PULSE

EEWeb

Issue 36 March 6, 2012

EEWeb.com

Dr. Jos Fernndez Villaseor

Freescale Semiconductor

Electrical Engineering Community

EEWeb
Electrical Engineering Community
Contact Us For Advertising Opportunities

advertising@eeweb.com www.eeweb.com/advertising

1.800.574.2791

TA B L E O F C O N T E N T S
Dr. Jos Fernndez Villaseor
Freescale Semiconductor
Interview with Dr. Jos Fernndez Villaseor - Medical Product Manager

4 9

TABLE OF CONTENTS

Telemonitoring Solutions to Prevent Chronic Degenerative Disease Complications

BY DR. JOS FERNNDEZ VILLASEOR
How Telehealth Monitoring Systems help health care providers adequately monitor patients with chronic degenerative illnesses.

Featured Products Repeaters: Learn to Love em

Michael Steinberger explains why digital repeaters will become present in every stationary system and demonstrates the type of analysis required for their design.

12 14 20

BY MICHAEL STEINBERGER WITH SISOFT

Would You...Could You...Should You... Compile Your FPGA Design on the Cloud?
With Cloud storage technology becoming more omnipresent, Phil Simpson weighs the pros and cons of using the Cloud for your projects.

BY PHIL SIMPSON WITH ALTERA

RTZ - Return to Zero Comic

22

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

INTERVIEW

Dr. Jos Fernndez

FEATURED INTERVIEW

Villaseor
How did you get into electrical engineering and when did you start? During high school, I loved all the science classes such as physics, math and biology. In my third year of high school, I chose to focus on physics and mathematics. The advanced physics, electronics and calculus classes were amazing because they let me use my imagination and create things. At the same time, it was almost like playing with the circuits and applications. After that, I decided to pursue a career in electrical engineering. Even though I enjoyed electronics, I always wished there was a medical/ biology group available when I was in high school. Can you tell us about your work experience/history before becoming the Medical Product Manager at Freescale? Before joining Freescale, I worked as a field application engineer for an electronics design house focused on health and automotive applications. I also worked in a clinic for Mexicos Health Ministry where I provided preventative and treatment medicine to high-risk, low-income communities. I did research on protocols for
Dr. Jos Fernndez Villaseor - Medical Product Manager

Freescale Semiconductor
renal and liver donors with the transplantation unit and worked with the internal medicine and gastroenterology department at a non-profit, public hospital in Guadalajara. For the last 13 years, Ive also taught courses in electronics, medicine

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

INTERVIEW
and surgery at various universities in their electronics and biomedical departments. What have been some of your influences that have helped you get to where you are today? While at the university, I earned a scholarship to study Japanese in Kyoto. This experience was a crossroad in my life. I learned how to improve my time management skills, perfect and master my daily work, and ultimately decide that I had just enough time and energy to push myself to the limit and pursue an additional career. When I was in my fourth semester of electronics engineering, I started my first semester of medical school. The universities were far apart, and my days were longstarting at 5:00 a.m. and finishing around 1:00 a.m. This discipline has continued to motivate me today as a practicing surgeon and as an electronics engineer for Freescale. Have you always been interested in practicing medicine? Yes. During my childhood, I visited hospitals frequently and was thrilled by the work the physicians did. However, I always thought I had a bad memory and at some point I decided that medicine was not for me. As it turned out, I was not as bad as I thought at memorizing information. What made you decide to study both electronics and medicine? It was always in my mind. I enjoyed both things, so why not study both? Why not apply technology to medicine to fully understand the needs of both the market and the user (patient)? This way, we can offer better solutions and improve the health of everyone. Can you tell us more about aesthetic medicine? Aesthetic medicine is focused on improving the human bodys aesthetics and helping in all antiaging medical treatments. Reasons for this could be due to accidents, and diseases such as vitiligo, acne, or just because patients do not feel comfortable with the way they look. What type of work do you do with this area of study? Some of the medical and surgical procedures we do are to correct or improve these conditions. One of the things I also focus on is medical sport supplementation. With that, my major objective is to improve an athletes performance. For this we learn how to detect and correct muscular, neural and joint imbalance. What are your favorite hardware tools that you use? I really enjoy using the Freescale Tower System development board, which is a designers platform to easily prototype and test home portable medical equipment. The best part about it is that you can just plug in the boards you need, such as the serial or LCD boards, and you dont need a hardware design at all. You just use them as stackable boards. What are your favorite software tools that you use? CodeWarrior, Micrium RTOS, Mathlab and IAR. What is on your bookshelf? I like to read non-technical books so that I can rest my mind from the things that I read on a daily basis at work. I am reading Pedro Paramo by Juan Rulfo, a Mexican authors short novel about death, and Genji no Monogatari from Murasaki Shikibu, a Japanese lady from the court, which is sometimes called the worlds first novel. But besides that, I really enjoy writing. I have just recently coauthored a book based on Micriums

FEATURED INTERVIEW

During my first year at engineering school, I kept blowing out the capacitors. Electrolytic capacitors had a figure similar to a 1 which marked the leg that was supposed to be connected to the ground. I kept saying to myself, This must mean the digital 1 so it should go directly to Vcc. Well, after seven or eight capacitor explosions, I found out it didnt!

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

INTERVIEW
RTOS with specific applications for medical devices. In this book, we focus on home portable medical devices, explain the basics of physiology and how easy it is to create a device using Freescales microcontrollers and sensors. I have also published some articles in Germany, Korea, China, USA, Japan and Brazil about how to use technology and medical devices to monitor and prevent complications in chronic degenerative and cardiac diseases. In this way, I focus on what is needed from the patients perspective to improve his or her quality of life, rather on what the technology can do by itself. I also participate in Freescales blog Medical by design, where we talk about how to solve real medical problems like avoiding acute complications of chronic degenerative diseases, Parkinsons disease, emerging technologies and others. If you would like to know more about this and others, take a look at the Medical by Design blog from Freescale. What has been your favorite project? One innovative project Ive been involved with is an emotion sensing application for consumer and automotive applications . In this project, we tried to capture how emotions are created and felt by humans. The emotions and the way we react to certain stimuli evoke a cascade of brain reactions, mostly triggered by our unconscious zones. This produces specific neurotransmitters that are released through the brain and the bloodstream and produce specific responses in our body. We can try to fool ourselves by saying we dont feel scared, but our body says the contrary. Picture yourself jogging when suddenly a dog attacks you; your heart rate will increase, your muscles will respond to move away, your face will be pale so that all the blood is available for you to run away! In our reference design, we use the most common responses induced by hormones, such as sweating, heart rate and muscle contraction among other variables to detect emotions in the user. Imagine this concept being applied to a steering wheel. It would be able to detect a heart attack or fainting and stop the car, park it and call 911! Another one of my favorite projects is the design of field effect transistors for specific medical needs like detecting biochemical compounds that help provide early diagnosis of a wide range of pathologies. Do you have any note-worthy engineering experiences? I was awarded the Best Intern Award during my last year at medical school, and I graduated with honors at my residency. I have also been granted the Microcontrollers Solutions Group Excellence award for the research and development and enablement work for the medical market at Freescale. Do you have any experiential stories you would like to share? During my first year at engineering school, I kept blowing out the capacitors. Electrolytic capacitors had a figure similar to a 1 which marked the leg that was supposed to be connected to the ground. I kept saying to myself, This must mean the digital 1 so it should go directly to Vcc. Well, after seven or eight capacitor explosions, I found out it didnt! What are you currently working on? My colleagues and I are working on achieving the lowest power modes for microcontrollers so that home portable medical devices can run for longer periods of time and monitor the patient using battery operation. This includes transistor-array specifics for the medical market such as powerful measurement enginesso that medical sensors can be easily instrumented. As a Medical Product Manager at Freescale, what type of work do you do? I participate in the market analysis, product definition and conception and launch of microcontroller products, enablement tools design, demos and reference designs, as well as customer and product support throughout the product life. How does Freescale continue to be a global leader in embedded processing solutions, advancing the automotive, consumer, industrial and networking markets? Freescale is a global leader in the design and manufacturing of embedded semiconductors for the automotive, consumer, industrial and networking markets. The company is based in Austin, Texas, and has design, research and

FEATURED INTERVIEW

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

INTERVIEW
development, manufacturing and sales operations around the world. In the medical market, Freescale develops innovative embedded technologies ranging from microcontrollers and sensors to analog and wireless products that help device manufacturers achieve major advancements in nextgeneration medical and healthcare applications. These embedded technologies are ideal for use in health and wellness, home portable diagnostics and therapy, and medical imaging devices. What direction do you see your business heading in the next few years? Diagnostic and therapy devices will need to become more portable and easier to operate so people can quickly detect any signs of disease from their homes and diagnose treatable diseases early before it is too late. For this, we need to develop security for data transmission to health providers, establish appropriate communication protocols and improve low power/ low cost devices to be defined as standards. This work has already started, but a lot more needs to be donenot only on the technology side but also with regard to patient education. What are some of your hobbies outside of work and design? I love animals and am a huge supporter of animal shelters. I have adopted five dogs that I enjoy walking and taking on hiking adventures. I love to cook and went through master chef training for two years at a culinary school. I like experimenting with ingredients and different styles of food, and I enjoy having friends over to cook for them.

FEATURED INTERVIEW

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

Avago Technologies Motion Control Solutions

Worlds Smallest Miniature Reflective 3-channel Encoder

Features 3-channel encoding (AB and I) Miniature size 304 LPI Advantages Index Signal I Surface mount leadless package: 3.95 mm (L) x 3.4mm (W) x 0.95mm H) High encoding resolution Benefits No need for separate components to generate the index signal Ability to fit into miniature motor designs Various CPR capable by adjusting to the matching ROP of the codewheel Base CPR resolution can be interpolated by end user Corresponding high RPM performance with increased frequencies Catering for various user gating requirements Covering consumer, commercial and industrial applications

Avago Technologies AEDR-850x three channel reflective encoders integrate an LED light source, photo detector and interpolator circuitry. It is best suited to applications where small size and space matters! Applications include medical hand held devices, camera phones, wheel chairs, actuator, vending machine applications, just to name a few.
To request a free sample go to:

Built in Interpolator of 1x, 2x, and 4x High operating frequencies: 55 kHz at 1x interpolation Index gating -20C to 85C

1x, 2x and 4x via external pinouts Operating frequencies can be increased by external interpolator pinouts by maximum of 4x Options available for both gated and ungated versions Industrial application capable

www.avagotech.com/motioncontrol

All rights reserved.

PROJECT

Telemonitoring Solutions to Prevent Chronic Degenerative Disease Complications

By Dr. Jos Fernndez Villaseor
Aging Population and Chronic Degenerative Diseases The average age of the American population is continually increasing. Baby boomers are now becoming our senior citizens, and with this, drastic changes in our health system are necessary. According to the World Health Organization (WHO), non-communicable diseases account for nearly 50 percent of the global burden of disease. Among them, the most common are chronic degenerative diseases such as cardiovascular diseasemore specifically hypertension, which plagues roughly 600 million peopleand metabolic diseases like diabetes, from which roughly 90 million people suffer. With so many people suffering from these diseases worldwide, the ability of healthcare providers to adequately monitor their patients has become a major issue. Telehealth monitoring systems are providing the solution. Telehealth Monitoring Systems Help Prevent Acute Complications of Chronic Degenerative Diseases Telehealth monitoring systems use telecommunications to collect information regarding patients vital signs, which is then relayed to a remote healthcare provider for further analysis. The systems transmit data such as a patients glucose level, heart rate and blood pressure. They can also remind patients and healthcare providers of the proper time to take or administer a medication. The system can be customized to acquire different data related to a patients respective treatment. The ability of a healthcare provider to remotely monitor a patient helps prevent acute complications relating to a patients condition because the healthcare provider can immediately receive data that helps track the evolution of a disease or a post-operational treatment. For efficiency purposes, these systems are mostly developed for use by the patients themselves. They guide the patients through the process of measuring vital signs using a rich graphical user interface. This article

FEATURED PROJECT

Telehealth
MCU AC Mains or Battery Optional Peripherals Analog Sensors USB PHY or RS232 xcvr or Enet PHY PC/Broadband or POTS connection Speaker Display MCU/MPU
(WiFi, Zigbee, Bluetooth)

Keypad

Voltage Regulation

RF Transceiver

IR Interface

Figure 1: Telemonitoring System

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

PROJECT
provides an overview of how a custom-built, diseasespecific telehealth monitoring system is implemented in a home environment using the Freescale Solution Enablement Layer (SEL) for increased portability across Freescales 32-bit architectures such as i.MX, Power Architecture Technology and the ColdFire Family of processors. Freescale Solution Enablement Layer The Freescale Solution Enablement Layer (SEL) is an embedded software platform running with standard operating systems such as Linux and uCLinux to provide application framework capabilities and abstracted hardware drivers (called Services). The SEL is designed to support a compile and deploy model of software reusability across a range of Freescale 32-bit processors. The SEL Service is the primary abstraction mechanism designed to allow the partitioning of applications into software components that are hardware specific. By writing services for specific hardware, the application source code contains control and consistent behavior without being tied to a specific processor. Moving from one platform to another becomes as simple as partially re-implementing the service for the new hardware device. Moreover, services are RTOS agnostic and can be shared by multiple applications. Services are designed to be reusable between applications, and indeed, suites of services can be provided by Freescale or by third parties to eliminate the redundant portions of software solutions while still getting the most from specific processor capabilities. Application Frameworks and SEL Services Are the Primary Elements of the Solution Enablement Layer (SEL) Technology. 1. Application Frameworks: Define pre-validated application frameworks that exist for rapid prototyping and application development. Most application frameworks are suites of C++ classes designed to interact and define consistent application behavior, look and feel, and often implement a rich user interface. 2. SEL Services: A mechanism of application partitioning is available for the express purpose of partitioning software components from the underlying hardware design such that applications need to only be recompiled to migrate from one platform (hardware and RTOS) to another. Conceptually, the SEL is an extension of the operating system running on the embedded processor that allows another level of application abstraction. SEL Services are therefore sub-components of the application that are not operating system specific, and can be shared. SEL Services SEL Services are central components in a software solutions implementation. As applications begin to use the SEL, they can start by only abstracting a single service for a particularly complex piece of hardwarespecific code, while leaving the rest of the application directly calling the operating system. Over time, more and more of the functionality can be divided into services so that the application code becomes more and more abstracted from the hardware without losing any of the underlying hardware functionality. This process of gradually converting to the SEL allows the conversion to take place over the life of one or more projects. SEL Services Properties Dynamically loaded at runtime Interface is directly useable from within an application or the command line Applications do not compile or link to services to use them. Insulate the application against both OS and HW differences. Extended services may derive functionality based on existing SEL services Telehealth Monitoring System Services Working down from a software solution through a hardware implementation in a telehealth monitoring system, developers want to write application code that is easy to migrate among hardware device implementations and RTOS platforms. The Solution Enablement Layer allows applications to be segmented to define graphical user interfaces (GUIs) independent of SEL services: Main control with personalized items for vital signs on patient GUI Blood pressure with symptoms for acute

FEATURED PROJECT

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

10

PROJECT
complications GUI Glucometer with symptoms for acute complications and prevention of double intake of dosage GUI Pulse oximeter for Chronic Obstructive Pulmonary Disease GUI The applications are also allowed to define a service of the application that is hardware or RTOS independent, such that re-implementing part of the service allows the user easy migration across the 32-bit Freescale portfolio: Blood pressure Service (systolic, diastolic and mean arterial pressure) Glucometer Service Pulse Oximeter Service Thermometer Service (infectious disease complications) Digital Weight Scale Service (for monitoring water retention in patients with congestive heart failure) Software partitioning a telemonitoring system application might have multiple services running in a high-end processor such as i.MX, Power Architecture Technology or ColdFire Family, or the application might be tailored to implement a couple of services in a low-end processor. Figure 2 depicts a comprehensive telemonitoring system using several SEL medical services. This same system can be partioned to target other Freescale 32 bits processors Conclusion Modern American society faces many public health issues with the rapidly increasing average age of the population and the pathology demographics. This means that for people to age independently, it is important for healthcare providers to be able to adequately monitor vital signs and drug intake from a distance. Taking this into account, Freescale offers advanced hardware tools and a new software platform (Solution Enablement Layer) to the community of medical equipment designer and OEMs, which enables concurrent software and hardware development for hardware designers and developers to bring solutions faster to market.
mcf52277 mcf5329 Thermometer Graphical User Interface

FEATURED PROJECT

Main GUI Blood Pressure GUI Glucometer GUI

Application Framework GUI Framework GUI Widgets

SEL Architecture SEL - Interfaces

Medical Services Blood Pressure Oximeter Weight Scale Glucometer

SEL OS fsl_os_linux

mx31

mpc5121e

mpc8360

Figure 2: Medical SEL SERVICE

Graphical User Interface Main GUI Blood Pressure GUI Application Framework GUI Framework GUI Widgets

SEL Architecture SEL - Interfaces

Medical Services Blood Pressure

SEL OS fsl_os_linux

mcf5329

Figure 3: Medical SEL SERVICES in a MCF5329

Reusing SEL services and spanning its usage across the 32-bit Freescale Portfolio enables the proper and rapid development of health telemonitoring equipment by creating a virtual bridge between doctor and patient.

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

11

F E AT U R E D P R O D U C T S
High-Speed FPGA 100G Interfaces
Xilinx, Inc. announced the industrys first dual 100 Gbps Gearbox solution for connecting 100G interfaces with the newest generation of high-density, 100 Gbps CFP2 optical modules. Incorporating a single 28nm Virtex-7 HT FPGA and Xilinx Gearbox intellectual property (IP) cores, the solution overcomes the initial hurdles of leveraging the new CFP2 optics supporting 100GE, OTU4 and 10 10 MSA specifications. This allows higher density 100G line cards and transmission equipment while lowering overall system power consumption and solution cost through integration. By offering the implementation based on the Virtex-7 HT FPGA family with 28 Gbps transceivers, Xilinx can provide communications customers with twice the density and more advanced debugging features compared to competing devices as well as replace what would be a three chip ASSP design. For more information, please click here.
PVDD THERM BOOTR+ OUTR+ GAIN CONTROL MODULATOR (-) FET DRIVER OUTR BIAS BOOTR

FEATURED PRODUCTS

SSM3302
INR+ 40k 40k INR

2x10 W Filterless Stereo Class-D Audio Amplifier

The SSM3302 is a fully integrated, high efficiency, stereo Class-D audio amplifier. The application circuit requires minimal external components and operates from a single 7 V to 18 V supply. The device is capable of delivering 2 10 W of continuous output power into a 4 load (or 2 8 W into 8 ) with <1% THD + N from a 12 V supply. In addition, while mono mode is activated, the user can drive a load as small as 2 up to 20 W continuous output power by stacking the stereo output terminals. The SSM3302 features a high efficiency, low noise modulation scheme that requires no external LC output filters. This scheme continues to provide high efficiency even at low output power. The SSM3302 operates with 90% efficiency at 7 W into an 8 load or with 82% efficiency at 10 W into 4 from a 12 V supply, and it has an SNR of >98 dB. For more information, please click here.

SDNR

MONO

INTERNAL OSCILLATOR

EDGE CONTROL

EDGE

SDNL INL+ 40k 40k INL GAIN CONTROL

BIAS

BOOTL+ OUTL+

MODULATOR (-)

FET DRIVER OUTL BOOTL

VREG GAIN VREG (AVDD) AGND REGEN PGND

Real-Time Clock Calendar IC

Microchip Technology Inc., a leading provider of microcontroller, analog and Flash-IP solutions, announced the expansion of its stand-alone Real-Time Clock/Calendar (RTCC) family with the I2C MCP7940M RTCC device. This new device is designed for the price-competitive consumer products market, and includes 64 Bytes of SRAM as additional scratchpad memory, as well as a digital-trimming circuit that can compensate up to 11 seconds per day for crystal error. The MCP7940M devices provide accurate timekeeping at a low cost for applications in the home-appliance (e.g., microwaves, washing machines, dryers, ovens, thermostats); audio/video (e.g., radios, televisions, set-top boxes, digital recorders); and consumer-electronic markets (e.g., printers, network routers, cameras ), among others. For more information, please click here.

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

12

EEWeb
Electrical Engineering Community

Contact Us For Advertising Opportunities

1.800.574.2791
advertising@eeweb.com

www.eeweb.com/advertising

Repeaters: Learn to Loveem

Michael Steinberger
Lead Architect, Serial Channel Products

Eventually, any stationary system youre working on will use electrical repeaters. This article explains why and demonstrates the type of analysis thats required when designing with repeaters. 1.0 The Need for Speed Many years ago, a director at Bell Labs told me that the cost of a piece of equipment was roughly proportional to its weight. We sell equipment by the pound, he said. While that was a slight exaggeration, he had a point. Printed circuit boards, power supplies, connectors, and sheet metal represent a significant percentage of the total cost of any shelf or rack of equipment. Some years later, I had an

opportunity to explore this principle in detail. I was the circuits manager working with the manager of the mechanical design group, helping to choose the technology for a very high-performance system. Instead of trying to reduce the cost of the system at a fixed performance level, we decided to evaluate the cost and performance for each possible set of technology choices. Because the performance of the system was known to be directly proportional to the bandwidth of the interconnect, my group was responsible for estimating the maximum achievable data rate, and therefore the performance. My partners group was responsible for estimating the cost. We put all our data into a sophisticated

spreadsheet that plotted a scatter graph of the performance versus the cost for each one of the ten thousand possible technology choice combinations. Our study demonstrated that the highest performance-to-cost ratio was consistently achieved by the technology combinations that used electrical repeaters to support a higher data rate in the system interconnect. I have since lost access to that data, and so can no longer estimate cost accurately. Nonetheless, Figure 1 gives some sense of what those results looked like. Figure 1 assumes an equipment shelf containing sixteen line

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

14

TECHNICAL ARTICLE
Cost vs. Capacity
No repeaters 20000 18000 16000 14000 12000 10000 8000 6000 4000 2000 0 0 2 4 One repeater Two repeaters

TECHNICAL ARTICLE

Cost (zorkmids)

Bad

Good

10

12

14

16

18

Data Rate (Gb/s)

Figure 1: Capacity vs. relative cost for a hypothetical shelf of equipment

cards and two switch cards. The technology choices were as follows: 1. The material for the line card, switch card and backplane could be either FR4 or a lower loss material. The material was chosen independently for each of the three printed circuit boards. 2. There could either be no electrical repeaters in the system, repeaters on the line cards only, or repeaters on both the line cards and switch cards. The capacity of the processing ICs for the line and switch cards was held constant for all configurations. It was assumed that the capacity of the processing ICs was high enough, and that it was practical to populate the cards with enough ICs to consume the maximum interconnect bandwidth. If this assumption isnt valid, then

repeaters arent required in the first place. Figure 1 shows that the most cost effective configurations are the ones that use repeaters on both the line and switch cardsin other words, the configurations that can support the highest data rate. Most studies of this type reach a similar conclusion. In other words, the most cost effective system designs are the ones which achieve a high enough interconnect bandwidth to pack as much processing power as possible onto a printed circuit board. If the highest processing density requires electrical repeaters in the interconnect, then repeaters are going to be an indispensable part of the optimal solution. Note that the above conclusion applies primarily to high-capacity stationary systems such as core

data routers and high-performance computers. The power limitations in portable devices typically reduce the processing power to such an extent that interconnect bandwidth is not a limiting factor. 2.0 Rules of the Road Its tempting to analyze a channel with electrical repeaters by breaking it into individual segments and analyzing those segments independently. Unfortunately, this approach cannot be relied upon to produce accurate results. Analyzing the entire channel in a single analysis is much more reliable, and there are products out there such as SiSofts Quantum Channel Designer which are able to perform such analyses. There are several effects that need to be considered: 1. Many repeaters are either linear

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

15

TECHNICAL ARTICLE
or quasi-linear. That is, they do not recover the clock and regenerate the data. The effect of such repeaters is cumulative across the entire channel and is not isolated to any individual segment of the channel. 2. If the repeater is nonlinear, then the placement of the repeater in the channel has an additional constraint in that placing the repeater too close to the transmitter could reduce the effectiveness of the repeater. 3. If the repeater has clock recovery and data detection, then the individual segments of the channel are more nearly independent. That may or may not be a good thing depending on the placement of the repeater. Furthermore, the recovered clock from the repeater becomes the clock that the receiver must track. The resulting increase in clock phase noise should be accounted for in the analysis. The following sections briefly illustrate each of these effects. 2.1 Linear Repeater Figure 2 is a schematic for a simplified channel with repeater. In Figure 2, the length of transmission line from the transmitter to the repeater and from the repeater to the receiver are variable. To make observations about repeater placement clearer, the sum of the transmission line lengths was constrained to be a constant 100. Observations were made at the output of the repeater and at the decision point of the receiver. Both the transmitter and the repeater output driver had 3dB of precursor deemphasis, both the repeater input and receiver input had a continuous time linear equalizer (CTLE) and the receiver had five taps of decision feedback equalization (DFE). The data rate was 5 Gb/s. Figure 3 shows eye diagrams at the input to the repeater driver and at the receiver decision point for three different repeater locations: 10, 50, and 90 from the transmitter. As shown in Figure 3, even though the eye diagram at the input to the repeater driver varies considerably as a function of repeater location, the eye diagram at the receiver remains almost constant. This is consistent with the assumption that the repeater is linear. There is a slight variation of the eye diagram at the receiver due to changes in reflected waves between the repeater, the transmitter driver,
TX1 sisoft_serdes SiSoft_AMI_Tx W1 *_1_diff_strip_w... $len1 RP2 sisoft_serdes SiSoft_AMI_Repea... W2 *_2_diff_strip_w... $len2 RX1 sisoft_serdes SiSoft_AMI_Rx

TECHNICAL ARTICLE

A I

I A I

Input to repeater driver

Receiver decision point

Figure 2: Simplified channel with repeater

Input to repeater driver

Receiver decision point

Repeater location = 10

Repeater location = 50

Repeater location = 90
Figure 3: Repeater and receiver eye diagrams for three different repeater locations

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

16

TECHNICAL ARTICLE
and the receiver buffer amplifier. However, these variations are relatively small and therefore difficult to see in Figure 3. Note that, as demonstrated in Figure 3, the eye diagram at the output of the repeater is no indication whatsoever of the performance of the end-toend channel. The eye diagrams at the output of the repeater are very different, and yet the end-to-end channel performance is almost identical in all three cases. For later comparison, Figure 4 shows the eye height for a 10-12 BER at the input to the repeater driver and the receiver decision point, each as a function of repeater location. 2.2 Nonlinear Repeater In general, repeaters are not perfectly linear; and in many applications, the non-linearity of the repeater is significant. One of the primary functions of a repeater is to insert gain into the channel; a number of repeater designs
Repeater 0.7 0.6
Eye Height (V)

achieve this by providing a high gain saturating amplifier.

versus the repeater location for a nonlinear repeater in the same way that Figure 4 does for a linear repeater. In particular, Figure 5 demonstrates that if the repeater is placed too close to the transmitter, then the saturation of the repeater will severely degrade performance. 2.3 Retiming Repeater As mentioned above, there are also repeaters that include clock recovery and data detection. These are often called retimers, or in telecommunications transmission systems theyre often called regenerators. While a channel with retimers can be analyzed as the concatenation of independent segments, there are practical considerations and cumulative effects which should be understood. Figure 6 is analogous to Figure 4 and Figure 5 in that it shows eye height as a function of repeater location. In this case, however, the eye height at the input to the repeater driver is particularly important in that for a retimer, it is the node where the data is detected. Note in Figure 6 that for repeater locations beyond 60, the data detection in the repeater fails, causing failure of the entire channel. Even though the receiver has more than enough equalization capacity for the failing repeater locations, that excess capacity cannot be used to improve performance because it occurs after the bit errors have already been made. Figure 6 also shows that with a retimer, the channel length could be extended

TECHNICAL ARTICLE

The analysis of this type of system is perhaps the most complex. The repeater is not linear, so cascading transfer functions is not valid. Yet the repeater does not recover the clock and regenerate the data, so the channel cannot be broken into independent segments. The most reliable way to analyze a channel with nonlinear repeaters is a time domain simulation. However, as demonstrated in [1], time domain simulations cannot produce a statistically significant sample of the data waveform. There are approximate methods in statistical analysis and statistical extrapolation that can produce relatively accurate results; however, those methods are beyond the scope of this article. As a demonstration of the practical considerations associated with the application of nonlinear repeaters, Figure 5 shows the eye height
Receiver

0.5 0.4 0.3 0.2 0.1 0.0 0 10 20 30 40 50 60 70 80 90 100

Repeater Location (inches)

Figure 4: Eye height as a function of repeater location

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

17

TECHNICAL ARTICLE
Repeater 0.4 0.35
Eye Height (V)

Receiver

to a total of at least 13060 from the transmitter to the retimer and 70 from the retimer to the receiver. When using retimers, clock phase noise also becomes a more complex phenomenon. Rather than recovering the transmitter clock, the receiver recovers the clock from the retimer, which in turn recovers the transmitter clock. Thus, there are two clock recovery loops in series. There is an advantage in that any clock phase noise at the transmitter gets filtered twice, and so its effect at the receiver is considerably reduced. The disadvantage is that each clock recovery loop introduces pattern dependent jitter. Figure 7 shows the clock phase noise spectral density produced in a simulation in which a specific phase noise spectrum was injected into the transmitter and then the receive phase noise spectrum was measured with and without a retimer in the channel. In Figure 7, the transmit spectrum is shown in green, the spectrum without retimer is shown in grey, and the spectrum with retimer is shown in light green. The transmit phase noise spectrum in Figure 7 was generated to demonstrate a behavior and does not attempt to represent any real system. In particular, the spectral components in the transmit spectrum are used to make it easy to distinguish between transmit phase noise and pattern dependent jitter. When the retimer is disabled, the transmit spectral components are still readily visible in the receiver clock spectrum. However, when the retimer is enabled, then the transmit spectral components are no longer visible. The trade-off, however,

TECHNICAL ARTICLE

0.3 0.25 0.2 0.15 0.1 0.05 0.0 0 10 20 30 40 50 60 70 80 90 100

Repeater Location (inches)

Satisfactory Design Region

Figure 5: Eye height vs. repeater location for a nonlinear repeater

Retimer 0.7 0.6
Eye Height (V)

Receiver

0.5 0.4 0.3 0.2 0.1 0.0 0 10 20 30

Satisfactory Design Region

40

50

60

70

80

90

100

Figure 6: Eye height vs. repeater location for a retimer

Retimer Location (inches)

Transmitted and Received Jitter Spectra

Retiming Enabled Retiming Disabled Transmit Injected Jitter

Transfer Function

-90.0
Transmitted spectral components filtered out by retimer

-100.0
DB

-110.0

-120.0

Increased pattern dependent jitter with retimer

20.0

40.0

60.0
Hertz (MHz)

80.0

100.0

120.0

Figure 7: Clock phase noise spectra with and without retimer

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

18

TECHNICAL ARTICLE
is that there is increased pattern dependent jitter when the retimer is enabled. 3.0 Conclusion For high performance stationary systems, electrical repeaters are eventually going to become necessary to achieve a competitive performance-to-cost ratio. Designing with repeaters has its own set of considerations which at first may not be intuitively obvious. When incorporating electrical repeaters into the system design, the performance analysis must consider the entire channel, end to end. 4.0 References [1] Mike Steinberger, Accuracy of the Computational Experiments called Time Domain Simulations, EEWeb, http://www.eeweb. com/blog/michael_steinberger/ accuracy-of-the-computationalexperiments-called-time-domainsimulations, July 4, 2011. About the Author Michael Steinberger, Ph.D., is responsible for leading SiSofts ongoing tool development effort for the design and analysis of serial links in the 5-30 Gbps range. Dr. Steinberger has over 30 years experience in the design and analysis of very high speed electronic circuits. Dr. Steinberger began his career at Hughes Aircraft designing microwave circuits. He then moved to Bell Labs, where he designed microwave systems that helped AT&T move from analog to digital long-distance transmission. He was instrumental in the development of high speed digital backplanes used throughout Lucents transmission product line. Prior to joining SiSoft, Dr. Steinberger led a group of over 20 design engineers at Cray Inc. responsible for SerDes design, high speed channel analysis, PCB design and custom RAM design.

TECHNICAL ARTICLE

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

19

Would you...Could you... Should you...

Sr. Manager, SW Product Planning

Phil Simpson

Compile your FPGA design on the Cloud?

What if FPGA vendors offered access to their software via Cloud? If you are the end user of FPGA Design Software, a Cloud Computing model would provide you with elasticity and flexibility. You would have access to more computer power when needed; the availability of compute servers would become on-demand. You could select your server size from small (2 Virtual Core and 7.5GB RAM) to large (8 Virtual Cores and 70GB RAM), upload your design and scripts to your dedicated workspace and go. There is the potential to check the compile status, compilation reports and launch compiles from your smartphone or tablet. Many companies have their own compute farms and may not see the need for a Cloud solution. However, when it is crunch time on the project and you need to compile multiple variations of your design to reach timing closure, extra compute resources are appealing. Imagine running the equivalent of the Altera Design Space Explorer on the Cloud. You could compile multiple design or setting iterations in parallel, reducing the closure timing (e.g., for a four-hour compile time), compiling 10 variations or seeds in 4 hours as opposed to 40 hours. A Cloud solution could also offer convenience. It provides the convenience of avoiding those long software download and installation times. It would provide an easy way to evaluate new releases of the software without having to commit to the long installation process. It could also reduce your Information Technology (IT) costs in adding new hardware to your network and maintaining it. Security is interesting. There are many Web-based software packages and services available on the Cloud today. Most offerings use the Amazon Elastic Compute Cloud (EC2) and use SSH with private keys for access. For most of our everyday tasks we use the Cloud, and never think to question the security as we provide credit c a r d

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

20

TECHNICAL ARTICLE
information and other personal details; but will a company trust the upload of its IP to the Cloud? In addition to the SSH with private key access, the issue of IP security can be addressed via the deployment of the encryption technology that the FPGA vendors use on their own IPs. While certain business segments will not be satisfied with this solution, others may. This also opens the door to different business models for the FPGA design software. One option could be the payas-you-go model (i.e., only pay for licenses when they are being used). Or alternatively, you use your existing licenses and pay for the compute power. FPGA vendor software is significantly less expensive than traditional EDA software, thus the business model would likely differ from the model that the EDA industry will inevitably adopt. So, why have the FPGA vendors not provided Cloud access? GUI response due to network bandwidth could be an issue and require GUI redesign. However, script based compiles, which most designers are using today, would work perfectly. The real reason is more business uncertainty rather than technical reasons. Would you use it? If you would use it, what price are you willing to pay for this capability? About the Author Phil Simpson is Alteras senior manager for software technical marketing, product planning, and EDA relationships. In this role, he is responsible for Alteras Quartus II software and third-party EDA interfaces product planning and the creation of the Altera design flow software roadmap. Prior to joining Altera in 1997, Phil held several engineering roles at various EDA and semiconductor companies, including EDA Solutions, Data I/O, and Graseby Microsystems. He holds a B.S. (with honors) in Electrical & Electronic Engineering from City University, London and an M.S.C. (with distinction) in system design from the University of Central England, Birmingham, England. Phil is a published book author on team-based FPGA design. In addition he has written and had published numerous technical articles on topics related to his experience.

TECHNICAL ARTICLE

Master Web App

Compile Node Servers

Get your results anywhere

Figure 1: Potential Cloud Set-up

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

21

RETURN TO ZERO
RETURN TO ZERO

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

22

RETURN TO ZERO
RETURN TO ZERO

EEWeb | Electrical Engineering Community

Visit www.eeweb.com

23