NLC Timing System Baseline Design

 

Baseline stabilization system principal of operation:

Standard Fiber-optical cable has a temperature coefficient of delay of approximately 30ps/C/Km. For 20 degree of X-band, and 15Km transmission length, this would require a temperature stability of about <.01 degree C. The (very expensive) temperature compensated fiber from Sumitomo has a coefficient approximately 10X smaller, but would still require 0.1 degree C temperature stability for the phase transmission cables. This temperature stability requirement was believed to be too stringent, and it was decided to use active feedback on the fiber phase length.

The fiber phase length is measured by comparing the phase of the signal reflected from the fiber receiver with the transmitted phase. In order to minimize problems with reflections from fiber connections and splices, the transmitting diode is switched off for about a hundred microseconds each machine cycle. During this time, the reflected signal can be measured without interference from the transmitted signal. The phase length of the fiber is controlled through the fiber dispersion by varying the laser wavelength.

Fiber link stabilization system block diagram

PhaseSystem.GIF (34139 bytes) (Canvas 5 document)

 

Tunable Laser Options

The laser at the transmitter must provide approximately one milliwatt of power at 1550nm, with a tuning range of tens of nanometers. Note that the required tuning range is related to the temperature stability of the fiber trenches and is approximately 4nm/C. The line width must be less than about 0.25nm (DiodeSpectralEffects.pdf). The short term center wavelength stability (for timescales of 1 msec to 1 second, the expected feedback timeconstant) must be <5e-4nm. ( or about 60MHz) Note that transmitter powers in excess of a few milliwatts do not improve signal to noise due to non-linear loss mechanisms in the fiber. Several tunable laser options are being considered.

1. Fabry Perot laser diode: These diodes temperature tune. Experiments indicated that the bandwidth of these laser is too wide for use for long fiber runs. In addition it is believed that the multi-line nature of the diode spectrum may prevent the correct operation of the feedback.

2. DFB laser diode: These diodes can also be temperature tuned, although through a smaller range than for Fabry Perot laser diodes. They produce single narrow line outputs which are suitable for use with this system. The system was setup and tested with a DFB laser diode (from Thorlabs). The laser diode was found to mode-hop approximately every degree C of temperature, and was unusable for temperature tuning. Additional work was done with the diode temperature held constant . We are presently continuing work with this type of diode.

3. Grating tuned laser diode: These use a laser diode with an external cavity and diffraction grating tuning. They are commercially available, but expensive, approximately $45K for units which meet our specifications. A disadvantage of this type of diode is its use of moving mechanical parts (the grating) which may not be suitable for long term, continuous feedback.

4. Tunable fiber laser: We are starting development of a tunable Erbium fiber based laser system. This system should provide smooth tunability over the required wavelength range (~50nm). In addition it will have wider bandwidth than the DFB laser, which may reduce the polarization sensitivity of the system. Note that the bandwidth should still be narrow compared with the requirements discussed in  DiodeSpectralEffects.pdf.

5. Micromechanical tuned laser. CoreTech has recently introduced a VCSEL 1550nm laser tuned with a micromechanical mirror. We are investigating this option. Papers on similar design (but longer wavelength) micromechanically tuned VCSELS can be found at Papers.

Receiver requirements

Fiber-optical systems are fundamentally noisy when compared to RF systems. The large total phase noise from a fiber system can be corrected by using a narrow band phase locked loop at the receiving end of the fiber. The requirements on this loop are based both on RF phase noise and on machine protection issues. Phase Noise requirements

System configuration

The timing system will use centrally located transmitters, arranged with a point to point connection to each sector. As the timing system is critical to all NLC operations, a redundant system is proposed: Timing System Configuration. In addition to providing the timing signals for the machine, the timing system provides the RF phase reference for the RF system. The connection between the timing system and the RF system are described in RF Distribution System

The timing system components are described in: Timing System Components

The modifications to the linac timing system required for damping ring operation are described in: Damping ring timing system

Alternate Feedback Systems

A number of alternate fiber length feedback systems are under consideration. Most of these schemes rely on comparing the phase of the signal reflected from the far end of the fiber with the out-going phase.

  1. An additional fiber, 20% of the length of the main fiber is connected in series with the main fiber. The additional fiber is placed in a temperature control oven. The temperature of the additional fiber is controlled to make the combined phase length of the fibers constant. The primary advantage is the elimination of the technically risky tunable laser. The primary disadvantage is the added size, complexity, and cost of the fibers and ovens. In addition, the response time of the oven may make it difficult to close the feedback loop.
  2. A mechanical optical delay line is installed in series with the fiber. The primary advantage is the elimination of the tunable laser. The disadvantage is the potentially poor reliability of the moving mechanical stage. In addition the stage may be bulky and expensive.
  3. Precision trench temperature control: With the fiber phase lengths as inputs, it should be possible to control the trench temperatures to maintain each fiber at a constant phase length. The advantage is the elimination of the tunable laser, or additional fiber length control. The primary disadvantage is the difficulty of controlling the temperature feedback loops due to the long time constants.
  4. Electronic phase shifters in place of the optical length control. The phase shifter would be installed so that the outgoing, and returning signals pass through the same shifter. This eliminates the tunable laser system. The primary technical risk is amplitude sensitivity of the phase shift.
  5. Low temperature coefficient fiber cables (in a linear configuration) can be used. In this case, the phase drift with time may be sufficiently slow to allow the use of beam phase references in each sector. This system is considerably less complex than the other proposals. The primary disadvantage are the high cost of the phase stable fiber, and the unknown rate of phase drift over time. In addition, full re-phasing would be required after a long beam interruption.
  6. Direct RF distribution: This is similar to #5 above, except the phase stable fiber is replaced with coax cable in the (temperature controlled)  NLC tunnel. This system also relies on the electron beam for phasing. This is considered the least technically challenging fall-back option for the timing / phase distribution.

Page by Josef Frisch frisch@slac.stanford.edu 04/22/2002