2-29

A Low-Cost Surface MountPIN Diode π Attenuator

Application Note 1048

IntroductionAnalog attenuators find wide ap-plication in RF and microwavenetworks. Realized as either GaAsMMICs or PIN diode networks,these circuits are used to set thepower level of an RF signal from avoltage control. In commercialapplications, such as cellular tele-phone, PCN (Personal Communi-cation Networks), wireless LANs(Local Area Networks) and porta-ble radios, cost is a significantconsideration in the design of suchattenuators. This paper describesa low cost wideband PIN diode π

(Pi) attenuator which utilizes plasticpackaged surface mounted devices.

BackgroundThe basic π fixed attenuator isshown, along with its design equa-tions, in Figure 1. Shunt resistorsR1 and the series resistor R3 areset to achieve some desired valueof attenuation A = 20 log(K) whilesimultaneously providing an inputand output impedance whichmatches the characteristic imped-ance of the system.

When operated at frequencies wellabove its cutoff frequency fc (see

Appendix A), the PIN diode can beused as a current controlled vari-able resistor. Three diodes can beused to replace the fixed resistorsof the π circuit to create a variableattenuator, and such circuits havebeen described in the literature.For example, a three diode π atten-uator1 is shown in Figure 2 whichprovides good performance overthe frequency range of 10 MHz toover 500 MHz. However, the use ofthree diodes as the three variableresistors in a π attenuator leads toasymmetry in the network, whichresults in a rather complicatedbias network.

1 “The PIN Diode,” from the Hewlett-Packard RF and Microwave ApplicationsSeminar, 1973. Figure 2. 3 Diode Attenuator.

Figure 1. Basic π Attenuator.

Zo

R1 R1

R3

Zo

R1 = ZoK + 1K - 1

R3 = K

K -2

1

WHERE K IS THE INPUT TO OUTPUT VOLTAGE RATIO AND Zo IS THE IMPEDANCE OF THE SOURCE AND LOAD

Zo

6.81 kΩ 4.7 µH 10 µH 4.7 µH 21.5 Ω

.14700 pF4700 pF.1

V+ Vc

.02

IN/OUT

.02.02 D2

D3D1

450 Ω

.02 .02

1.47 kΩ

IN/OUT

D1, D2, D3 = HP 5082-3081

5966-0449E

2-30

COMPONENT VALUE MFG./PART NUMBER

R1,R2 560 Ω KYOCERA CR21-561JB1

R3 330 Ω KYOCERA CR21-331JB1

R4 1640 Ω KYOCERA CR21-162JB1

R5 680 Ω KYOCERA CR21-6B1JB1

C1-C5 47000 pF KYOCERA 0805Z473M2P03

D1-D4 - HEWLETT-PACKARD HSMP-3814

Four Diode πAttenuatorIf resistor R3 is replaced by twodiodes, as shown in Figure 3, sev-eral benefits result. First, since themaximum isolation of the networkis set by the capacitive reactanceof the series diode(s), the use oftwo diodes in place of one willincrease the maximum attenuationor double the upper frequency lim-it for a given value of attenuation.Second, the twin diodes whichoccupy the position of the seriesresistor are physically set up 180°out of phase, resulting in the can-cellation of even order distortionproducts2. Third, the resultingattenuator network is symmetricaland the bias network is substantial-ly simplified. V+ is a fixed voltage,and Vc is the variable voltage

which controls the attenuation ofthe network. The only drawback tousing two series diodes in place ofone is the slight increase in inser-tion loss, amounting to less than0.5 dB additional loss. R1 and R2serve as bias returns for seriesdiodes D2 and D3; they must be sethigh enough to minimize insertionloss; however, if they are set toohigh, an excessively large controlvoltage Vc will result. If the design-er does not require very largebandwidth, some savings in inser-tion loss can be achieved by add-ing chokes between R1 and R2 andthe RF line, using these inductorsto decouple the resistors from theRF portion of the network. R3 andR4 are chosen to match the charac-teristics of the specific PIN diodesused; properly selected, they will

Figure 3. Wideband 4 Diode π Attenuator.

provide for the correct split of biascurrent between series and shuntdiodes required to maintain goodimpedance match over the entiredynamic range of attenuation.While analysis can be used to de-termine the values of R1 through R4,it is much quicker and easier to selectthem empirically.

The HP HSMP-3810 series of sur-face mount PIN diodes featuresgood distortion performance, lowcutoff frequency and low price. Tosave cost and space on the board,two HSMP-3814 common-cathodepairs were chosen over four indi-vidual HSMP-3810 diodes. Havingchosen these diodes, and selectingV+ = 5V and 0 ≤ Vc ≤ 15V, the val-ues of R1 through R4 were empiri-cally determined. Values for all com-ponents used in the tested circuit areshown in Figure 3.

The attenuator was laid out on a2 inch square of 0.032" thick HT-2PC board, as shown in Figure 5.This material, a high performancealternative to conventional FR4, isdescribed in detail in Appendix B.Using chip resistors and capacitors,the entire attenuator occupies a0.5 in 2 space as shown in Figure 5.

Test ResultsIn Figure 6, the measured attenua-tion vs. frequency is given for sev-eral values of control voltage.Good performance is obtained overthe frequency range of 300 KHz to3 GHz. Figure 7 contains the plot ofreturn loss vs. frequency at themaximum and minimum values ofVc. For all other values, the return

2Raymond Waugh, “A Low Distortion PINDiode Switch Using Surface MountDevices,” Proceedings of RF EXPO WEST,pp 455 - 461, Feb. 5 - 7, 1991.

D1

IN/OUT

C1

R1

C4

R4 R5R4

C5

R2

D2 D3

D4

C2

Vc

R3

C3

IN/OUT

V+= 5.0 V

HSMP-3814 HSMP-3814

2-31

Figure 4. Circuit Board Layout.

loss was higher; the data for Vc = 0represents the worst case. In Fig-ure 8, a plot is given for attenua-tion vs. control voltage at a numberof frequencies. Finally, the inter-modulation distortion performanceof the attenuator is plotted inFigure 9. The data are given as

intercept points; for a detailedexplanation of intercept points,see Appendix C.

ConclusionAs can be seen from these data,the four diode π attenuator pro-vides very good match and very

flat attenuation over an extremelywide band. Using surface mountdevices, it has the additional bene-fit of being low cost. Realized as athin-film or thick-film hybrid cir-cuit with chip PIN diodes, it wouldfit within a TO-8 can.

1 2

Vc

V+

2-32

Figure 8. Attenuation vs. ControlVoltage.

Figure 7. Return Loss vs. Frequency.Figure 6. Attenuation vs. Frequency.

47,000 pFKYOCERA

CHIPRESISTOR

CHIPRESISTOR

47,000 pFKYOCERA

47,000 pFKYOCERA

CHIPRESISTOR

47,000 pFKYOCERA

47,000 pFKYOCERA

CHIPRESISTOR

VIA HOLE TO GROUND

Figure 5. Detail of Circuit Layout.

100

10

2

AT

TE

NU

AT

ION

, dB

0.3 1 10 3015CONTROL VOLTAGE Vc, VOLTS

10 MHz

100 MHz

1 GHz

3 GHz

0

10

20

30

40

50

RE

TU

RN

LO

SS

, dB

10 100 1000FREQUENCY, MHz

Vc = 0 V

Vc = 15 V

0

10

20

30

40

50

60

70

80

90

100

INS

ER

TIO

N L

OS

S, d

B

0.3

6 V

1.7 V

2.4 V

1.3 V

1.17 V

1.1 V

1 10 100 1000FREQUENCY, MHz

Vc = 0 V

Vc = 15 V

2-33

Appendix A - PIN DiodeCutoff FrequencyThe PIN diode is generally consid-ered to be a current controlled RFresistor. However, this model isaccurate only at frequencies wellabove the diode’s cutoff frequency,fc = 1 / 2πτ, where τ is the minoritycarrier lifetime of the device. Atfrequencies 10 times fc, a PIN di-ode can accurately be modelled asa current controlled resistance inparallel with a small (and constant)junction capacitance (neglectingpackage parasitics). At frequenciesunder 0.1 fc, the PIN diode behavesas an ordinary PN junction diode.For 0.1 fc ≤ frequencies ≤ 10 fc, thecharacteristics of the PIN diodebecome very complex; it willgenerally behave as afrequency-dependent resistanceshunted by a very large frequencyand current dependent inductanceor capacitance. Additionally, dis-tortion performance will usually bevery poor when operating in thisfrequency range. For theHSMP-3810 series of diodes,τ ≅ 1500 nsec, resulting in a cutofffrequency of 100 kHz. This diodeshould therefore providefrequency-independent values ofpure resistance at frequenciesabove 1 MHz. However, becausethis diode has been optimized forwideband attenuator applications,its characteristics remain generallywell behaved down to frequenciesbelow fc, as can be seen from the300 k Hz measured data shown inFigure 6.

Appendix B - BoardMaterialSeveral printed circuit board mate-rials are in common use for RFcircuits such as this one. Two ofthe most popular are FR4 andfiberglass reinforced PTFE(Teflon®). The former provides

good mechanical stability anddurability at low cost. However, itsuffers from high losses and adielectric constant which is poorlycontrolled and stronglyfrequency-dependent. The latterexhibits very good RF properties,but is expensive, suffers from poormechanical stability, and cannotsurvive certain SMT (SurfaceMount Technology) processingsteps. Hewlett-Packard’s newHT-2 board material provides dura-bility and high temperature perfor-mance which are actually superiorto FR4 with a controlled dielectricconstant (εr ≅ 4.3) and a loss tan-gent which is half that of FR4.These properties make it ideal formicrostrip circuits operatingbeyond 6 GHz.

At the time of this printing, HT-2 isavailable through Dan Schutte ofInternational Circuits, 1319 SouthArkle Street, Visalia, CA.

Appendix C - TheIntercept PointOf the several types of distortionproducts, one of the most trouble-some is intermodulation distortion.Unlike harmonic distortion, this isa multi-tone product resultingwhen two or more signals of equal(or unequal) amplitude mix in anon-linear device such as a PINdiode. The frequency of the result-ing unwanted signal is related tothose of the original input voltages.In certain industries, the numberof input signals may exceed 10,and both test and analysis becomevery complex. To keep matters assimple as possible, many semicon-ductor manufacturers maketwo-tone measurements using twovoltages which are equal in ampli-tude and closely spaced in frequen-cy. Given two such input signals atfrequencies f1 and f2, one can com-

pute several significant intermodu-lation distortion products from theequation

Kf1 ± Mf2

where K, M = 1, 2, 3, ....

The order of the distortion productis given by the sumN = K + M.

Of the infinite number of distortionproducts described by this equa-tion, one is of special significance.The third order products given inFigure 1 are important becausethey exist on either side of the orig-inal signals f1 and f2 and cannot beremoved by filtering.

The behavior of all types of distor-tion products is shown on Figure10. As can be seen, an increase of 1dBm in the applied signal’s powerwill result in a 2 dBm increase inthe second order products and a 3dBm increase in the third orderproducts. Since the level of mea-sured distortion is dependent uponthe level of the input signal, it isconvenient to specify distortion in

Figure 9. Measured DistortionPerformance.

50

45

40

35

30

25

INP

UT

INT

ER

CE

PT

PO

INT

, dB

m

0 10 20 30 40 50ATTENUATION SETTING, dB

300 MHz

100 MHz

30 MHz

10 MHz

TWO SIGNALS, SEPARATED BY

1 MHz, CENTERED AT . .

TWO-TONE, THIRD ORDER INTERMODULATION

DISTORTION INPUT INTERCEPT POINT

vs. ATTENUATION SETTING

2-34

terms of a fictitious constant, theintercept point. This is the point atwhich the extrapolated fundamen-tal signal and the extrapolated dis-tortion product meet. In makingdistortion measurements, it is mostconvenient to measure the inputpower of the signal(s) applied tothe DUT (Device Under Test) and

the output power of the distortionproducts. For this reason, and be-cause the input intercept pointvaries less with attenuation, theinput intercept point is the onemost often calculated and speci-fied. Using it, we can neatly specifythe performance for a given type ofdistortion using a single number.

Figure 10. Behavior of Distortion Products.

OU

TP

UT

PO

WE

R, d

Bm

Pout = Pin - α

INPUT POWER, dBm

INPUTINTERCEPTPOINT

OUTPUT INTERCEPT POINT

FUNDAMENTALSLOPE = 1

DISTORTIONPRODUCTSLOPE = N

Pin

PDIST

The equation for input interceptpoint is

N(Pin - α) - PdistIPin = ______________ + α , in dBm

N - 1where N = order of distortion prod-uct, and all power levels are speci-fied in dBm.