17987-100302di

Upload: ankita-verma

Post on 07-Apr-2018

214 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/4/2019 17987-100302di

    1/6

    Acurrent-feedback amplifier is awell-known componentwith many uses. Its basic

    block diagram shows that its input stageis a voltage followerin practice, a sym-metrical emitter follower (Figure 1).The

    configuration samples the output cur-rent, converts it to voltage across a largeimpedance, and amplifies it to the outputusing a high-power, low-output-imped-ance amplifier. The idea is to use the am-plifiers input stage as a voltage followerin a basic Colpitts oscillator. This circuituses the noninverting input of the cur-rent-feedback amplifier as the followerinput and the inverting input of the am-plifier as the follower output.You use theoutput amplifier to obtain a relativelyhigh-power buffered output. The circuit

    in Figure 2 shows a basic Colpitts oscil-lator that uses the amplifiers input-volt-age follower as the active element of theoscillator.

    Take note of two aspects of this oscil-lator circuit: First, back-to-back diodesconnect across the resonator to limit theoscillations to a specific level, thus main-taining the linearity of the voltage fol-lower. Second, the voltage follower out-put connects to the resonator tap through

    resistor ROSC

    to improve the linearity anddefine the feedback magnitude. The val-ue of R

    OSC, 330, lets you obtain soft

    clipping operation of the diodes acrossthe resonator (V

    RES1V p-p, which is

    0.5V peak across each diode). Figure 3shows V

    RES, the measured voltage at the

    top of the resonator. RF

    is the amplifiersfeedback resistor; the amplifiers manu-facturer recommends its value. This de-sign uses the LM6181 from NationalSemiconductor (www.national.com),and the value of R

    Fis 1 k.

    It is easy to calculate the output volt-age:V

    RES1V p-p, and V

    INVV

    RES1V p-

    p. The voltage-buffer gain is unity:

    V(ROSC

    )VINV

    VRES

    /2.The voltage at theresonator tap is V

    RES/2, because the res-

    onator capacitors are equal in value.V(R

    OSC)V

    RESV

    RES/20.5V p-p.I(R

    OSC)

    V(ROSC

    )/ROSC

    .I(RF)I(R

    OSC). The neg-

    ative feedback nulls the amplifiers in-

    verting-input current. VOUTV(RF)VINV

    RFI(R

    F)V

    INV1000(0.5/33

    0)12.51V p-p.If you need more volt-age, you can add R

    Gin this case,

    100from the inverting input toground. I(R

    G)V

    INV/R

    G. Now, the cur-

    rent through RF

    is the sum of the currentsthrough R

    OSCand R

    G. So, V

    OUTV(R

    F)

    VINV

    RFI(R

    F)V

    INV1000(0.5/33

    01/100)112.51V p-p. Figure 4shows the measured output voltage.

    The LM6181s maximum output cur-rent is 100 mA,so it can easily drive a cur-

    rent of63 mA p-p (6.3V/100) intoa total load of 100 (50 output-termi-nation resistor and 50 load resistor).The voltage across the 50 load is 3.15Vpeak, or 2.23V rms, which is close to 20dBm (100 mW).This power level can di-rectly drive high-level diode double-bal-anced mixers, or it can drive a higherpower amplifier while delivering a cleansinusoidal waveform.You can modify theresonator circuit to accommodate differ-

    IEZT

    VN

    VP

    VOUT+1

    IE

    +

    Figure 1

    RF oscillator uses current-feedback op ampVictor Koren, Tioga Technologies Ltd, Tel Aviv, Israel

    In a typical current feedback amplifier, the

    input stage is a voltage follower.

    RLOAD50

    RTERM50VOUTVRES

    VINV100 nF

    1 nF

    100 nF

    1 nF

    100 nF

    RF1k

    RG100

    ROSC330

    BAV99

    0.8 H

    +

    + +

    LM6181

    12V

    12V

    F igure 2

    This Colpitts oscillator uses a current-feedback amplifier to provide a clean sinusoidal output.

    www.edn.com O ctob er 3 , 2002 | ed n 83

    ideasdesign

    RF oscillator usescurrent-feedback op amp ............................83

    Simple tester checks LCDs............................84

    Circuit drives mixed types

    and quantities of LEDs..................................86

    MOSFET serves as ultrafast

    plate driver ......................................................88

    Parallel port provides high-

    resolution temperature sensing..................90

    Publish your Design Idea in EDN. See theWhats Up section at www.edn.com.

    Edited by Bill Travis

  • 8/4/2019 17987-100302di

    2/684 edn | O ctob er 3 , 2002 www.edn.com

    ideasdesign

    Manufacturers of electronicequipment use LCDs for calcula-tors, watches, mini-videogames,

    and pagers, for example. In com-parison with LED-based displays,which consume power on the order oftens of milliwatts, an LCD consumes onlya few microwatts. The LCD thus savespower by a factor of approximately 1000.Checking an LED is as simple as check-ing a semiconductor diode but, in thecase of LCDs, involves some added com-

    plexity. An LCD requires an ac electricfield to excite the organic compound inthe display. Applying a dc voltage couldpermanently damage the LCD. The cir-cuit in Figure 1 is a simple configurationto test the performance of an LCD. Thecircuit produces biphase square waveswith negligible dc content. The circuit isbased on a CD40106 hex Schmitt-triggerinverter. The circuit comprises an oscil-lator, IC

    1A; a phase splitter, IC

    1B; and a pair

    of buffer/drivers comprising IC1C

    /IC1D

    and IC1E

    /IC1F

    .

    The buffers and drivers connect to test

    probes through 47-k series resistors,which protect the IC in the event of shortcircuits. With the component valuesshown in Figure 1, oscillator IC

    1Apro-

    vides a square-wave frequency of ap-proximately 45 Hz. The circuit can oper-ate from 3 to 5V. To test any segment ofan LCD, touch the backplane using eitherof the two test probes while touching the

    segment with the other probe. If the seg-

    ment under test is operational, it willlight up. If the LCD under test is a mul-tiplexed type, then all segments, whichare connected, will glow if they are oper-ational.Usually, the rightmost or leftmostconnection is the backplane of the LCD.If it is not, you have to find it by trial anderror.

    IC1B

    IC1F

    R1

    IC1E

    IC1D

    IC1C

    IC1A

    13

    11

    12

    10

    9

    5

    8

    6

    220k

    C110 F

    25V

    C20.47 F

    R247k

    R347k

    PROBE 1

    PROBE 2

    314

    4

    1

    7

    2

    4.5V

    F igure 1

    Simple tester checks LCDsD Prabakaran, NL Polytechnic College, Tamil Nadu, India

    This simple circuit tests LCDs by producing a biphase square wave with no dc component.

    Is this the best Design Idea in this

    issue? Select at www.edn.com.

    ent tuning elements. You can use the cir-cuit as a crystal oscillator by changing theinductor to a crystal and changing the

    resonator capacitors to an appropriate

    value, such as 268 pF. You need a high-value, such as 10-k, bias resistor fromthe noninverting input to ground to pro-

    vide bias current to this input.

    This clean sinusoid is the signal at the top of the res-

    onator, VRES

    , in Figure 1.

    The Colpitts oscillator in Figure 2 produces a pure

    sinusoidal output.F igure 4F igure 3

    Is this the best Design Idea in thisissue? Select at www.edn.com.

  • 8/4/2019 17987-100302di

    3/686 edn | O ctob er 3 , 2002 www.edn.com

    ideasdesign

    Portable systems often useLEDs of different colors andin varying quantities of each

    color. Some examples are white forthe display backlight, green forkeypad illumination, and red forpower. Typically, the LEDs derivepower from at least two powersupplies: one for standard LEDs(red and green) and one for whiteLEDs. (White LEDs exhibit

    a higher forward voltage.)The keypad and other indicatorLEDs have current-limiting resis-tors associated with them. Toeliminate these resistors and drivegroups of dissimilar LEDs fromthe same source,you can regulate the cur-rent through multiple strings. Fourstrings of varying LED types derive pow-er from a single power source (Figure 1).The circuit mixes LEDs of different for-ward-bias requirements, yet keeps theloads reasonably well-balanced through

    use of a current mirror comprising tran-sistors Q

    1through Q

    4. It also eliminates

    the need for a separate current-limitingballast resistor on each LED orstring of LEDs and provides acommon control point (IC

    1s ADJ pin)

    for adjusting the LED intensities.Transistors Q

    2through Q

    4mirror the

    current in the diode-connected transis-tor, Q

    1. Note that the Q

    1current-set string

    (LEDs D3

    through D5) should have an

    equal or larger voltage than that of sub-sequent LED strings. (If it doesnt, the

    current-mirrored strings may have toolittle voltage overhead to function prop-erly.) You can easily meet that require-ment in the first string by placing eitherLEDs with larger forward voltage drops,such as the approximate 2.8 to 3.7V rangeof white LEDs, or more similar LEDs.Then,the circuit can easily accommodatethe subsequent strings with lower voltageburdens.The matched-transistor currentmirrors maintain a constant and equalcurrent in all LEDs, regardless of quan-tity and type. That configuration allows

    the use of a single power supply and a sin-gle point for adjusting LED brightness.

    Any power difference between the refer-ence string and a mirrored string dissi-

    pates in the current-mirror transistor forthat string: P

    MAX(transistor)

    (VOUT

    300mVVLEDs

    )ILEDMAX

    . Thecurrent-sense resistor value is R

    2300

    mV/ILEDMAX

    , where ILEDMAX

    is the sum ofcurrents from all the strings. (For a com-prehensive circuit and parts list, refer toMaxims MAX1698 (www.maxim-ic.com) EvKit data sheet.)

    When driving the same LEDs withoutthe current mirror, you can reduce pow-er dissipation in the sense resistor andballast resistors by substituting a mi-

    cropower op amp across the current-sense resistor (Figure 2). That circuit im-

    proves efficiency by reducing the resistorvalues and their associated loss. Increas-

    ing the gain of the current-sense signal byapproximately 16 allows an equivalent re-duction in the value of R

    2and the ballast

    resistors. A typical value for R2

    is 15,which represents a loss of 18 mW: (20mA)215 for each of three resistors. IfR

    2R

    5R

    60.931, then the resistor

    power loss drops to 1.12 mW. The opamp draws only 20 A maximum,whichrepresents a dissipation of 100 W.

    VCC

    REF

    ADJ

    GND

    SHDN

    EXT

    CS

    PGND

    AGND

    FB

    MAX1698IC1

    C110 F

    R1500k

    MPQ3904Q3

    MPQ3904Q4

    MPQ3904Q2

    MPQ3904Q1

    R2

    L110 H

    D1MBR0540T1

    D3 TO D5NSPW500BS

    D6 TO D8NSPW500BS

    Q5FDN337N

    LEDSTRING

    1

    LEDSTRING

    2

    LEDSTRING

    3

    LEDSTRING

    4

    C21 F

    D224V ZENERCMPZ5253B

    STANDARD-COLORLEDs

    STANDARD-COLORLEDs

    VCC2.7 TO 5.5V

    F igure 1

    Circuit drives mixed types and quantities of LEDsMark Pearson, Maxim Integrated Products, Sunnyvale, CA

    In this LED-driver circuit, a switching converter, IC1, and associated components lets you mix LED quantities

    and types.

    MAX4040

    VCC

    REF

    ADJ

    GND

    SHDN

    EXT

    CS

    PGND

    AGND

    FB

    MAX1698IC1

    C110 F

    C41 nF

    R31M

    R465.8k

    R2

    R5R6

    VCC

    R1500k

    L110 H

    D1MBR0540T1

    D3 TO D5NSPW500BS

    D6 TO D8NSPW500BS

    D9 TO D11NSPW500BS

    Q5FDN337N

    LEDSTRING

    1

    LEDSTRING

    2

    LEDSTRING

    3

    C21 F

    C31 nF

    D224V ZENERCMPZ5253B

    VCC2.7 TO 5.5V

    CURRENT-SENSERESISTOR

    BALLAST

    RESISTORS

    Figure 2

    Modifying Figure 1 as shown reduces the overall power dissipation in a standard application.

    Is this the best Design Idea in thisissue? Select at www.edn.com.

  • 8/4/2019 17987-100302di

    4/688 edn | O ctob er 3 , 2002 www.edn.com

    ideasdesign

    The circuit in Figure 1 provides a20-MHz square wave across a set ofhighly capacitive ion-deflection

    plates in an experimental instrument. Toget the required deflection, the plate volt-age must be 20 to 30V, much higher volt-age than conventional logic or driverfamilies can provide. To minimize arti-facts, the rise and fall times must be very

    fast, with a minimum of overshoot andringing. Identical circuits, phased 180apart, drive the plates. The driver uses aDirected Energy (www.directedenergy.com) DEIC420 high-speed MOSFETgate driver to drive a 1000-pF capacitiveload from 0 to 25V in less than 5 nsec.With smaller loads of a few hundred pi-cofarads, the rise time decreases to ap-proximately 3 nsec. Series resistors R

    1and

    R2

    control the output rise and fall times,allowing you to trade off the rise and falltimes against overshoot and ringing.

    A high-speed Analog Devices (www.

    analog.com) ADUM1100BR ferromag-netic signal isolator prevents systemground loops by providing dielectric iso-lation for the input signal; you could alsouse high-speed optocouplers.A low-pow-er MC78L05CD regulator provides pow-er for the signal-isolator output stage.

    A snubber network, composed of athin-film, high-power resistor R

    3in a TO-

    220 package and high-quality NP0 ca-pacitors C1

    and C2, terminates the load

    at the plates. You empirically determinesnubber values by observing the radiat-ed field on an RF spectrum analyzer us-ing a passive RF probe. You tune thesnubber network to reduce higher ordersignal harmonics. Note that placing anoscilloscope probe on the outputs signif-icantly increases the observed higher or-der harmonics, indicating that adding theprobe to the circuit increases ringing andovershoot.The DEIC420 is mounted in a

    high-speed, high-power package that

    minimizes lead inductance. The part re-quires multiple bypass capacitors at eachof its power pins. You should choose thecapacitors so that their self-resonant fre-quencies do not significantly overlap.Having a full ground plane and usinghigh-speed and RF-signal-layout tech-niques are critical to the proper operationof this circuit. The input must be well-

    isolated from the output. Double-puls-ing, ringing, and even oscillation may oc-cur if you dont strictly follow thesepractices. The tracks or cabling betweenthe driver and the load should be im-pedance-controlled and should be asshort as possible. The DEIC420 requiresgood heat-sinking when you operate it athigh speeds and high voltages. When op-erating at 20 MHz from a 25V supply, thetwo drivers and snubber together dissi-pate 130W.

    VI V0

    GND GND GND GND

    2 3 6 7

    IC2MC78LO5ACD

    IC3DEIC420

    R382

    35W

    VDD1 VDD2

    OUT

    GND2

    GND2

    VDD1

    IN

    GND1

    IC1ADUM1100BR

    0.1 F

    0.1 F

    C222 pF

    C122 pF

    0.22 F

    50V

    8 1

    G1

    0.22 F

    50V

    G1

    0.1 FG1

    G1

    G1

    G1G1

    8

    6

    7

    5

    13

    2

    4

    0.1 F

    50V

    G1

    G1

    1

    2

    1

    2

    0.22 F

    50V

    G1

    1

    2

    10 F

    35V

    G1

    1

    20.01 F

    50V

    G1

    1

    2

    1

    2

    AXIAL WIDEBAND CHOKE, 2.5 TURNS

    WB2-2.5T

    2

    2

    2

    1

    1

    1

    2

    1

    3

    64

    5 1 2

    CONTROL RISE TIME

    1, 1W, 5%R1

    1 2

    1, 1W, 5%R2

    2

    1DEFLECTION

    PLATES

    12 TO 25V POWER SUPPLY

    1

    2

    5V ISOLATED SUPPLY

    DRIVE SIGNAL

    ISOLATION BOUNDARY

    KEEP TRACK LENGTHS DOWN

    KEEP TRACK LENGTHS DOWN

    IDENTICAL CIRCUIT FOR SECOND PLATE

    DRIVEN 180 OUT OF PHASE

    INVERTED DRIVE SIGNAL

    SNUBBER

    NETWORK

    F igure 1

    MOSFET serves as ultrafast plate driverClive Bolton, Bolton Engineering Inc, Melrose, MA

    You can use a high-speed, MOSFET-driver IC to drive ion-deflection plates.

    Is this the best Design Idea in this

    issue? Select at www.edn.com.

  • 8/4/2019 17987-100302di

    5/690 edn | O ctob er 3 , 2002 www.edn.com

    ideasdesign

    High-resolution temperature sens-ing at low cost is possible using onlyone chip attached to the PCs

    parallel port (Figure 1). The Dal-las Semiconductor (www.dalsemi.com)DS1722 digital thermometer allowsmeasurement resolution as fine as0.0625C in digital form and with linearresponse. The accuracy specification isonly 2C, but you can improve this figure

    by careful calibration. Moreover, the ac-curacy spec is unimportant in applica-tions in which you measure only changesin temperature or in which you mustclosely maintain a noncritical tempera-ture. The measurement range is 55 to120C, the part can use either three-

    wire or SPI interface, and the cost is ap-proximately $1. The eight-pin part isavailable in SO or SOP packaging andin large quantities as a flip-chip measur-ing only about 1 mm sq.

    In this application, the chip attaches

    directly to the PCs parallel port througha male DB-25 connector. Because the de-vice draws a maximum of 0.5 mA, theport can supply the power, and its sup-ply range tolerates variations in voltagelevels that may exist on varying ports.The chip is in SPI mode with the SCK

    1 VDDD

    3 SCK

    4 GND

    1 VDDA 8

    2 CE SERMODE 7

    SDI 6

    SDO 5

    DS1722 1

    D0

    D7

    NACK

    PAPER

    25

    5V

    F igure 1

    Parallel port provideshigh-resolution temperature sensing

    Martin Connors and Mike Foote, Athabasca University, AB, Canada

    The DS1722 connects to a PCs parallel port

    through a male DB-25 connector, seen from the

    pin side.

    LISTING 1TURBO C FOR DATA-TRANSFER CYCLE

  • 8/4/2019 17987-100302di

    6/692 edn | O ctob er 3 , 2002 www.edn.com

    ideasdesign

    clock signal supplied by the PC; in thisway, data-transfer timing is noncritical.

    A simple Turbo C program (Listing 1)running in DOS mode effects the data-transfer cycle in the PC, whereas thetransfer is automatic in the chip upon re-ception of SCK. The routine reads a lowbyte and a signed high byte and creates afloating-point value by simply adding thelow byte,divided by 256, to the high byte.In the highest resolution mode, whichthis design uses, a data read can occuronly every 1.2 sec,and you should adjustthe timing loops accordingly. You mayalso need to adjust the settling time,

    DELTIME, depending on the speed ofthe PC you use. The sample programprints the bytes transferred as well as thetemperature, and you can easily modifyit. The data sheet explains the use of theconfiguration register and changes tomake if you need a higher data rate with

    lower resolution.You can download List-ing 1 from the Web version of this Design

    Idea at www.ednmag.com.The data transfer takes place beginningwith the write of an address byte to thechips SDI in the order A7 to A0 (high bitto low bit). If A7 is high, a write takesplace; otherwise, a read occurs. For awrite, D7 to D0 route to the chips SDI.For a read, D7 to D0 are available on thechips SDO. The program always usesboth SDI and SDO and ignores whichev-er it doesnt need.For example, data goesto the chips SDI even during a read, butthe chip ignores this data. Each byte

    transfers as 8 bits, and each transfer in-volves the following steps:1. The PC raises D1/SCK and places 0

    or 1 on D2 for the chips SDI.2. The PC then reads PAPER.3. Finally, the PC drops D1/SCK.This action repeats for each bit of the

    pair of bytes being transferred (one in,one out). By using the other parallel

    ports output pins as chip selects, youcould string together several devices. Youcan also use these pins to control a heaterby use of a switching transistor or anSCR. With this scheme, you can achievehigh-resolution temperature controlwith minimal parts and a simple pro-gram.Alternatively, if you need only lowaccuracy, you can implement a very-low-cost thermostat with this part.

    Is this the best Design Idea in thisissue? Select at www.edn.com.