Transcript Final Presentation

Final Presentation
ECE 4006 C
G3: Karen Cano, Scott Henderson, Di Qian
April, 23 2002
I: Project Tasks and Theory
• 1. Research on the transmitting and
receiving modules.
• 2. Examine the testing board
• 3. Search for the components
• 4. Testing the evaluation board with
purchased components
• 5. Connecting the purchased components
with parts from other groups.
Project Goal
• Duplicate the data transmitting and
receiving module functionality of the
Gigabit Ethernet technology with purchased
components that provide optimum
performance at a minimum price.
Possible Solutions
• Transmitting module (laser source)
– VCSEL
• Receiving module (Photo-detector)
– PIN photodiode
• Other Specs:
- SC connectorized (optical)
- SMA connectorized (electrical)
- 850nm
- Multimode (fiber)
- relatively low cost
Laser Basics
• What is a Laser?
– Light Amplification by Stimulated Emission of
Radiation
• How?
1) Electrons in low-energy levels bumped into
high levels by injection of energy
2) When an electron drops to a lower energy
level, excess energy is given off as light.
VCSELs
• Vertical Cavity Surface Emitting Lasers
• Physical makeup
– Bragg mirrors
– Active region
• Fabrication techniques
– Molecular beam epitaxy
– Vapor phase epitaxy
VCSELs
• In EELs no pre-cleaving tests can be
performed, testing VCSELs is much
cheaper
• Less current required for VCSELs
• Output beam easier couple into fiber and
much less divergent than EELs
• Smaller and faster than EELs
VCSELs vs. EELs
• Edge Emitting Lasers - give out their light
from the sides or edges, therefore no precleaving tests can be performed
• Since VCSELs emit light from the top and
bottom, they do not have this problem.
Testing them is much cheaper
Multimode
• Multimode- light is injected into the core
and can travel many paths through the cable
(i.e. rattling in a tube).
• Each path is slightly different in length, so
the time variance this causes, spreads pulses
of data out and limits the bandwidth.
Singlemode
• Fiber has such a narrow core that light takes
one path only through the glass.
• Not limited to modal-bandwidth.
• Very small amount of pulse-spreading is
consequential only in Gigabit speed
applications.
Photodetectors
• Necessary for light pulse detection
• Wide variety of of types
–
–
–
–
Photoconductors
Avalanche photodiodes
PIN photodiodes
MSM photodiodes
Avalanche Photodiodes
• Exemplify the “gain-bandwidth” tradeoff
• Use the p-n junction model to operate
• Take advantage of the avalanche effect
–
–
–
–
Carrier multiplication
Associated gain
Time constant associated with avalanche
Bandwidth penalty
PIN Photodiode
• PIN
–
–
–
–
Reason for name
Doped region, undoped region, doped region
Unity gain
Functions under reverse bias
• Most important parameter for operation
– Transit time
Bandwidth vs. Depletion Width
• Transit time
– Time for subatomic
particle to get from one
electrode to the other
• Based on quickest,
typically electron
– e- mobility > h+ mobility
• Capacitance limited
Transit Time (continued)
• Dependence on intrinsic
region length
• Minimizing this region
• High bandwidth
applications
MSM Photodiode
• Metal-Semiconductor-Metal
– Associated work functions
– Atomic level metal-semiconductor marriage
• High speed (up to 100GHz)
• Majority carrier devices
• Not developed for Gigabit Ethernet on scale
as large as PIN
II: Design Overview
Keeping track of Amps & Watts
Design Specifications (Lossless System)
Wavelength 850nm
SC Connectorized
TX & RX
DC Bias
TX DC Bias = 0-300mA
Modulation Current = 30mA
RX DC Bias = 80 uA
VCSEL
Threshold Current (Ith)
Slope Efficiency (mW/mA)
DC Bias =
1.2 X Ith < TX DC Bias
PD
Responsivity
Current into RX =
Power Emitted X Responsivity
Power Emitted =
DC Bias X Slope
Signal Specification Overview
Honeywell VCSELs
•
•
•
•
•
•
HFE4380-521
Slope Efficiency 0.04 mW/mA
I threshold - 1.5 - 6 mA
HFE4384-522
Slope Efficiency 0.15 mW/mA
I theshold - 1.5 - 6mA
Link Budget
• Analysis of current and power
throughout the system, starting at the
TX and ending at the RX or transimpedance amplifier.
• Losses
– Not all of the light emitted by the VCSEL will reach the
PD.
– Losses are incurred from the fiber and the SC
connectors.
At the PD Side
• Once the losses have been calculated into
the output power range, this new range of
power is to be converted back into current.
• When the power or light hits the PD, it is
multiplied by the responsivity of the PD,
expressed in A/W.
• This value is the current coming out of the
PD and into the trans-impedance amp.
At the RX Side
• The current coming out of the PD has to be
large enough to drive the trans-impedance
amp, which takes at least 80 uA.
Honeywell Option #1
•
•
•
•
•
•
•
•
•
•
•
•
Ith = 6 mA
DC bias of laser = 6 (1.2) = 7.2 mA
Slope efficiency 0.04 mW/mA
Power output at DC bias = 0.04 * 7.2 = 0.288 mW
Max TX modulation current. 300 mA
Power output at TX modulation current = 04 * 30 = 1.2 mW
Range of emission of light coming from the laser (lossless) =0.288 - 1.2 mW
Losses3dB or 1/2 output power
Range of emission of light coming from the laser (with losses)=0.144 - 0.6 mW
Responsivity of lasermate's PD
= 0.35 A/W
Min. current from PD = .000144 W * 0.35 A/W= 50.4 micro Amps
Max current from PD = .0006 W * 0.35 A/W =
210 micro Amps
• Table 1. Link budget for low slope efficiency VCSEL
(HFE4380-521).
Honeywell Option #2
•
•
•
•
•
•
•
•
•
•
•
•
Ith
6 mA
DC bias of laser = 6 (1.2) =7.2 mA
Slope efficiency 0.15mW/mA
Power output at DC bias = 7.2 * 0.15 =1.08 mW
Max TX modulation current. 300 mA
Power output at TX modulation current = .15 * 30 =
4.5 mW
Range of emission of light coming from the laser (lossless) =1.08 - 4.5 mW
Losses3dB or 1/2 output power
Range of emission of light coming from the laser (with losses)=0.54 - 2.25 mW
Responsivity of lasermate's PD =
0.35 A/W
Min. current from PD = 0.00054 W * 0.35 A/W = 189 micro Amps
Max current from PD = 0.00225 W * 0.35 A/W= 787 micro Amps
• Table 2. Link budget for high slope efficiency VCSEL
(HFE4384-522).
In Conclusion: Purchasing Choice
• Prefer VCSEL with higher slope efficiency
because it can drive the trans-impedance
amplifier. More importantly, it can do this
with the same amount of input current that
is needed for the other VCSEL.
• If the price difference is not too significant,
this VCSEL is the most reliable option.
III: Construction and Testing
• Interface optical components with Maxim
• Extensive testing to meet standards
• Develop more optimal design utilizing
newly ordered VCSEL and PD
• Repeat testing procedures
Circuit Layout for PD
• Optical connection on top, (SC - Light In).
Electrical connections on left and right (SMA)
• R=1/(2*pi*f*CDET), where CDET = 1.5 pF (from
spec. sheet). So, R = 53.1Ω.(impedance of PD).
• Capacitors were chosen based on their frequency
response at 1.25 GHz. From Murata’s site: C =
.01uF.
Circuit Layout for the VCSEL
• Electrical connections on the left (SMA - In),
optical connections on the right (SC - Light Out).
• Constraint: Circuit’s RTOT has to be 50Ω, due to
equipment requirements.
• VCSEL’s RTYPICAL=25Ω (from spec.sheet)
• So, a 25Ω resistor was placed in series with the
VCSEL.
Constructing the VCSEL Board
• Radio Shack purchased “through” boards
• SMA connectors, SMT components, and
VCSELs
• Impracticalities of initial design
• Remedy for small holes and lack of a drill
• Problem: GTS-1250’s outputs are ACcoupled
• Bias-T applied and shown in new circuit
VCSEL Testing Procedure
• GTS-1250
• New VCSEL
board
• Old opto-board
• Scope
• Electrical
connections and
optical loop
Testing Results
• IEEE 802.3z standard
mask application
• Bit error rate settings
in scope
• New VCSEL vs. old
VCSEL
• Failure of complete Tx
module thus far
VCSEL PCB Design
• Five components:
- SMT Resistor,
- LED,
- Power Connector,
- 2K ohms Resistor,
- SMA Connector
VCSEL PCB Design (Con’t)
•
•
•
•
SMT Resistor – 0.1 inch
LED – Two Leads
2K Ohms Resistor – 0.5 inch
Power Connector – hole 0.2 x 0.1 inch
The Complete PCB Design
(Left: PD Circuit, Right: VCSEL Circuit)
PCB Dimensions
• SMA Leads – 0.045”X0.045”, Pad .09”X0.09”. (given in the spec sheet)
• VCSEL and PD holes – 0.02”X0.02”
• Power Connectors – 0.125”X0.125”, Pad –
0.25”X0.25”
• SMT – 0.01”
• Resistor – Template in Library
The Completed PCBs
Top Layer and Bottom Layer
Adjustment to the PCB
The SMA hole diameter was increase to 0.06-7” by using a
razor
Optimizing Circuit Design
• Minimizing inductors
and component
quantities in general
• Utilizing existing
schematics and
general EE knowledge
• Soldering followed by
repeated testing
Intel/Agilent PC Interface Card
IV: Construction: Photodetector
(Connectorized)
• Initial “through-component”
board
• Connectorized OSI PD
• Standard SMT components
• SMA connector and DC bias
via 5V connection
• PD maximum solder
temperature of 500oF for
total time of ten seconds
V: Testing: Photodectector
(Connectorized)
• Lack of detectable eye
• Simple signal analysis
• 50% duty cycle,
1.25GHz square wave
• Signal averaging
functionality of scope
• Comparison to input
• Weak output (top)
Testing (cont’d)
• Fourier analysis via
oscilloscope
• Corresponding frequency
concentrations, but
apparent degradation
• Averaging functionality in
combination with Fourier
• Input and output (bottom)
still weakly associated
with simple square pattern
Testing (cont’d)
• K28.5 (pseudorandom) bit
test using Fourier
• Averaging still turned on
• Further degradation of
signal with large amounts
of stray artifacts
• Corresponding
frequencies still appear to
be present
Testing (cont’d)
• Drastic measures to
ensure PD board
integrity
• Fiber input to PD
disconnected and new
signal compared
• Small amount of
change in Fourier,
zoomed in farther
Testing (cont’d)
• Scope (BNC)
connection removed
• New signal compared to
previous noisy Fourier
• Signal dissipated to
nearly zero, proved that
scope was not causing
all noise
Possible Problems at this Stage
(1)
• Ascertained that the noise was coming from
PD board itself
• Possibly slow or defective PD, inadequate
board or construction
• Process repeated, similar results
• Circuit design errors unlikely
VI: Construction: Photodetector
(Unconnectorized)
• Initial “throughcomponent” board
• Unconnectorized
Hamamatsu PD
• Standard SMT
components
• SMA connector and DC
bias via 5V connection
• PD maximum solder
temperature of 500oF for
total time of ten seconds
VII: Testing: Photodetector
(Unconnectorized)
• “X-Y-Z” stage setup
– Cleaved fiber (emission)
– SC connector (other end)
• Test setup identical to
previous connectorized test
sans the “stage”
• Simple square wave input
• Fine adjustment of fiber-vise
• No apparent output
Possible Problems at this Stage
(2)
• Previous semesters board tested
• Satisfactory on simple pattern, not on
K28.5
• Power connector on old board fickle
• Possible broken solder joint on capacitor
(VCSEL side)
• Incorrect construction unlikely, but
inadequate construction possible
Concluding Remarks
• Insight into opto-electronics gained
• Work experience in team environment
• Soldering, test equipment, and test
methodology learned
• Report writing and presentation experience
for future job use