so IP3 (the third-order intercept point) was not given much weight in deciding on amplifiers. Filters were simply chosen to cut out the unwanted frequencies at various points in the circuit, while not overly attenuating the desired frequency.

The solid state switch was chosen largely based on having high isolation, as well as the convenience of a built in TTL driver so that the switch can be used directly with TTL logic levels.

References

[1] ATF2 Proposal

[2] Mini-Circuits email correspondence with Doug McCormick

[3]