No abstract

No abstract

No abstract

the transportation and intercomparison uncertainties. The first may be accomplished by more frequent and more carefully studied voltage transfers or by performing the experiments at the site of the NBS voltage standard. The latter, however, does not solve the problem of disseminating the volt to other experimenters, in particular to other national laboratories such as the National Physical Laboratory (Great Britain). With regard to the second uncertainty, we note that the uncertainty associated with our own intercomparisons on our local voltage standard is~0.005ppm. This suggests that significant improvement on the 0.2 ppm quoted by NBS for cell intercomparisons is possible. In any case, the present experiments indicate that the ultimate limit on the absolute accuracy of the ac-Josephson-effect determination of e/h is the accuracy with which the primary voltage standard can be established, maintained, and referred to. We thank B. N. Taylor for his encouragement, RCA for supplying evaporation masks, and W. J. Hamer for arranging the calibration of the standard cells. One of us (A.D.) thanks the National Research Council for a NRC-NAS-NAE Postdoctoral Research Associateship which he held at NBS during an intermediate stage in the planning of the present experiments.

The integration of Er-doped optical amplifiers has been explored for several years with the goal of being able to compensate for splitting losses in passive components without the necessity of getting off the chip. However, previous results on Er-doped planar optical waveguide amplifiers (POWAs) have suffered from a limited gain spectrumi'l, large area coverage121, and incompatibility with conventional waveguide fabrication technologyi31. Here we report on the performance of an Er-doped aluminosilicate POWA with a broad gain spectrum integrated with our standard Silicon Optical Bench (SiOB) waveguide techno log^^^^.To make the POWAs, E?'-doped films of aluminosilicate glass were RF-sputtered on 5" oxidized Si substrates. Waveguide cores were defined using conventional lithography and etching. Finally, our standard silica upper cladding was deposited by LPCVDf41.coupled directly to a high numerical aperture fiber that was fusion-spliced to a standard single-mode telecommunication fiber. The insertion loss at 1.3 pm was 5.5 dB and 7.1 dB for the 7-cm and 14 cm devices, respectively. The typical fiber-waveguide coupling loss was 1.25 dB/interface. An additional 0.8 dB loss was incurred in each pigtail, which included the splice between the dissimilar fibers mentioned above and the FCFC connector.The small signal gain at 1549 nm versus input power at 980 arid 1480 nm is shown in Figure 1 for a 14-cm-long POWA. At 980 nm, a net gain of 15 dB with 150 mW of pump power and a threshold of 22 mW was achieved. At 1480 nm, a net gain of 9 dB at 100 mW and a threshold of 16 mW was measured. The gain at 980 nm seems to be limited by the available pump while the gain at 1480 nm reaches saturation at about 80 mW. gain spectrum is expected in a glass with such a high aluminum content. At the peak (1532 nm), a gain of 22 dB is recorded. From 1540-1 560 nm, only 1.4 dB gain variation is measured, similar to what is observed in non-gain-flattened Er-doped fiber amplifiers (EDFAs)IS. The gain spectrum closely resembles the ASE spectrum in this device. Noise figures derived from direct measurements of the ASE are in good agreement with noise figures obtained from the Bit-Error-Rate (BER) measurements described below.The gain saturation curves and noise figures at 1549 nm for 980 nm and 1480 nm pumping are shown in Figure 3. There is less gain saturation for 1480 nm pumping than for 980 nm pumping. That is partly due to the lower gain at 1480 nm, and partly due to more efficient use of pump photons at this wavelength. With 980 nm pumping, a 3 dB compression level is achieved at 0 dBm output power with 130 mW of pump power. At low input signal levels, a noise figure of 5.3 dB is measured. With 1480 nm pumping, the 3 dB compression level increases to 5 dBm output power with only 1 14 mW of pump power. In this case, a noise figure of 6.1 dB was measured at low input signal levels. As expected due to lower inversion, the noise figure is higher than for 980 nm pumping.The packaged POWA was tested as the first stage of an optical preamplifi...