Test report on the Detexis optical link
Marc Torres, April,1999

The test was performed in the lab, with the prototypes modules ME3 and RC2, on loan from Detexis. The fiber was a 2-m long monomode jumper, equipped each end with Diamond E-2000 connectors. The tests are focussed on the parameters which are critical for Radioastronomy applications, namely phase stability of CW signals, (for LO uplink) and dynamic range on Gaussian white noise (for IF downlink). The amplitude stability (mainly versus Temperature) is also of interest and will be checked later.

After good thermal stabilisation, the conditions were:
Tx:            +15V, 220 mA
Laser:                   120mA
Rx:             +15V, 65mA
Photodiode: -8V, 2.7 mA

The laser module housing temperature was 34 C


Dynamic range

1/ Distorsion
Bandpass filtered white noise (1280MHz CF, 320MHz BW) with a density of -56 dBm/MHz is sent as input (blue). The link actually acts as an amplifier, with an average gain of 17dB, and a small rolloff (green curve) at the low end. The distorsion generates a noise band at twice the input frequency, caused by assymetrical clipping of the peaks of the gaussian noise. The input level was chosen so that the density of this band is kept 30dB below the useful band.

2/ Noise
The output of the module exhibits three different noise floors: (no signals transmitted)
- Noise floor of the instrumentation (yellow)
- Noise floor when the receiver module is turned on (deep blue)
- Noise floor when the laser module is turned on (light blue)

When the signal is applied the green curve is obtained. For proper transmission of IF signal, its density must be 30dB above artefacts. In the picture above it is 30 dB above 2nd harmonic and 45 dB above noise floor. This allows for increasing the bandwidth (from 320 MHz to 1 GHz, the density must by lowered by 5 dB to keep the total power constant), and for some additional margin in operating level.


Phase noise

1/ Direct measurements
A CW signal at 1.8 GHz was sent thru the link and compared to the same signal connected directly. Generator is a hp8644B and spectrum analyser is RS FSEA30. Both are phase-locked to high-quality 10 MHz external reference.


Far from the carrier (15 kHz in the picture above) very little degradation can be observed. Blue trace is transmitted signal and green trace is the noise floor of the equipment. The noise level is -119dBc/Hz @ 5 kHz, which is typically 10dB better than current Plateau de Bure system.
Close to the carrier ( picture below) the two traces are almost at the same level. The 50 Hz line is an artefact from the synthesizer ( but useful as a beacon).


2/ Long term measurement
A mixer is fed with a CW signal @ 2GHz on the LO port, and the same signal transmitted thru the link is applied on the RF port. A low noise, low frequency amplifier picks up the IF voltage which represents the phase shift of the transmission. A digital scope displays the slow drift of the phase. A sensitivity of 5 millidegrees per division is obtained reliably. Quadrature and calibration are obtained by tweaking the frequency.
The sensitivity to mechanical vibration of the devices is quite high. The receiver module is far more sensitive than the transmitter. The connectors themselves are affected by mechanical stress, but significantly less than expected. The fiber is more sensitive to stretching than to bending or twisting.
Long-term surveillance of the phase drift was performed. The coefficients of sensitivity to power supply of the modules were found to be :
Laser @ 120 mA : 0.02 degree/mA
Photodiode @-8V: 0.71 degree/volt
The 15 Volt supplies do not affect the phase significantly.
The intrinsic noise of the power supplies used was less than 2 millivolts.

The following picture shows a typical 10-second sample of the phase drift. The vertical scale is calibrated to 0.04 degrees full scale.


The Fourier transform of repeated such shots has been observed and does not show any particular frequency, but exhibits a dominant 1/f component.

Conclusion

The prototype link in its present status is directly useable for IF downlink transmission up to several GHz bandwidth in the interferometer. The field reliability and practicality of the connectors in the PdB environment will be checked over next year on station N46. For reliable LO transmission an improvement in stability of a factor 5 seems to be necessary. Thermal effects are time-consuming measurements and have not been tested so far but will be investigated.