Optical Gigabit Ethernet Group F O R E
Optical Gigabit Ethernet Group F. O. R. E. Final Presentation Chris Abbott, Ronen Adato, David Larado 4/21/2005
Overview n n n Purpose was to build physical layer of optical gigabit transceiver Selected active components based on availability, connectorization, quality Designed and laid out two transceiver printed circuit boards Built and tested boards Compiled prototype budget and productionscale bill of materials
IEEE 803. z Standard n n n Extinction Ratio Eye-safe power level Criteria for testing n n Definition of Bit Error Rate Use of eye as guideline, but not formal test
Test Board n Built test board to learn soldering and testing techniques Soldered Test Board Test Setup (To oscilloscope)
Test Board Results No resistance/Full Setup High resistance/Full Setup <-- High resistance / test board only Ideal resistance / test board only --> Ideal resistance/Full Setup
VCSEL Selection n Criteria for selecting VCSEL: n n Good performance (slope efficiency, threshold current) Availability Cost Connectorization (SC over LC)
VCSEL Performance Analysis
AOC vs Lasermate n n Performance Similar Lasermate costs more in small quantities (costs comparable in bulk) AOC is LC connectorized, Lasermate is SC Chose Lasermate due to SC connectorization (no patch cords)
ROSA vs. PD n Past years have used PDs n n ROSA contains TIA inside package n n n Requires external TIA Possible reduction in noise Easier layout Comparable prices, easier to find Leading PD vendor too expensive Chose ROSA
ROSA Selection n Lasermate RSC-P 85 P 416 1. 25 Gb/s SC ROSA n n SC connectorized ROSA Costs: n n 2 -3 = $36. 00/unit >10, 000 = $6. 00/unit
Optical Link Budget
Schematic Design n Schematics derived from: n n n Maxim datasheets Maxim application notes Past group’s work Used three element power supply filters Conservatively coupled Vcc and ground with decoupling capacitors Designed for common anode VCSEL configuration
Transmitter Schematic
Receiver Schematic
PCB Layout Considerations n n Kept traces as short as possible to prevent transmission line behavior Avoided sharp corners on high-speed lines (used 45° angles when needed) Kept high-speed differential lines symmetrical Effectively separated ground plane of transmitter and receiver
Conservative Layout
Aggressive Layout
Conservative Board 1 n n Soldered all components and tested the Rx circuit for a clear eye SMA/QMA fiasco prevented BER testing for the Rx circuit, and any Tx circuit testing Rx Circuit Test Setup Eye Produced by Rx Circuit, Signal PRBS 7
Conservative Board 2 Rx n n Reused SMAs from test board, soldered a new board The Rx circuit failed to produce an eye n Rx Circuit Waveform Measured Before LA It was determined that the LA (MAX 3264) had been damaged Rx Circuit Waveform Measured After LA
Conservative Board 2 Tx n n Soldered board, but had to reuse VCSEL from conservative board 1 Voltage drop across VCSEL was 5 V n n n VCSEL could be damaged VCSEL could be connected incorrectly Removed VCSEL and checked behavior by pin-pin resistance
Aggressive Board 1 n Decided to try Emcore VCSEL on aggressive design to get Tx working n n Aggressive Tx very similar to conservative Tx Made sure we knew how Emcore fit into board
Aggressive Board, Tx. Rx We varied the attenuations with, the transceiver: 4 d. B 7 d. B Using different combinations of the optical attenuators, we were able to try these also: 9 d. B 12 d. B
Aggressive Board, BER n n BER of less than 10^-9 when each signal was tested PRBS 7, at least 5 minutes test runtime
Financial Budget
Bill of Materials (BOM)
Gantt Chart
Future work n n Get Lasermate VCSEL working Write final report
Conclusions n n Project was a success Learned about entire design process
- Slides: 28