Testing elements in a fast communication channel 100GB/s Final Presentation Spring 2010 Developers: Hanna Alam and Yousef Badran Project supervised by: Yossi Hipsh Technion - Israel Institute of Technology
Feb 23, 2016
Testing elements in a fast communication channel 100GB/s
Final PresentationSpring 2010
Developers: Hanna Alam and Yousef BadranProject supervised by: Yossi Hipsh
Technion - Israel Institute of Technology
Agenda Brief IntroductionProject GoalsTransmitter and Receiver SchematicDesigning the Microstrip LineLosses ConsiderationsTesting the Microstrip LineComponents and Suitable DevicesConclusions and next steps
Introduction
• The Fast evolving high-speed digital systems in today’s technology and in daily life
• Using tools and technologies which we currently possess in order to create a more advanced design
• A method for breaching the 100 GB/s barrier is through channels with lower frequencies
Goals
• The ability to transfer information in a single 100GB/s rate channel between two units located on the same printed board
• Designing and characterizing the microstrip line
• Suggest theoretical and practical way to test the microstrip line in terms of signal integrity and internal losses
General Schematic
Unit1– Transmitter’s side
Sharpening the signal Reducing
tr/tf
Reducing the bit time
to 10ps
Adding Different delay to
each channel
Buffer
Sigma: Adding all of the 4 narrow signals
25GB/s100GB/s
Unit1– Transmitter’s side
• Sharpening the signal to reduce rise and fall time:
• Problem : finding a suitable buffer • In lower frequencies we use step recovery diodes, in higher frequencies the issue requires further researching
Unit1– Transmitter’s side
• Reducing the bit time is achieved by a mixer and a power divider• Delay is required in order to achieve the desired result
• Also requires additional researching
Unit2– Receiver’s side
Equalizer Splitting into 4 channels
Filtering Data Buffers Monostable
100GB/s25GB/s
Unit2– Receiver’s side
• The attenuation of the microstrip line is not uniform. The higher the frequencies the greater are the losses
• Equalizer - to compensate on the different losses preserving SI
• Monostable - to increase the bit time back to the original 40ps state
The µstrip Line
• primary objective is designing a microstrip line on a printed circuit board
• Our focus will be on a 15 cm long microstrip line.
• How to begin our design ?
Designing the µstrip
• First step is finding a suitable dielectric material for high frequency as 100 GHz
• Next, calculating and choosing microstrip line parameters considering technological constrains
• Additional possible modification that can be added
Duroid 5580
RT/duroid 5880 Properties(typical Values)
2.2 Dielectric constant -
0.0009 Dissipation factor -
Isotropic
Uniform over volume
• Duroid 5580 manufactured by Rogers Corporation
r
tan
Modeling the µstrip line
The demand for Uniform EM field in the H«λ µstrip line
“H” is the dielectric height, What is λ?
0 0
0 0
max
1 1 1c
100
0.2 «
r r r r r
c c
f GHzc cm Hf
We chose H=0.127mm = 5 mil
Modeling the µstrip line
• What about conductor’s width, W?
• W is determined by the requirement for a standard resistance of 50Ω
• As shown in the booklet we acquire that: W=15.5 mil
Basic µstrip line
• Summing up basic µstrip line parameters:
Line modification
• Shielded microstrip line:
+ Reduces cross-talk + Significantly improves energy advancement
through the µstrip line
Line modification
• Suspended microstrip
+ Foam-like material dielectric layer+ dielectric constant close to unity + Much lower losses approximately 1:5 ratio
- complicates manufacturing process
Losses Considerations
• Accurate losses equations are complicated, they’re explained in chapter 3 in the booklet
• Approximation:Typical Duroid loss = 0.3 dB/λMicrostrip length = 15 cmλ(min) = 0.2 cmExpected total losses = 22.5 dB
• This graph can be verified by SI simulations
Inte
rnal
Lo
ss
f[GHz]
Testing the µstrip line• We need to measure the losses in the
microstrip line in order to configure the equalizer or the pre-emphasis unit
• 100GHz pulses are hard to create, therefore we can calculate the Fourier series and deal with several lower frequencies individually
• This can be easily created using a simple signal generator
Testing the µstrip line
• A pulse signal can be split to a sum of sinusoidal signals in different frequencies by calculating its Fourier series
example:
Testing the µstrip line
• Multiplying the signal to reach 100GHz frequency
• Running it through the microstrip line• Using a mixer we decrease the frequency to
a measuring level • We repeat the process for additional various
frequencies• Enough points can give the losses vs.
frequencies graph
Testing the µstrip lineFirst schematic for 100GHz components :
Source-1Eva-ADF4350
MultiplierX2
MiteqMAX2M080160
4.16GHz 8.33GHz
MultiplierX12
MillitecAMC-08-RFH00
100GB/s information rate MicrostripWaveguide-microstrip adaptor
Waveguide-microstrip adaptor
100GHz
LecroyOscilloscope
Buffer
Waveguide
100GHz
Ducommun Balanced Mixers
FDB-10-01
12GHz
Source-2Eva-ADF4350
MultiplierX3
Hittite HMC916LP3E 11GHz
MultiplierX8
Millitech AMC-12-RFH00
Buffer
3.667GHz
88GHz
RFLO
IF
Testing the µstrip lineSecond schematic for 50GHz components :
Source-1Eva-ADF4350
MultiplierX4
MiteqMAX4M160169
4.16GHz 16.667GHz
MultiplierX3
Millitech MUT-19
100GB/s information rate MicrostripWaveguide-microstrip adaptor
Waveguide-microstrip adaptor
50GHz
LecroyOscilloscope
(or simple oscilloscope)
Buffer
Waveguide
50GHz
Ducommun Balanced Mixers
FDB-10-01
6GHz
Source-2Eva-ADF4350
MultiplierX4
Hittite HMC370LP4 14.67GHz
MultiplierX3
Millitech MUT-22
Buffer
3.667GHz
44GHz
RFLO
IF
Testing the µstrip lineThird schematic for 25GHz components :
Source-1Eva-ADF4350
MultiplierX3
Hittite HMC916LP3E 4.167GHz 12.5GHz
MultiplierX2
Hittite HMC576LC3B
100GB/s information rate MicrostripWaveguide-microstrip adaptor
Waveguide-microstrip adaptor
25GHz
LecroyOscilloscope
(or simple oscilloscope)
Buffer
Waveguide
25GHz
Ducommun Balanced Mixers –
FDB-42-01
1GHz
Source-2Eva-ADF4350
MultiplierX3
Hittite HMC916LP3E 12GHz
MultiplierX2
Hittite HMC576LC3B
Buffer
4GHz
24GHz
RFLO
IF
Suitable Components• EVA-ADF4350 is the supposed unit to
generate a stable 0.5:4.4 GHz signal for the testing environment
• We deal with this unit as a black box with given parameters
• Other components can be found in our booklet
appendix B
Conclusions & Next Steps• Researching devices to sharpening the signal in
unit-1, and additional units presented in the booklet
• Simulating the microstrip line using 3D simulation software such as “CST”
• Fabricating the design and testing it
• We’ve found various suitable devices that can operate in 100GHz frequency and more
• We have presented a HF medium in addition to testing environment
thank you …
Questions?