1. ECE4006 – Final Presentation Group 6, Spring 2003 Gigabit Ethernet Vikas Parekh

2. Group Management Plan Vikas Parekh – Group manager, background research, scheduling and budgets. Theron Davis – Webmaster, chief editor and solder expert. Dominic Saltarelli – Design and testing of the Rx/Tx board. Gautam Khanna – Design and testing of the Rx/Tx board. Akil Sutton – Optical Link Budget and inventory management.

3. Gigabit Ethernet OLB OLB 4 Corners Analysis – Components used in fabrication highlighted

4. Problems with the First Run No decoupling capacitors for the input and output pins of the Transmitter MAXIM chip – design problem. Bad soldering. Too much board surgery – inexperience.

5. Further Problems VCSEL set up backwards – plain dumb. Ground plane was shorted at several joints – again bad soldering. Very poor results. The output is shown here with the mod current at 7 mA and the bias current at 9.38 mA.

6. Second Board Design Layout After debugging our problems with the transceiver board, we realized that most of the problems were mainly design related. Hence, we decided to design a second run in order to get better results. Solving these problems in the second design involved five revisions to the original design and a more careful approach to the soldering of the second run.

7. Issues considered in the 2 nd run The ground plane was separated from all non-ground elements by at least 2mm. The power supply jack was switched so the plug would face away from the board. Three extra capacitors were added to the design, decoupling the input and output pins connected to power supplies on the laser driver chip. Also, the under board power modulation current trace was moved up, providing a straight line of conduction between the laser ground and the laser driver ground. Lastly, shielding was added to the top of the board to restrict crosstalk from the transmitter to the receiver.

8. Progress in the Last week Second run assembled. Extensive Testing of the Transceiver board. Alignment testing. Final Paper completed. Bill of Materials prepared. Final Financial Budget drawn up.

9. Second Run of Rx/Tx board Soldered the second board. Extra care taken to obtain clean soldering joints and avoid damage to the components due to heat of the solder.

10. Presentation of the Results Independently, both the transmitter and the receiver worked. The top picture shows the Rx baseline test with PRBS7. The bottom picture shows the Tx baseline test with PRBS7.

11. Problems with Rx sensitivity The problem lay with the actual design – as opposed to a problem in the selection of components or soldering. The garbled picture depicts the test for Rx attenuation of 5dB.

12. No problems with Tx sensitivity Figure: Tx PRBS7, 0.5 mW output, 5dB attenuation Figure: Tx PRBS7, 0.9 mW output, 5dB attenuation Figure: Tx PRBS7, 0.9 mW output, 10dB attenuationFigure: Tx PRBS7, 1.25 mW output, 15dB attenuation

13. Loopback Test When the loopback test was performed, the output was the laser was reset back to 0.9mW of output, achieved by setting the bias current to 9mA and the mod current to 7mA.

14. Sensitivity/Crosstalk Issues (again) The problem with receiver sensitivity arose when the output of the laser, which was already running at close to eye safe, was attenuated by 5dB. Even the D21.5 pattern showed jitter with this setup, and PRBS7 was unrecognizable. This loss of signal could not be attributed to receiver sensitivity alone, since 5dB of attenuation of a 0.9mW signal should still be robust enough to drive the PD. The crosswalk between the transmitter and receiver was measured by feeding a signal into the transmitter, and reading the output of the receiver without a fiber connecting the laser to the PD. The results show that the interference is not due to crosswalk alone.

15. Continued The lowest frequency signal, D21.5, translates rather cleanly from the transmitter to the receiver. But with higher frequency patterns, more frequencies are being transmitted to the receiver.

16. Gigabit Ethernet Alignment Tolerance Unconnectorized S7912 Rx produced eyes with significant jitter off of the “sweet spot” “Sweet Spot” deviation as Z increases. Increased coupling in the lateral direction possibly on account of fiber orientation and PD positioning Sweet Spot Off Center

17. Financial Budget

18. Bill Of Materials (BOM)

19. Final Gannt Chart

20. Special Thanks TA: Nickolas Kingsley ECE Administrators: Gwen Selena Martin Brooke/ Nan Jokerst