Electronicspecifier Design 201108

8/4/2019 Electronicspecifier Design 201108

1/48

Avoiding EMI In Capacitive Touch Screens, p4True Grid Independence, p30Driving Revolution In Power Electronics , p34

Avoiding EMI In Capacitive Touch Screens, p4True Grid Independence, p30Driving Revolution In Power Electronics , p34

Driving Efficient

Power Solutionsfor LED Lighting

Driving Efficient

Power Solutionsfor LED Lighting

August 2011 / Volume 1, Issue 7


2/48
http://www.electronicspecifier.com/


3/48www.electronicspecifier.com electronicspecifierdesign | august 2011 | 3

august 2011

4 Displays: Avoiding EMI In Capacitive Touch

Capacitive touch has become the technology ofchoice for high end consumer applications but thetechnology is particularly susceptible to EMI. Howdo major manufacturers deal with this problem?

10 System Security: Wireless Access ControlsWeakest Link

Wireless access control is only as good as itsweakest link, so check out your options!

16 ASICs : Shrinking Chip Geometry for MaximumReturn

How did two chip companies reach almost identicalSoC unit prices and which achieved the best result?

22 Test & Measurement:AOI: Agile Optical Inspection

How one company adopted AOI to achieve theinspection resource it needed.

26 Smart Metering: Building A Smarter Meter

The move to smart energy brings opportunity acrossthe supply chain but choosing the right solution

needs careful consideration at every level.28 Connector Solutions: Holistic Approach Is

Beacon For Waterproof Connectors

Waterproof connectors conceived from a problemno one else had managed to solve.

30 Power: True Grid Independence

Energy harvesting for wireless sensors using apiezoelectric energy harvesting power supply.

34 Power: Driving Revolution In Power Electronics

Heralding a new era in power electronics is the firstcommercially viable application of GaN basedpower device technology.

38 Powering Light: Driving Efficient Power Solutionsfor LED Lighting

Best practice in developing efficient power supplysolutions for solid state lighting reviewed.

42 Smart Design: Maintaining More For Less

Defence budget cuts call for smarter design practices.

44 Design: Meeting The ChallengeOf High Speed Design

Altium Designer for high speed digital board design.

CONTENTS

ElectronicSpecifier Ltd.Ellion House, 6 Alexandra Road, Tonbridge, Kent TN9 2AA, UK

Editor: John Taylor BA, IEng, MIETEmail: [email protected]

Tel/Fax: +44 (0) 1843 866 607

Copy Editor: Raj Joshi

Advertising: Tim AnsteeEmail: [email protected]

Tel: +44 (0) 1732 366 624 Fax: +44 (0) 1732 366 052

Publisher: Steve RegnierEmail: [email protected]

Tel: +44 (0) 1732 366 617 Fax: +44 (0) 1732 366 052

Subscriptions Subscribe to the Digital

Edition of ElectronicSpecifier

Distribution Blog Find out whats happening

in the world of distribution

Page 34.

Page 22.

Page 28.

Page 10.
http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/SubscribeMag.asphttp://www.electronicspecifier.com/SubscribeMag.asphttp://www.electronicspecifier.com/SubscribeMag.asphttp://www.electronicspecifier.com/SubscribeMag.asphttp://distributor.electronicspecifier.com/http://distributor.electronicspecifier.com/http://distributor.electronicspecifier.com/http://distributor.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/SubscribeMag.asphttp://www.electronicspecifier.com/SubscribeMag.asphttp://www.electronicspecifier.com/SubscribeMag.asphttp://distributor.electronicspecifier.com/http://distributor.electronicspecifier.com/http://distributor.electronicspecifier.com/http://www.electronicspecifier.com/


4/48www.electronicspecifier.com4 | august 2011 | electronicspecifierdesign

Capacitive multi-touch technology is the mostwidely adopted technology in these high endconsumer devices as it promises a raft of bene-fits over its competitors. However, the technol-ogy does face some technical challenges.

One of the biggest challenges is that of EMI:there are many sources of electromagnetic in-terference around and capacitive technologyis particularly susceptible.

The most commonly cited noise source isfrom the LCD itself, which of course is perma-nently in close proximity to the touch sensor.Every LCD panel is different and noise pro-duced by LCDs has a correspondingly widerange of intensities. OLEDs, by comparison,are thought of as a noise friendly choice.

Charging devices are another common

source of noise. Since the standardisation ofmobile phone chargers to use micro USB,manufacturers no longer have control over thequality of the charger that is plugged into their

touch screen mobile phone, and unqualifiedchargers can introduce significant noise.

So how do high end touch controller manu-facturers mitigate EMI in their products?

Superior algorithmsAtmel recently released the maXTouch

mX122E series of capacitive touch controllersoptimised for 2.8 to 3.5in touch screens such

as digital cameras and GPS devices.They allow up to four simultaneous touches,

a level of performance previously seen only inthe smartphone market.

Touch processing rejects unintentionaltouches caused by a gripping hand whilst theyinterpret light touches correctly for gesturesmade on the device. The part can support anarrow passive stylus input even when theusers hand in resting on the screen in a natu-ral writing position.

Atmel says these devices deliver the most ad-vanced signal processing and algorithms tomitigate noise from after-market chargers andother sources.

The common technique to guard against dis-play noise is to add in a physical shield layer

between the sensor and the display, whichadds both cost and thickness to the design, ex-plains Helen Francis, Atmels Senior MarketingManager. With noise processing on Atmels

displays

Avoiding EMI In

Capacitive Touch ScreensCapacitive touch is now the technology of choice for high end consumer

applications but the technology is particularly susceptible to EMI.Sally Ward-Foxton asks how do major manufacturers deal with this problem?

Touch screen technology has been incor-porated into practically every con-sumer device with a screen, from the

large display of the iPad down to the small-est digital camera or GPS screen.

Millions of devices integrate touch technolo-gies, employed to create interesting user expe-riences utilising touch, multi-touch and gesturerecognition.
http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/



latest E series devices, this isno longer required and Atmelcan support shieldless designsfor devices ranging from hand-sets to tablets. This results inthinner, lighter, more appeal-ing products.

For many competing solu-tions RF noise is also a chal-lenge, she adds. However,the Atmel touch screen controller operates out-

side this frequency range, sothis is easily avoided.

Dynamic rangeMaxim has developed the

Max11871 as part of its Tac-Touch series which, it says,has such a high dynamic

range that it can detect evena touch from a gloved hand,which is notoriously difficultfor capacitive touch. This

product can detect and track up to a ten fingersimultaneous touch.

Maxim says the Max11871s analogue frontend is built from the ground up for capacitivetouch by the company's high resolution dataconverter IC design team, and provides a near60dB signal to noise ratio performance

Maxims Max11871 provides near 60dB SNR per-formance, a 1000:1 equivalent ratio betweentouch and no-touch.

Atmels maXTouch mXT112E al-lows up to four simultaneoustouches, a level of perform-ance previously only seen in

the smartphone market.
http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/



which is equivalent to a 1000:1 ratio betweentouch and no-touch. This is said to be some tentimes more than existing products.

Noise is the number one problem for our

handset customers [when selecting a capaci-tive touch controller], says Bart DeCanne,Business Director for application specific dataconverters at Maxim. EMI immunity is reallya problem of dynamic range: you want to de-tect a weak signal, like a touch from a glovedhand or stylus, in the presence of a largenoise. Most touch controllers on the markethave a broadband analogue front end con-

troller but you need to scale back the gain ofthe front end controller or it will be over-loaded [with noise] and the weak signal willbe lost.

Maxims im-pressive SNRenables detec-tion of veryweak (in the

femtofaradrange) touchvariations, suchas from a handwaving near thescreen (proxim-ity detection),touch from a

fine tip stylus orballpoint pen,or a glovedhand. Also, itmeans that thetouch point canbe further away from the sensor, allowing forthicker cover glass for improved ruggedness.

We also have a system which rejects noise

at frequencies other than the frequency we ac-cept from the touch screen, he says, referringto the Max1871s proprietary architecture thatrejects noise (by over 40dB) from external

sources such as AC USB chargers, LCDs, orCFL lights.

Frequency hopping

The STMicroelectronics solution to displaynoise is based on synchronising the acquisitionof touches with the display blanking time. Andthe companys latest multi-touch controller, Fin-gerTip, provides both frame and line sync meth-ods. This single chip solution is for touch screensup to 10in diagonal and features a 32bit DSPengine to help drive its EMI immunity.

The company also cites conducted noise from

chargers, present on both power supply andground, as a problem since it creates falsetouches. Giuseppe Noviello, Director of Tech-nical Marketing for the STMicroelectronics

Sensor BusinessUnit, says thatfalse touchescan also resultfrom noise in-

jected by theusers fingerstouching thepanel. In thecase of fingernoise, he says,it is the humanbody that cap-

tures radiatednoise, like thenoise generatedby fluorescentlamps or comingfrom other low

frequency (below 1MHz) sources and trans-fers it to the screen. There is no easy counter-measure because it is in the same frequency

range as the panel touch acquisition circuitsand even if its periodic, it is affected by widejitter and most of the time it is modulated.

According to Noviello, the common solu-

displays

ST used its know-how in analogue and mixed signal interfacingto develop an innovative analogue front end that detects

capacitance variations in atto-Farads - 1018F.



tion for conducted noise is based on frequencyhopping and a heavy firmware heuristic algo-rithm to eliminate suspected false touches.Using the MCU to create filters can reduce or

eliminate the unwanted touches but it will re-duce the scan rate and response time. Fre-quency hopping also cannot deal effectivelywith noise when there is modulation and jitter.

FingerTip, like other solutions, uses frequencyhopping to scan the panel in a noise-free areain the frequency domain. This area may bepretty narrow be-cause of jitter and

modulation, causingcorruption of thevalid signals comingfrom any fingermovement.

FingerTip uses anarrow band digitaldemodulationmethod with propri-

etary DFT filteringthat is able tostrongly attenuatethe adjacent noisecomponent. All thesignal processing isdone in the ana-logue front end using a state machine: the

processor is not involved so the response timeis not affected and the raw data coming in isalready filtered to an SNR level that can pro-vide the required accuracy and jitter, even inthe presence of tens of volts of noise injectedon the panel.

In order to overcome these limitations,Noviello says, ST is adopting a series of tech-niques that come from its strong know-how in

capacitance to voltage conversion alreadyused in MEMS analogue front end design.

In a MEMS device, the mechanical sensoroutputs capacitance variations as low as atto-

Farads (1018F) when the capacitance of thesystem is a million times bigger. The detectionof such small signals requires sophisticatedanalogue filtering techniques in order to get

fast response time with a strong SNR.The capacitive touch screen. Noviello ex-plains, also sends small capacitance varia-tions created by the finger touch to the ana-logue signal conditioning circuit. Like theMEMS case, these variations are very small,typically in the range of femtofarads but the

panel capacitanceto earth is a thou-

sand times bigger.The analogue frontend resolution mustbe enhanced to asignificant level fora touch to be recog-nised, and Finger-Tips ADC and sig-nal conditioning

circuits use similarmethods to MEMSdevices to achievethe best SNR.

Capacitive touchscreen controllermanufacturers need

to protect against noise of many different

types from many different sources.Advances in design for EMI immunity nowallow todays touch screens to pick out eventhe smallest signals from an ocean of electricalnoise. And manufacturers are continuing to re-fine their approaches to EMI in capacitivetouch and are also introducing new techniquesbased on their accumulated experience inanalogue electronics and MEMS. TR

More from Atmel More from Maxim More from STMicroelectronics Return to contents page.

displays

STs MEMS sensors and FingerTip technology share asimilar architecture whereby the sensing element is con-

nected to a high performance capacitance sensing circuit.
http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/Atmel-Corporation.asphttp://www.electronicspecifier.com/company/Atmel-Corporation.asphttp://www.electronicspecifier.com/company/Atmel-Corporation.asphttp://www.electronicspecifier.com/company/Maxim.asphttp://www.electronicspecifier.com/company/Maxim.asphttp://www.electronicspecifier.com/company/Maxim.asphttp://www.electronicspecifier.com/company/STMicroelectronicshttp://www.electronicspecifier.com/company/STMicroelectronicshttp://www.electronicspecifier.com/company/STMicroelectronicshttp://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/Atmel-Corporation.asphttp://www.electronicspecifier.com/company/Maxim.asphttp://www.electronicspecifier.com/company/STMicroelectronics


8/48

rswww.com/electronics

Make us part of your electronics line-up.

FIND IT Quickly.

DESIGN IT For Free.

BUY IT In a Click.



With such a wide range of component andtechnology choices, designers should not sim-

ply look for specific security features on a de-vice datasheet but engage in a detailed discus-sion of all design options with themicrocontroller supplier.

One wireless security solution simply doesnot fit all applications and various factorsneed to be balanced if designers are to find asystem that delivers adequate, affordable pro-

tection. Latest microcontrollers feature ad-vanced security encryption algorithms, on-chipRF communication or low power technology,together with dedicated RF ICs, and can helpdesigners easily develop a complete, securewireless system that delivers the right balanceof cost, size and functionality.

No wireless access control device is unbreak-able! It only takes time and money for crypto-

attackers to find a way to break the deviceand read the protected information. With asystem level approach to security, designerscan develop a powerful arsenal with which to

system security

Wireless

AccessControls

WeakestLink

Wireless access control is only as good as its weakest link,says Microchips Vivien Delport & Cristian Toma.

Designing low cost, secure wirelesssolutions has been significantly simpli-fied with recent advances in micro-

controllers, RF ICs and compact security al-

gorithms. However, wireless design stillrequires a strong understanding of the latestattack methods used to break security sys-tems. Finding the right counter-measuresthat are affordable is only possible if thedesigner takes on a system wide approachto security.



protect wireless access control devices. Secu-

rity is a system wide issue and it is crucial fordesigners to consider the security of the mo-bile/transmitter and the base/receiver sectionof their design, as well as potential weak-nesses in their hardware.

Security algorithmsUnless a designer is fully familiar with differ-

ent types of potential crypto-attacks such asplaintext, side channel, differential crypto-analysis, meet in the middle and slide attacks,choosing a security algorithm can be difficult.

Using a public algorithm such as DES, thedata encryption standard, encryption is pro-vided by a 56bit encryption key whilst AES,the advanced encryption standard can use ei-ther 128 or 256bit keys. Designers can also

use proprietary algorithms such as Mi-crochips Keeloq which combines a strongcryptographic algorithm with code hopping.

Code hopping, or rolling code, provides an

additional level of security by changing the ci-

pher message each time it is transmitted, thisthen prevents the re-use of previously transmit-ted messages.

Unfortunately, higher security often meanshigher cost. The stronger the algorithm, themore complex the calculations and there liesthe need for larger software memory. This typ-ically requires a more expensive microcon-troller which adds to the overall cost of the se-

curity solution and its complexity.Stronger algorithms also typically result in

longer crypto messages that need to be trans-mitted. This will add longer time delays andalso increase the power consumption whensending the radio packet over the air, becausethe packet takes more time to send. Longerdata transmission is not always desirable andcan negatively influence the field acceptance

of the product.

Hiding the keyEffective key management is as important

as the choice of security algorithm. Kerck-hoffs Principle states that a security systemshould not rely on the security algorithm beingsecret but rather on the key being secret. It isalways safest to assume that both the en-crypted message and the algorithm will even-tually be known to the public, even if it is aproprietary algorithm.

System security should, therefore, neverrely solely on the algorithm being secret butalso consider how the security encryptionkeys will be generated, exchanged, stored,safeguarded, used and replaced throughout

the system to decipher or unscramble en-crypted messages.A critical element of any key management

scheme is that not all devices use the same

Advanced microcontrollers are more affordable,in wireless access control, says Microchip.



secret key. This helps to increase overall sys-tem security, so that if a single mobile unit iscompromised, it does not compromise the en-tire security system. The easiest way to imple-

ment this is to give each mobile unit its ownunique secret code or encryption key.One method often used to accomplish this is

to serialise each unit with a unique numberand then base the calculation of the uniqueencryption keys on this serialnumber and a master man-ufacturers code. A re-ceiver unit that

needs to supportmultiple mobileunits at the sametime can then eas-ily use the serialnumber to derive theencryption key needed todecipher information transmit-ted from that specific mobile device.

Mobile unit serialisation is typicallycarried out at the time of production,either by preprogramming the embed-ded microcontroller with this informationbefore placing it on the printed circuitboard or by using an in-circuit serial program-ming interface to programme the microcon-troller after board assembly.

It is essential to protect the encryption keysat all times, including during manufacturingand especially if assembly is carried out by athird party contract manufacturer. It is prefer-able to provide the contractor with prepro-grammed, code protected microcontrollersrather than try to secure a complete produc-tion flow against the illegal copying of encryp-tion keys. Most microcontroller suppliers, such

as Microchip, provide serialised quick turn-around programming options on all of theirmicrocontrollers. By providing the manufac-turer with device serialisation information, they

can preprogramme both the application soft-ware and serialisation information into the mi-crocontroller during production testing.

A further good way of protecting system se-

curity is to make regular changes rather thankeeping the same security solution with theexact same security key information for pro-longed time periods. Make changes to eitherthe key management scheme, the master en-cryption code used to derive the unique en-cryption keys for each mobile unit, or even mi-grate to next generation security algorithms as

they become available.

The downside to change isthe loss of backward

compatibility but thisis a design trade-off that systemdesigners needto evaluate. Inthese types ofdesigns, anembedded mi-crocontrollermakes it eas-

ier to implementon the fly changes,

without the need for a com-plete redesign and allowing the same hard-ware design to be used for different products.

Hardware securityAttacks on security systems reach beyond

analysing data and trying to perform mathe-matical attacks on the security system. Theyalso include analysing the application circuitto see if hardware tampering allows access tothe secured system. If the receivers output sim-ply pulls a digital line high to activate a relay

that presents an easy point of attack. Ofcourse, this only works if an attacker can getphysical access to the receiver units hardwarewhile in use.

system security



Another attack scheme involves analysingthe mobile sender units from the physical com-ponent side. This involves analysing the actualcircuit and applying specification voltages that

signal the microcontroller or current starve theapplication to see if this allows the attacker toread the secured information stored inside thedevices non-volatile memory.

There are also other invasive and non-inva-sive methods of attack that try to break thecode protection locking that has been builtinto microcontrollers.

With cyber attackers continually trying to de-

vise new threats, component manufacturersare constantly adding more layers of physicalobscurity to protect algorithm codes and keysstored in microcontrollers. It is always best towork closely with a microcontroller supplier tounderstand which devices incorporate the lat-est tamper-proof circuitry to protect the infor-mation stored inside the device.

RF parametersThe frequency used will depend mainly upon

the application and regulations. For example,in the US, ISM bands are 315 and 915MHz,whilst in Europe, they are 433 and 868MHz.

The distance covered by the radio link is alsosubject to guidelines. A typical RKE applica-tion requires at least 20m and there can some-

times be a maximum distance requirement. InJapan, for example, the maximum coveredrange is just 5m and this is simply down totougher RF regulations.

One of the most common mistakes in designis to focus on the maximum transmitter rangeand forget that the transmitter and receiverare equally important. Good antenna designcan significantly improve the reception from a

weak transmitter.The RF modulation scheme and data rate

also have a major impact on operational relia-bility of the radio link. Frequency modulated

radio links are typically less subject to noise.However, such technology adds cost. A moreadvanced radio link adds cost both to the re-mote unit and to the receiver. However, with

todays advances in integrated RF transmitters,receivers and transceivers, these devices canfall into the same price bracket as low cost hy-brid RF modules.

Managing costDesigners need to fully understand what they

are trying to protect and then decide on whichsecurity solution to use. This of course will af-fect total system cost in a number of ways.

Using a microcontroller based solution, in-stead of an ASIC based design, adds flexibil-ity. Microcontrollers enable designers to makechanges simply by altering the software. Thisis also true if minor code changes are neededto support multiple countries regulations using

the same hardware design.

Microcontroller selectionThe latest microcontrollers enable easier de-

system security



velopment of wireless products while deliver-ing high levels of security through the use ofsoftware blocks that support most of the en-cryption algorithms in a high level language

such as C. This signifi-cantly simplifies the devel-opment of a secure wire-less application, whichcan easily be tailored tothe rapidly changing con-sumer markets.

Some microcontrollerscan reduce design com-

plexity by integrating on-board wireless peripher-als. Microchips rfPICdevices, for example, in-tegrate UHF wirelesstransmitters for low

power RF applications whilst supporting spaceconstrained applications with a small packageoutline and a low external component count.

Other microcontrollers, such as XLP based

eXtreme Low Power PIC devices are optimisedfor low power applications. These XLP PICMCUs feature sleep currents down to 20nAand provide compatibility with dedicated RF

modules for IEEE 802.11 WiFi or IEEE802.15.4 ZigBee, as well as transceivers andreceivers for ISM band applications.

RF componentsAnother significant advance now enabling

shorter time to market is a wider selection ofintegrated RF transmitter receivers or trans-ceivers. These devices help to reduce the com-

plexity of RF design by integrating most of theRF circuitry needed into a single chip.

Next generation RF ICs only need a fewbasic external components to enable the fullimplementation of a high performance RFwireless implementation. These devices also

typically incorpo-rate an SPI inter-face for an easy

connection to amicrocontroller.This MCU thenconfigures the RFradio to the ap-propriate requiredsettings and sendsand receives the

demodulated datapackets. TR

Vivien Delport isDirector of Applica-tions EngineeringSecurity at Mi-crochips Microcon-troller & Technol-

ogy Division and Cristian Toma is MicrochipTechnologys Applications Engineer. More from Microchip Return to contents page.
http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/Microchip-Technology-Ltd.asphttp://www.electronicspecifier.com/company/Microchip-Technology-Ltd.asphttp://www.electronicspecifier.com/company/Microchip-Technology-Ltd.asphttp://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/Microchip-Technology-Ltd.asphttp://www.electronicspecifier.com/



ing the cost and availability of suitable IPblocks, as well as the costs to acquire the neces-sary knowledge and to update the company in-frastructure to work at the most advancednodes. Furthermore, non-engineering factorssuch as short term and strategic corporate ob-jectives can also have an important influence on

node selection.

Competing approachesThe two specialist chip companies described

asics

Recently, two competing chip companies, each producing ICs in high volume,chose significantly different process nodes for new SoC products targeting the

same application. How did the two reach an almost identical unit price and which

one achieved the best result overall? Shimon Raviv unravels the conundrum.

As semiconductor fabrication technolo-gies continue to advance, the most ad-vanced independent semiconductor

foundries are able to support processes rang-ing from mature geometries such as 0.18m tocutting edge nodes such as 28nm.

Newer, smaller geometries are understood to

carry higher NRE and per wafer costs, while thelarger number of dice/wafer results in a favour-able unit price after production ramp-up.

Other factors must also be considered, includ-

Shrinking Chip GeometryShrinking Chip GeometryFor Maximum ReturnFor Maximum Return


15/48



produce relatively high volumes of ICs, andthe devices are divided approximately equallybetween analogue and digital features.

Company A saw 0.18m as the optimum

technology for its custom SoC. Typically, fabri-cation at this node is on 8in/200mm wafers.By targeting this node, the company avoidedthe time and cost to acquire knowledge and IPat finer geometries. There is some sound rea-soning behind this decision: miniaturising ana-logue circuitry for very small process geome-tries delivers a relatively low saving in terms ofdie area.

Company As view that 0.18m can deliver astrong combination of dense digital circuitry andcompetitive analogue circuitry is reasonable.

Company B chose 65nm for its competingchip design. The resulting cost for a12in/300mm wafer was a little over five timesthat for company As 8in solution. This trans-lates into a price/mm2 about double. How-ever, since the 65nm chip was less than half

the size of the 0.18m design, the smallergeometry yields a small advantage. This is di-minished, however, when taking into accountthe higher costs for mask sets and any ROMcode changes at 65nm.

Ultimately, the two companies have achievedbroadly similar costs per chip. Arguably, com-pany A has realised its product for a lower ini-

tial financial commitment.However, company Bs investment in expertiseand infrastructure at 65nm, which should con-tinue to deliver benefits in future product gener-ations, may prove to be a more beneficial strat-egy in the longer term.

Node selection guidelinesIn this example, the 0.18m process was

able to achieve a competitive unit price. Thedesign contained a relatively high proportionof RF and analogue circuitry, and the digitalcircuitry was already well defined in relation

to mature governing standards.In other applications, particularly for emerg-

ing generations of smart products requiring in-tensive digital functionality integrated on the

chip, a more advanced node may present abetter option.On the other hand, it is important

to consider the size of thetarget market, bearingin mind that smallerprocess geome-tries requireprogres-

sivelyhigherproduc-tion vol-umes tooffsetthe in-creasedNRE

andwafercosts.

All ofthese factorsmust be consid-ered, and a deci-sion taken, before

the design can move for-ward, since the target nodehas an important influence on thechoice of foundry. It is vital to select a waferfoundry partner offering the best price andsupport at the chosen node. Node selectionmust also be determined before any thirdparty IP is sourced so that the detailed chipdesign can begin.

Analogue or digital miniaturisationThe relative proportions of analogue and RF

to the digital circuitry included in an SoC de-

asics



sign have an important bearing on the eco-nomics of migrating to a smaller process node.

Since digital feature size reduces accordingto a square function, halving the process

geometry yields features occupying a quarterof the previous die area.On the other hand, the same change

in design rule will yield savingsup to 50% for analogue

circuit features. Hence,systems containing

a significantquantity of ana-

logue or RFcircuitry willdeliver arelativelysmall re-turn in ex-changefor thehigher in-

vestmentand cost/

mm2 that areincurred at

progressivelysmaller process

nodes.

IP assessmentFor many projects at maturenodes, most or

all of the required IPblocks, for example,cell libraries, I/Osand memories, aswell as more complexblocks including PCI

Express or USB func-tions, can be pro-vided by the foundryat no extra charge.

However, for more advanced nodes, third-party IP is often needed.

Larger semiconductor companies may devel-op their own functions as required but today,

IP is more typically bought on the open mar-ket. If the functions needed are not availablefor the desired node, or if the costs are consid-ered excessive, a less advanced node may bea more viable choice.

As a guide, third party Flash IP blocks forleading edge nodes may cost in the region of$10~30,000, while more complex blocks suchas PCI Express functions, USB or advanced

memories such as DDR3 are typically priced$100~700,000.

Selecting IP may not be straightforward, par-ticularly for more advanced nodes. It is worthpointing out that the most advanced nodes,generally, do not benefit from a full portfolioof supporting IP, some custom IP developmentmay be necessary.

NRE & wafer pricing assessmentThe results achieved by the two example chip

design projects discussed earlier highlight howdiffering NRE costs and per wafer prices influ-ence the final unit cost for each die. Table 1compares the combined costs of NRE and maskset, as well as wafer prices, for currently activeprocess nodes from 0.18m~28nm. Its also

worth noting that todays most advanced nodesdeliver greater cost benefits at high volumes

Table 1: Comparison of NRE/mask costs and wafer price for foundry nodes.



significantly over 100 million units/yr.An accurate guide is to consider the most ad-

vanced nodes only for projects where very highproduction volumes are projected and the de-

sign contains a large quantity of digital cir-cuitry. For lower volume designs, or where asignificant proportion of analogue circuitry isincluded, a more mature process may result inthe lowest cost per unit.

Projected production volumeThe relatively easy accessibility of todays

more mature nodes, such as 0.18m and

0.13m, can allow fabless chip companies todeliver commercially viable custom chip solu-tions for lower volume sectors, such as someindustrial and medical markets.

Here, the high initial costs of the more ad-vanced nodes can demand production vol-umes well beyond the total available market inorder to become economically viable.

Negotiable factorsBefore committing to a node, serious dia-

logue with potential foundry partners is essen-tial. Better pricing may be available if thefoundry is particularly interested in the proj-ect: high projected production volumes mayallow the foundry to offer one of its more ad-vanced nodes at a competitive rate. Alterna-

tively, foundries may offer lower rates tolongterm, high volume customers.For this reason, foundries tend to favour

larger customers. An alternative is to gothrough a broker or consultant that can pro-vide the benefit of established relationshipswith the major foundries.

In practice, working through a consultingpartner can overcome many of the potential

pitfalls that can be encountered when workingwith an offshore foundry, across cultures andgeographical boundaries. With experience,some fabless chip companies may feel that

one particular foundry is easier or more reli-able to work with than some others. In fact,most foundries are equally reliable provided amutual understanding can be established.

Its also important to bear in mind that thechosen process will continue to maturethroughout the duration of the project, result-ing in improvements in aspects such as IPavailability as well as process yield. By thetime the project reaches production ramp-up,most foundries can be expected to achieve ac-ceptable yields at the target node. In additionfoundries will continue to develop new process

technologies, such that the most advancedprocess available at the beginning of the proj-ect will no longer be the most advanced bythe time full production is reached.

Node targetingThere is a case for targeting the most ad-

vanced node that is practicable at the time.On the other hand, a profitable result may be

achievable at a more mature node. The deci-sion is essentially commercially driven, basedon the projected sales volume and target sell-ing price.

Factors including the quantity of funding avail-able to start the project, the level of expertise inthe candidate nodes, and the corporateroadmap for future generations of the product

also have a critical impact on node selection.Arguably, todays mature sub-micron nodesallow fabless chip companies to realise newdesigns at a relatively low initial cost. On theother hand, choosing a more advanced nodemay deliver better results in the longer term,for those companies able to bear the higherup-front costs. TR

Shimon Raviv is Engineering Vice President atEquipIC supply chain. More from EquipIC Return to contents page.

asics
http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/EquipIC-supply-chain.asphttp://www.electronicspecifier.com/company/EquipIC-supply-chain.asphttp://www.electronicspecifier.com/company/EquipIC-supply-chain.asphttp://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/EquipIC-supply-chain.asp


19/48



Over the past 60 years, Potters Barbased Soundcraft and Studer havesteadily grown their dedicated audiomixing console design and manufac-

turing businesses, with global sales and sup-port networks. Its mixers are designed for livesound, recording, post-production, and TV andradio production.

This production facility is not currently used fora regular, high volume throughput of a stable,longterm product mix. Instead, its role is to pro-duce prototype and low production quantities of

a wide range of products as a service to Sound-craft and Studer as well as other companieswithin the Group, with volume production hand-led elsewhere. Typically, the facility processesbetween two and eight new products everymonth, for example, in-car microphones are cur-rently being prototyped for Harman AKG inAustria. PCBs for these products include compo-nents practically invisible to the naked eye.

This high and constantly changing product mixcreates complex issues, especially when some ofthe products are for sister factories with theirown part numbering systems. Speed and flexi-

bility are needed on theproduction lines to facili-tate rapid productchanges. To avoid bot-tlenecks, any test and in-spection facility has tobe equally as fast andversatile. When Sound-craft Studers x-ray in-

spection systems lack ofthese attributes became an issue, automated op-tical inspection, AOI, looked promising and sothe company invested in a Nordson YesTech BXbenchtop AOI system.

Adopting AOIAOI is fast, delivering cost-effective quality

control compared to in-circuit testing. Once a

system is programmed with the component, lay-out, orientation and soldering information, itsstrategy of inspecting by comparison to a li-brary of parts and packages allows rapid

test & measurement

AOI: Agile Optical InspectionWe sounded out one company about how it has adopted AOI

to achieve the inspection resource it needed.

Soundcraft Studer, part of Har-man International, manufac-tures advanced audio mixing

consoles on its SMT production lineand also produces PCBs for other

Harman companies. Production is lowvolume, high mix and constantlychanging so its inspection needed tobe fast and agile, responding to rap-idly changing production needs.



throughput of boards during production.Equally essential to Soundcraft Studer, however,was the BXs ability to be quickly reconfiguredand even reprogrammed fast enough to keep

pace with different areas demands for inspec-tion resource.As the company adopted the BX system, it

found that the time needed for programming newboards was heavily influenced by the availabilityof component data. Lead times for new pro-grammes were reduced as the component librarydeveloped and grew. Developing this componentlibrary called for careful management of the part

numbering for these components, as up to threenumbers related to eachone: the original manu-facturers part number; the part number assignedby Soundcraft Studer and if applicable, the partnumber assigned by the relevant sister company.

As with any AOI sys-tem, other parametersapply to each part too:board reference; its X

and Y co-ordinates; ori-entation and packagecode. These, however,are available from theCAD files used by thepick& place machines.Accordingly, the com-pany built a spreadsheet

database that integratedCAD data with part num-ber information, together with storage location.This comprehensively defined each componentused in PCB production and integrated with theBX library database.

Operating experienceAs the database has been growing, set-up

times for entire boards have got faster. For ex-ample, a board was programmed within 25minof receiving the CAD data even though it con-tained 123 new components not yet in the



library. After set-up, inspection has proved to befast too. An inspection of a double sided boardafter solder pasting took 23s to determine thatthe board was ready for soldering. Component

presence and position were checked.The machine also uses OCV to check polarityfor each component and also that it is the rightpart by reading its label. Another useful featurewas the AOIs ability to check componentheight above the board, as an over-height com-ponent could indicate a problem such as asmaller part trapped beneath a BGA being in-spected. A further 23s inspection after reflow

was sufficient to check for shorts, dry joints,lifted leads and other solder faults.

Soundcraft Studer mounted its AOI benchtopmachine on a trolley so it can be rapidly movedto any point in the production line where inspec-tion is required. This is a significant improvementon the earlier situation in which inspection wasperformed by an x-ray machine permanently inposition after reflow. And with just x-ray inspec-

tion, it was not possible to detect incorrect com-ponents or orientation, and solder is oftenmasked by components in the same position onthe top and bottom.

Batch sampling has proven to be well worth-while with the new system as it can catch prob-

lems within the manufacturing process that gobeyond a single board failure.

With smaller batches 100% inspection is bestas, for example, it is possible for a component

reel loaded into a pick&place machine to bepopulated with mixed components when kittingsmall lots. This fault would not be visible, andthe pick&place machine would be unable to de-tect it as it only reads the barcode label on thereel which would probably be as expected.However, a series of boards rejected by theAOI system for the same wrong componentfailure would soon alert an inspector to the exis-

tence, nature and source of the fault.Two months down the line following installa-

tion and Soundcraft Studer has found the AOIto provide the inspection resource essential forits low volume, high mix production environ-ment. Dave Birtwistle, SMT Team Leader atSoundcraft Studer says, We appreciate thebenchtop systems agility as well as its speed, asthis allows its rapid reconfiguration for different

product types and process stages, and it is easyto program as well as fast in operation. Aboveall we can report that to date it has found everyfault it has encountered! TR More from YesTech Return to contents page.

test & measurement
http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/YESTech-Europe.asphttp://www.electronicspecifier.com/company/YESTech-Europe.asphttp://www.electronicspecifier.com/company/YESTech-Europe.asphttp://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/YESTech-Europe.asp


23/48

Open-frame & Enclosed 5 - 3000 Watts

PCB or chassis mount

Ultra-compact design

Industrial & medical approvals

Green Power versions

DC-DC Converters 0.25 - 60 0 Watts

Regulated & unregulated versions

2:1 & 4:1 input ranges

Industry standard SIL, DIL

& SMD packages

External Power Supplies 8-250 Watts

Energy eff iciency level V

Industrial & medical approvals

Compact high efficiency design

Optional class II

DIN Rail Mount 5-960 Watts

Single or 3 phase input versions

DC OK signal & LED indicator

Rugged design for industrial

applications

Redefining

Industry Expectations

Visit our website to request a copy of ournew 2011/12 Power Supply Guide and

see our complete line of power products.

GREENPOWER



Measuring energy usage isnt fun-damentally difficult but creating a ro-bust infrastructure around smart en-

ergy requires many elements to cometogether. The underlying technology isdeveloping rapidly to enable the smartgrid and the physical devices put in placeover the next three to five years will needto remain in operation for many more: up to 20years and possibly longer.

Against todays fast moving technologicallandscape it would be difficult to imagine many

electronic devices that would stand that test oftime but smart energy measurement and controlsolutions have to meet this challenge.

Trends within the industry strongly indicate that

wireless communication between energy awaredevices promises the most viable solution interms of longevity and functionality. And this isreflected in the efforts of the ZigBee Alliance

which has worked co-operatively to develop theZigBee Smart Energy Standard, now

widely trialled byutility providers around the world in prepara-

tion for large scale roll-out.As this roll-out begins, demand will rise for sys-

tems that implement the Smart Energy Stan-dard. In parallel to ZigBee certification, these

solutions will also need to meet the commercialrequirements of a competitive market. To ad-dress these needs ByteSnap Design has devel-oped and launched the industrys first ZigBee

smart metering

Building A Smarter MeterThe move to smart energy brings massive opportunity across the supply

chain but choosing the right solution needs careful consideration atevery level, says ByteSnap Design Director Dunstan Power.

As consumers become more aware ofenergy management, the need to ac-curately measure and react to energy

consumption becomes greater. Evidence of in-

creased energy awareness is all around us, inpublic places, commercial buildings and ourown homes. The need for greater control overutilities such as gas, water andspecifically electricity, has led tothe emerging concept of smartenergy and the smart grid.The active management ofenergy consumption is a

core element of the smart en-ergy concept and deliveringthis management requires an ef-fective means of monitoring en-ergy consumption.



Smart Energy module which integrates meteringand control functions in a single, ZigBee compli-ant platform.

ZMM-01 provides a pre-integrated and pre-cer-

tified solution targeting OEMs and system integra-tors looking for a cost-effective solution to imple-menting a smart energy solution based on theZigBee Smart Energy Standard. Using industryleading discrete solutions from Ember and CirrusLogic, ZMM-01 enables system integrators tobring solutions to market faster and with signifi-cantly less development expenditure.

Creating a robust wireless connection is inher-

ently difficult and gaining compliance testing for

a wireless device can be both ex-pensive and time consuming. The ZMM-01 re-moves this pressure by delivering a fully inte-grated and compliant platform on which tobuild smart energy solutions. By implementingmetering and control functionality on a singlemodule, system integrators are able to target arange of applications where the ZigBee Smart

Energy protocol will be used.The modular design overcomes two significant

development challenges: developing a ruggedanalogue front end, AFE, for measuring mains

levels of voltage and current with creating areliable radio interface implemented using theZigBee protocol over the underlying IEEE802.15.4 layer.

The AFE is a Watt/hour meter on a chip, de-veloped with smart energy metering in mind.The AFE accepts two current and two voltage in-puts, with an energy linearity of 0.1% of read-ing over 1000:1 dynamic range. RMS voltageand current calculations are made on-chip, giv-ing active, reactive and apparent power/energydata with system level calibration and phasecompensation. The device meets the IEC, ANSI

and JIS accuracy specifications and also pro-vides voltage tamper correction and a powersupply monitoring function.

Based on the Arm Cortex-M3 core, the Zig-Bee SoC from Ember delivers ample

processing power and memorybandwidth for the implementation of

customer specific applications. Be-spoke customisation is enabled though

ByteSnaps ZDM-01 development kit whichprovides the perfect development platform

for end products based on the ZMM-01 module.It includes an optically isolated USB serial port aswell as a mains load simulator, giving the abilityto vary the current, voltage and phase anglewithout working with the mains.

Energy measurement and control are inti-

mately linked functions of the smart energy con-cept: ZMM-01 is the only solution that success-fully combines both in a cost-effective, lowpower platform, says ByteSnap. The potentialfor smart energy measurement combined withcontrol is only just being explored and as such,ZMM-01 is at the sharp end of integration, pro-viding a flexible wireless platform that can beconfigured to meet todays requirements whilst

providing the required reliability and robustnessto remain in service for many years to come. TR More from ByteSnap Design Return to contents page.

ByteSnaps developmentkit and module.
http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/ByteSnap-Design.asphttp://www.electronicspecifier.com/company/ByteSnap-Design.asphttp://www.electronicspecifier.com/company/ByteSnap-Design.asphttp://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/ByteSnap-Design.asphttp://www.electronicspecifier.com/



A thorough investigation of the way the bea-cons were being used was undertaken, and theproblem finally identified by Intelliconnect. Priorto the aircrew bailing out of an aircraft, theywould be inclined to twiddle with the PLB an-

tenna and in some cases accidentally disconnectit from the beacon. On bailing out, the PLB wouldthen fill with seawater, leaving the crew memberwondering why rescue took longer than antici-pated. With the PLB full of water its ability totransmit its location was severely limited and insome cases could even cease to operate.

So the challenge was set in providing a matedpair of connectors that would be waterproof to

minimum IP68/NEMA 6P ratings for up to 20mimmersion for a minimum period of four hoursunmated. And as usual with Intelliconnect de-signs, the company took its holistic approach,

with the complete system investigated.To provide a sealed connector, complete with

modularity and ease of assembly, a TNC toMCX adaptor was designed by Intelliconnect.This enabled a separate umbilical link to beused inside the equipment which greatly im-proved the flexibility of the manufacturingprocess. The mating half needed careful consid-eration, says the company, as the personal loca-tor beacon had to be used in confined spaces.So a low profile, TNC right angled plug, also

sealed to IP68, needed to be developed.A further part of the design brief required In-

connector solutions

Holistic Approach Is Beacon

For Waterproof ConnectorsIntelliconnects new range waterproof connectors were conceived, saysMD Roy Philips, when a Personal Locator Beacon manufacturerpresented a problem that no one else had managed to solve.

When a market leading manufacturerof personal locator beacons, PLBs,

was faced with a problem whenmilitary aircrew were using its beacons, thecompany called in Intelliconnect to help solvethe problem. Simply, the problem arose in anon-functioning beacon when crewman hadexited the aircraft or boat. But tracking downthe problem proved difficult to understand as,like any other piece of emergency equipment,these PLBs were subjected to rigorous testing

during manufacture and also a further fullfunctionality shake-down test prior to the manu-facturer despatching the equipment.



telliconnect to take into account the need to at-tach additional GPS and transmitting antennaswithout them being connected, inadvertently orotherwise, to the wrong ports. For this require-

ment, a solution was developed using standardand reverse polarity TNC connector designs thatwould clearly indicate if attempts were made towrongly connect them.

The final part of the system to be protectedwas the battery connector. By using the experi-ence and knowledge gathered on the antenna

connector design, it was a relatively simple taskto apply this know-how to Intelliconnects Nim-

Camac connectors and also seal them toachieve the necessary IP68 rating unmated.The resultant Intelliconnect solution emerged

from the companys holistic approach to solvingthe problem, says the company, rather than justaddressing the connectors, hence drawing onthe companys vast experience in designing forharsh environments.

The final solution featured a combination of O-

rings and proprietary methodology to protectthe connector, and thus the PLB, from moistureingress under seawater at depths down to 20m.

Notably, rigorous testing by Intelliconnect and

the PLB manufacturer proved that the new con-nector design was able to protect the equipmentto an equivalent depth of 60m, exceeding theoriginal product requirements and providing the

customer with a healthy threefold safety factor.The standard method employed to test con-nectors is pressurisation of the connectors usingcompressed air, followed by immersion inwater and subsequently checking for air bub-bles as a sign of leakage. This is simple enoughto be used as an in process operation and canquickly and easily highlight any process or as-sembly problems.

In addition to the sealing requirements, the re-sultant connectors are designed to meet all therelevant Mil specs, durability requirements, elec-trical, mechanical, and environmental conditionsthat all standard military equipment needs towithstand to be able to operate in an airborneor seaborne environment.

And of course now, Intelliconnect has incorpo-rated this waterproofing process for TNC connec-

tors into many of its other products including theN series, SMA, and custom designs such as theABMS, an MCX derivative used for a cochlearimplant, and the MPNC four way multi-pin con-nector for environmental corrosion monitoring.

These connector and adapter devices nowform the companys Pisces range and otherapplications include traffic monitoring systems,

rail traction systems, military and aerospaceapplications, medical, marine radar, batterypower, and corrosion monitoring for the oiland gas industry.

Intelliconnect is a UK manufacturer of RF con-nectors and cable assemblies, based in Chelms-ford and with facilities in the USA. It specialises inthe design and manufacture of waterproof con-nectors, with the capability to design, develop

and supply interconnect solutions to customerdriven specifications in weeks not months. TR More from Intelliconnect Return to contents page.
http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/Intelliconnect--Europe-Ltd.asphttp://www.electronicspecifier.com/company/Intelliconnect--Europe-Ltd.asphttp://www.electronicspecifier.com/company/Intelliconnect--Europe-Ltd.asphttp://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/Intelliconnect--Europe-Ltd.asphttp://www.electronicspecifier.com/



Energy harvesting alone often does not pro-duce sufficient power to continuously run the sen-sor transmitter energy harvesting can produceabout 1~10mw, where the active sensor transmit-

ter combination may need 100~250mw. Har-vested energy must be stored when possible,ready for use by the sensor/transmitter, whichmust operate at a duty cycle that does not ex-ceed the energy storage capabilities of the sys-tem. Likewise, the sensor/transmitter may need tooperate at times when no energy is harvested.

Finally, if the stored energy is depleted and thesystem is going to shut down, the system may

need to carry out housekeeping tasks first. Thismay include a shutdown message or storing in-formation in non-volatile memory. Thus, it is im-portant to continuously gauge available energy.

Complete energy harvesting systemFigure 1 shows a complete system implementa-

tion using an LTC3588-1 energy harvester and

buck regulator IC, two LTC4071 shunt batterychargers, two GM Battery GMB301009 8mAhbatteries and a simulated sensor transmittermodelled as a 12.4mA load with 1% duty cycle.

power

True Grid IndependenceLinear Technologys George H Barbehenn outlines the companys robust

energy harvesting system for wireless sensors using a piezoelectric energyharvesting power supply and Li-Poly batteries with a shunt charger.

T

heres an emerging and potentially largemarket for wireless sensors. By their verynature, wireless sensors are chosen for

use in inaccessible places or for applicationsthat require large numbers of sensors: toomany to easily hardwire to a data network.

In most cases, it is impractical for these systemsto run off primary batteries. For example, a sen-sor for monitoring the temperature of meat as itis shipped would need to be mounted in a tam-perproof way. Alternatively, HVAC sensorsmounted on every source of conditioned air

would be far too distributed to feasibly use bat-teries. In these applications, energy harvestingcan solve the problem of providing power with-out primary batteries.



The LTC3588-1 contains a very low leakagebridge rectifier with inputs at PZ1 and PZ2 andoutputs at Vin and GND. Vin is also the inputpower for a very low quiescent current buck

regulator. The output voltage of the buck regula-tor is set by D1 and D0 to 3.3V.The LTC3588 is driven by an Advanced Cera-

metrics PFCB-W14 piezoelectric transducer, ca-pable of generating a maximum of 12mW. Inour implementation, the PFCB-W14 providedabout 2mW of power.

The LTC4071 is a shunt battery charger withprogrammable float voltage and temperature

compensation. The float voltage is set to 4.1V,with a tolerance on the float voltage of 1%,yielding a maximum of 4.14V, safely belowthe maximum float allowed on the batteries.The LTC4071 also detects how hot the batteryis via the NTC signal and reduces the floatvoltage at high temperature to maximise bat-tery service life.

The LTC4071 is capable of shunting 50mA in-

ternally, says Microchip. However, when thebattery is below the float voltage, theLTC4071 only draws ~600nA from the bat-tery. The GM Battery GMB301009 batterieshave a capacity of 8mAh and an internal se-ries resistance of~10.

The simulated sensor transmitter is modelledon a Microchip PIC18LF14K22 and MRF

24J40MA 2.4GHz IEEE standard 802.15.4radio. The radio draws 23mA transmitting and18mA receiving. The model represents this as a12.4mA, 0.98% duty cycle (2ms/204ms) load,set with a self-clocked digital timer and a MOS-FET switching a 267 resistor.

Modes of operationThis system has two modes of operation:

charging sending and discharging sending. Incharging sending mode, the batteries arecharged while the sensor transmitter presents a0.5% load. When discharging, the sensor

Figure 1: Complete piezo based energy har-vesting system independent of the grid. This de-

sign uses thin film batteries to gather energycollected by the piezo for a wireless sensortransmitter, which operates on a 1% duty cycle.



transmitter is operating but no energy is beingharvested from the PFCB-W14.

Charging sending

When active, the PFCB-W14 delivers power atan average of around 9.2V 180A 1.7mW.The available current must charge the batteryand operate the buck regulator driving the sim-ulated sensor transmitter. The active sensortransmitter draws 12.4mA 3.3V 41mW ataround 1% of the time, or about 0.41mW onaverage, leaving some current to charge the

battery. Tak-

ing into ac-count the85% effi-ciency of theLTC3588 buckregulator, as-suming an av-erage Vin of9.2V, as

shown in Fig-ure 2, and abuck quies-

cent current of 8A, the average current con-sumed by the system without charging the bat-tery is:

Harvested energy can drive the sensor trans-mitter at a 0.5% duty cycle with about 120Aleft to charge the batteries. The GMB301009batteries have a capacity of 8mAh, so they com-pletely charge from empty in about 75 hours.

Discharging sendingWhen the PFCB-W14 is not delivering power,

the voltage at Vin drops to approximately:

So the reflected load current calculation

changes to:

The quiescent current of the buck regulator ishigher because the regulator must switch moreoften to regulate from 7.5V versus 9.2V. At78A, with no energy harvested, the battery isdischarged in around 115 hours. This indicates a

charge storage capacity of >8.95mAh. Thesebatteries when brand new could store approxi-mately 12% more charge than rated.

A more serious problem is what happenswhen the battery is fully discharged. If current isdrawn after the state of charge reaches zero,and the battery voltage drops below 2.1V, thebattery is permanently damaged. Therefore theapplication must ensure that the battery voltage

never falls below this limit. For this reason, bat-tery cut-off is set to 2.7V or 3.2V to ensuresome energy remains in the battery after thedisconnect circuit has engaged.

Simply stopping the transmitter or disconnect-ing the load will not protect the battery, as theLTC4071 draws a quiescent current of approxi-mate 600nA. Although this is extremely low, thetotal load, including the LTC3588V1, is nearly2A. A fully discharged battery will only beable to supply approximately 100A before itsvoltage drops enough to damage the battery.

A disconnect circuit is necessary to ensure thatthe battery does not discharge in a reasonableamount of time. The LTC4071 provides an inter-nal low battery disconnect circuit. This disconnectcircuit was measured to provide



nect state before the battery is damaged.

In Figure 3, the second battery (BAT2) is seento disconnect 50 hours after BAT1 due to the2A load.

Measured resultsThe system shown in Figure 1 was measured in

both operating modes: discharging-sending (Fig-ure 3) and charging-sending (Figure 4).

Discharging sendingIn Figure 3 the voltages of the two batteries

BAT1, BAT2 and VBuck are plotted against timewith the batteries supplying all the system en-ergy, none from the PFCB-W14 piezo.

The batteries slowly discharge until BAT2 acti-vates the LBO threshold of the LTC4071, where-upon the disconnect circuit activates and discon-nects BAT2 from all circuitry except U5. Thiscauses the voltage at Vin of the LTC3588 todrop below the UVLO for the regulator, and theregulator shuts off.

The load on BAT1 is the 2A quiescent currentof the LTC4071 and the LTC3588. This smallload slowly discharges BAT1 until the low bat-tery disconnect of LTC4071 is activated andBAT1 is disconnected.

Charging sendingWhen the PFCB-W14 once again starts deliver-

ing power to the system, Vin rises to 7V, which for-

ward biases the body diodes of the LTC4071s

disconnect FETs . This charges the batteries untilthe reconnect threshold is reached, allowing bat-teries BAT1 and BAT2 to be reconnected. Lookingat Figure 4, this can be seen as the voltage at V insnaps down to the battery stack voltage.

Since the voltage at Vin is now VBAT1 +VBAT2 + (180a15k) = 6.2V, the buck regula-tor on the LTC3588 restarts and 3.3V is onceagain available.

ConclusionWith a few easy to use components, it is possi-

ble to build a complete compact energy harvest-ing power sub-system for wireless sensor trans-mitters. In this particular system a piezoelectrictransducer supplies intermittent power, whiletwo batteries store energy for use by the sensortransmitter. An integrated disconnect switch pro-tects the batteries from over-discharge.

This system can fully charge the battery in 75hours, even while operating the sensor transmit-ter at 0.5% duty cycle.

The batteries allow the system to continue op-erating the sensor transmitter at 0.5% duty cyclefor 115 hours after the PFCB-W15 stops provid-ing power. If longer battery operating time is re-

quired, the sensor transmitter duty cycle can bereduced to accommodate this need. TR More from Linear Technology Return to contents page.

Figure 3: Discharge with battery under-voltage disconnect.

Figure 4:Battery disconnect recovery on charge.
http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/Linear-Technology--UK--Limited.asphttp://www.electronicspecifier.com/company/Linear-Technology--UK--Limited.asphttp://www.electronicspecifier.com/company/Linear-Technology--UK--Limited.asphttp://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/Linear-Technology--UK--Limited.asphttp://www.electronicspecifier.com/



been made to bring the inherent capabilities of

this exciting material system to bear in practicalsemiconductor power devices. The combinationof high breakdown field strength due to thewide band gap of the III-nitrides, high electron

mobility, as well as an unusually high channelelectron density yield a remarkably com-pelling drift resistance.

Such devices also benefit from the reducedgate charge requirements involved in switch-

ing the devices on and off. Probably the mostexciting attribute of the system involves theeasily isolating nature of the inherently lateraldevices, permitting unprecedented monolithic

power

Since the advent of the spontaneousAlGaN-GaN based high electron mobil-ity sheet formation, first discovered by M

Asif Khan in 1991, significant effort has

Driving Revolution

In Power Electronics

Michael A. Briere discusses the first commercially viable application ofGaN based power device technology, heralding a new era in power

electronics through the combination of high performance and

competitive costs for semiconductor power devices.



integration of power systems. From atechnologists point of view, it is unfor-tunate that, as in many such inherentlysuperior technological alternatives, eco-

nomics more than physics will deter-mine the rate and extent of adoption ofGaN based power devices.

The driving metric for adopting newsemiconductor technologies is perform-ance/cost. This is the same as used in thepopular Moores Law for data process-ing technology. It is therefore imperativethat focus be placed not on performance

alone but at least equally on the fundamentalfactors that drive the economics of the techno-logical proposition. In order to beat silicon, it isbest to stay as close to the incumbent silicon plat-form as possible, only deviating when required.

In the case of GaN based power devices,one of the single greatest factors affecting costis the choice of substrate material. As theadoption of any new device structure must

compete with that of the incumbent, the cost ofsubstrate and epitaxial growth should not farexceed the cost for a comparable fully fabri-cated silicon device.

This sets an upper limit of about$3/cm2 for the combined cost of sub-strate and epitaxial layer. Only siliconsubstrates meet this requirement. The use

of silicon substrates for GaN hetero-epi-taxy present several significant technicalchallenges due to large mismatches inthermal co-efficient of expansion as wellas lattice constants of the materials. As atechnologist, such a challenge must bemet and not avoided.

In addition, device fabrication process-ing costs must be competitive with

CMOS based silicon processing oftenperformed in large scale foundries pro-cessing >20,000 wafers/week. There-fore, in order to achieve the required

performance/cost, GaN based devices shouldbe processed in the same high volumefoundries, using as much of the same highthroughput, high yielding process technologiesas used for silicon based devices as possible.Today, this requires that the substrates be atleast 150mm, preferably 200mm.

Device performance requirements are not lim-ited to such figures of merit as on-resistance or

gate charge. Leakage currents, for instance,should achieve the performance levels of incum-bent silicon devices. This is to say that the

Figure 2: Forward biased SOA for low voltage GaNbased power devices intended for 12Vin power con-version applications.

Figure 1: Id (A/mm) vs Vds (Volts) for early HV GaN-powIR HFET, Vgate= 10V.



criteria for both breakdown voltage and leak-age currents (gate and source to drain) must beadjusted down from that commonly used whendiscussing GaN based devices of mA/mm of

gate periphery to a more practical0.1~1A/mm. In addition, device quality, stabil-ity and reliability are all to be considered firstorder requirements for successful commercialisa-tion of any new technology, with the incumbentsilicon platform setting the standard.

A technological platform thatmeets the performance and cost re-quirements described above has

been developed by InternationalRectifier, referred to as GaNpowIR.This platform uses multi-waferMOCVD reactors to grow GaNepitaxy on standard 675m thick-ness 150mm silicon substrates.Wafer bow for 2m of epitaxialfilms is routinely less than 20m.The device processing is fully com-

patible with silicon CMOSfoundries and is based on thin filmphoto and etch lithography. Theohmic contacts are formed withoutthe use of gold metallurgy.

Epitaxial quality is excellent, ri-valling the published performance ofany such film. Fabricated low volt-

age devices routinely exhibitIon/Ioff ratios of >1012 with leak-age currents less than 0.1pA/mm. Infact high voltage devices have beenroutinely shown to exhibit Ion/Ioffratios in excess of 107, where Ioff ismeasured at 600V at less than0.1A/mm. In addition, the use ofan insulated gate provides gate

leakage currents of less than0.1pA/mm as shown in Figure 1.

In addition, excellent deviceruggedness in the form of forward bi-

ased safe operating area, shown in Figure 2,and longterm stability of parametric performancesuch as Rdson as well as drain and gate leakagecurrents have been demonstrated to >6000hr. In

fact, to date, over 3million device hours underaccelerated stress conditions have demonstratedstable parametric behaviour for initial GaNbased devices. See Figures 3 & 4. TR More from International Rectifier Return to contents page.

power

Figure 3: Stability of gate leakage current over 3000hr with

50V applied to the 8.5V rated gates at 150C.

Figure 4: 30V GaNpowIR HEMT device stability over

3000hr under reverse bias stress with Vdrain-source of 26V,Vg=- 7 V and at 175 C. Ids measured at room temperaturefor Vds=14.5 V and Vg = 7V. Wg= 2550mm, Lg=0.3m.
http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/International-Rectifier.asphttp://www.electronicspecifier.com/company/International-Rectifier.asphttp://www.electronicspecifier.com/company/International-Rectifier.asphttp://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/International-Rectifier.asp


35/48

)01%-06%0)643:)56309)'38/

!!

:::43:)56309)'38/

>"%8735%2+-2+-2487

>####38748743:)55%7-2+

>-2+0)""35"%(.867%&0)3874879307%+)

>3:)533()137)2**

>8++)(1)7%0'%6)*35,%56,-2(8675-%0)29-5321)276

>$')57-*-'%7-32*35,%=%5(38603'%7-326347-32%0>((32*82'7-3213(80)6*355)(82(%2'



Whilst the LEDs themselves are highly effi-

cient, the benefits of this can be lost if energyefficiency is not optimised in the power supply.Correctly installed, the lifetime of an LED lightcan be 35 years or more but even a well de-signed power supply working within its operat-ing parameters will last only 10~12 years. Apoorly designed, lower efficiency power sup-ply unit running at high temperature will noteven last half as long.

Designing a power supply unit to achievemaximum operating efficiencies for LED lightingapplications can be a daunting task. Designershave to consider a wide range of issues when

conceiving power solutions to navigate the reg-ulations for LED installations and ensure apower supply with maximum service life.

RegulationsDue to specific operating voltages and the

configuration in which the LEDs are arranged,a standard power supply cannot be used withLED lighting. LED power supplies are gov-erned by specific legislation: EN61347-1/ -2-13, UL8750, UL1310 Class 2 for safety,EN55015 for EMC and EN61000-3-2 Class C

for harmonics. But not all these specs are rele-vant to lighting applications.

In addition, the EU Ecodesign Directive2009/125/EC sets out rules for reducing the

powering light

Efficient Power

Solutions ForLED Lighting

InstallationsCor van Dam, Avnet Abacuss Market-ing Director, reviews best practice indeveloping efficient power supplysolutions for solid state lighting.

The LED lighting market looks set to ex-plode, with some sources predicting thatin five years time 50% of the global

lighting market will be fulfilled by solid statesolutions. Whether street or office illumina-tion, refrigeration or architectural lighting,every solid state light needs power. And thepower supply is often the weakest link.



environmental impact of Energy using Prod-ucts, EuP and other Energy related Products,ErP, at all stages from design concept to manu-facturing. Street, office and domestic lighting

all fall within this directive, along with their re-spective power supplies.This legislation can sometimes be compli-

cated to interpret for lighting designers andLED engineers entering the market, so dis-cussing specific requirements with a specialistdistributor such as Avnet Abacus with its com-prehensive power portfolio and consultativeapproach can be valuable at the design stage

to ensure compliance from the start.In addition, a good distributor will work with

its suppliers to keep up with future regulationsand also participate in regular training.

Operating EfficienciesThe illumination output of solid state systems

depends on how the LED array is driven. Forexample, using a constant current power sup-

ply produces the lowest cost and highest effi-ciency solution but unbalanced current foreach parallel branch of LEDs can lead to un-even brightness and shortened life.

On the other hand, using a constant voltagepower supply with an LED driver IC will accu-rately control the current through the LEDs,delivering a more uniform light output andlong life, but at a reduced efficiency and a

higher cost.Designers will need to consider what applica-

tion they are designing for and if a return oninvestment could be achieved by using a moreexpensive but more uniform driver solution togive extended life, instead of a low cost, re-duced life option.

For small and medium power LED Lightingsystems, for example, sophisticated semicon-

ductor solutions from Power Integrations couldbe specified. PI leads on technology solutionsto design customised LED power supplies withthe minimum of component with their perfectworking software tools.

This could be a low cost option for largevolume applications, and one benefit, for ex-ample, is that the circuit can be trimmed for

the highest efficiency.However, this is not always the option withthe fastest turnaround time, given the approvalprocess required for some regulations.

The alternative is a ready to use, high qual-ity, highly efficient off the shelf solution result-ing in a faster time to market.

Typically the company has over 500 differ-ent models available with constant current out-

put power of 3~350W, and a growing num-ber with dimming options for use inapplications such as architectural lighting. Inaddition, a wide range of LED power sup-



plies with constant voltage output are available, rangingfrom 10W to thousands of Watts, suitable for low to highpower LED lighting systems and screens.

Maximising service lifeMuch has been written about the longevity of LEDs but it isimportant to be aware that they are not indestructible. A sim-ple voltage glitch has the capacity to damage the LED. Asidefrom that, the main area affecting service life of both thepower supply and LED lighting is that of excess heat.

Although the LED devices run at low power, the build-up ofheat in the array as well as in the power supply needs to beminimised by using heat sinks on mountings for example. Not

only will the lighting unit run more efficiently at a lower tem-perature but the risk of damage to the housing holding theLEDs will be significantly reduced.

Designers should be integrating heat dissipation solutionsinto the power supply design from the outset in order to max-imise service life. Consideration should also be given towhere and how the lighting units will be mounted to aid thecooling process.

Another issue to weigh up is minimising weak points within

the power supply assembly itself. Components such as elec-trolytic capacitors and opto couplers are known to impact onservice life and reliability. Increasingly, designers are specify-ing power supply solutions without these items.

In addition, care should be taken to ensure that high qualityconstruction methods are used to prevent against the potentialdegradation caused by heat, aggressive moisture ingress andUV light. A dry solder joint from poor quality construction meth-

ods could prove critical once the power supply is in the field.Selecting a power supply for LED lighting applications is notstraight forward, and not everyone will have the in-depthknowledge of the rules and regulations that must be compliedwith in each specific market.

In order to achieve the longest life cycle possible with LEDlighting, it is crucial to get the right advice and specify and se-lect the correct power supply from the many solutions avail-able and install it correctly. Only when all these things have

been taken into consideration can you expect to achieve themost efficient solution possible. TR More from Avnet Abacus Return to contents page.

powering light

Power Integrations LinkSwitch PH & PL.
http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/Avnet.asphttp://www.electronicspecifier.com/company/Avnet.asphttp://www.electronicspecifier.com/http://www.electronicspecifier.com/http://www.electronicspecifier.com/company/Avnet.asphttp://www.electronicspecifier.com/



AS one of Europes leading power, interconnect,

passive, and electromechanical distributors, AvnetAbacus is teaming up with some of the worlds lead-ing manufacturers of power components to deliver aseries of free technical sem

Electronicspecifier Design 201108

Documents

Transcript of Electronicspecifier Design 201108