NLC Timing System Initial Experiments

Experiments performed summer 1998

Overall Note: These experiments were performed with a Fabry Perot laser diode. This type of diode is now considered unsuitable for use in this system due to its large optical bandwidth.

Experimental Results - Phase Noise:

A fiber coupled laser diode (non-DFB) at 1550nm and a fiber receiver were used to test the phase noise and temperature tuning of a simple fiber link. A 2.2Km length of single mode fiber was used for the experiment. A modulation frequency of 120MHz was used due to the bandwidth limit of the fiber receiver. Optical power on the receiver was about -8dBm, similar to the levels used for the NLC timing system. Phase noise was measured with a spectrum analyzer.

Measured phase noise: -117dB/Hz at 100KHz from carrier.  -110dB/Hz at 5KHz from carrier (limited by instrumentation).

The expected spectral noise from the fiber link is flat. The noise level determines the minimum allowable integration time constant for the phase lock loop in the receiver. In order to reach the required -80dB (integrated) phase noise on the RF (corresponds to 1 degree X-band with a 120MHz carrier), the integration time constant for the PLL must be longer than about 3KHz. The resulting phase noise requirements on the PLL VCO can be met with a standard quartz oscillator. This demonstrates that the fiber link has sufficiently low phase noise for this application.

Experiments done at 622MHz with a detector similar to the type intended for use for the NLC, with an optical power of about -8dBm, and 2.2Km of fiber gave 0.75degrees X-band in a 1KHz bandwidth. Noise power was fairly flat with frequency.

Experimental Results - Bias current wavelength tuning:

A network analyzer was used to measure the phase delay through 2.2Km of optical fiber while the bias current in the laser diode was changed. A test with a short fiber was performed to check for phase shifts in the diode itself. Approximately 10psec/Km (about 40 degrees X-band / Km) were measured for a range from minimum to maximum bias current. This is sufficient to compensate for approximately 0.3 degrees C change in fiber temperature. No data was obtained on the response time of the wavelength to bias current changes.

A mixer was used to measure the change in phase vs. bias current (measured at 100MHz). A change of 27ps/Km was measured for the full range of bias current.This result is believed to be more accurate than the previous result.

Experimental Results - Temperature wavelength tuning:

The laser temperature was changed in a un-controlled manner with "freeze spray". For a temperature change on the order of 25 degrees C, a change of >10 degrees of 357MHz was observed. This corresponds to (very roughly) >3psec/C/Km. With a 50 degree C temperature tuning range, this will compensate for >5 degrees C of fiber temperature. Note that this result is very preliminary, more accurate results will be obtained with the fiber transmission test system. There was some evidence that the tuning Vs temperature curve was non-uniform. That, in itself, is not a serious problem, however if the tuning curve is discontinuous a different fiber length compensation system will be needed.

A mixer was used to measure the diode temperature effect on transmission time through a 2.2Km fiber. 3ps/Km/C was measured. Some possible phase jumps were measured (near the detection limit) at .03ps/Km. This would correspond to .5psec over 15Km, or 2 degrees X-band. Tuning was measured over a 35 degree C temperature range.

Experimental Results - Fiber length temperature coefficient:

The phase length of the fiber was measured over temperature. A change of 40ps/Km/C was measured.

Experimental Results Summary Table

Summer Science Student paper on fiber experiments - Johanna VanGiffen (word 97 format) Report.doc

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