'$ŽW 4 '

4,r

-A"-

September979

AD AO7SS~"mt 4

"'V A 7$4{'~A ~tt4 n

-do'~ Z2'Nl L4"f2y 5~ '4r4 •45f~j 4 ~St%4

Al~b.'rPtrdy Damplrlg LoCt 44'444 4

"Ade'hl, MO 'f%-07$3

BestAvai~lable

Copy

-Vi "5mw,;.m- * ~.whn Iw w~w~~s-~~~ t 5*~-'

Ž#~~~~ * ~ j;d~j,-x 44&'t 1.~ tt'½-ri n- -Žr~_ t~h-74 ;j

-~~ ?tft~~Ak t.~'~ ~MR;

V *~-~A-'7tt

Thefi-to be, s'V co- ½'t4, - w

- i - " '-- g - -V , -I'll At A '9- - -)2

00 A' dolifi-

-'5-'. ~~~~ UM:,''¾ ' " '

Sitsg&1 lkS""- "-'-5 A '

U NC LASS M-SECURITY CLASSIFICATION OF THIS PAGE (D1ten Dat. Kintered)

Viscoelastic Polymer for Printed-Circuit- echnicalke/

( ~ Mark R. ~Pr0bsL

Harr Di7n AREA 6 WORK UNIT NUMBERS

2800 Powder Mill Road Program Elem: 6.33.17.AAdeiphi, MD 20783

11. CONTROLLING OFFICE NAME AND ADDRESS(Ii.1ma

U.S. Army Missile Research and 111Sep~W9Development. Command Ill NUBE

Huntsville, AL 35809 13MONIITORING AGENCY NAME & ADDRESSIt dill@ Wruq 3LUA aL4I. SECURITY CLASS, (of this report)

I11a. DEC A5SI PIC ATI ON/ DOWN GRADING

IS. DISTRIBUTION STATEMENT (at this.Report)

Approved for public release; distribution unlimited.

17. O13TR19UTION STATE[MENT (of the abetrect entered In Black 20, It different from, Report)

IS. SUPPLEMENT4RY NOTES

HDL Project., 692964DRCMS Code: 6543317.1120012

It. KEY WORDS (Continue on toyeree side It nlecessary and identtify by block num~ber)

Viscoelastic ResonancePolymer DampingMissile AttenuationPrinted circuit board viscous

20, AUSTRACr (Caath~i. am ,oerwr 04b M new..y sad tdnitfr by block nuiuber)

--;A viscoelastic polymer was selected as a possible means ofvration damping for electronic printed-circuit-board (PCB)

assemblies in a surface-to-air guided-missile application. Thinlayers of the self-adhesive polymer were bonded to dummy and realPCB assemblies, and accelerations were recorded at various PCBlocations during sinusoidal vibration tests. For input levels of1 and 2 g and logarithmic frequency sweeps from 10 to 2500 Hz,

DID I JA* 7 1473 C''owo' OF I M "OV 1Si NL UNCLASSIFIEDSI CIJNTY CLAWSIFICATION OF TNIS PACE (win, Date Ufa"ed

1.6 3 OMD

-- - --- - - - - - ---- -- -- - , - - - - - - - -

UNCLASSIFIEDSECURITY CLASSIFICATION OF THIS PAGE.lohwi Dole Mnteod) - .

•- peak resonant acceleration leve)ý:, w•re reduced to 15 percent ofthe peak levels recorded under ,.he s&mie conditions with no dampinglayers added. C,

I

(

I

-i

I "I

Al

UNCLASSIFIED2 SECURITY CLASSIFICATION OF THIS PAGE1"f~en Dlt. &nt.,.d)

•t~i 4Ii

CONTENTS

19 INTRODUCTION.5..,...

2. VISCOELASTIc VIBRATION DAMPING ............. ............... ..... 5

13. PCB APPLICATION OF VISCOELASTIC DAMPING. ......................... 6

3.1 Stage One--Evaluation of Damping Capability ................. 73.2 Stage Two--Damping Layer Attachment ........................ 8

4. CONCLUSIONS AND RECOMMENDATIONS ................................. 12

DISTRIBUTION 9*..... o...................................... ........ 13

FIGURES

1 Viscoelastic damping system. ............... .................... 6• I ~ 2 PCB model ................................ 7

3 Function.al PCB assembly.............. ................ 11

!II TABLE

I'I 1 ~~1 Sinusoidal Vibration Test Data...................... 9

IDl>',J AB

JL-v._ If i uati~on

Dict •ibutio/

!ýWi I• bi

3 Aval an &d/o.L.

3l

•2~

1INTRODUCTION

Printed-circuit-board (PCB) encapsulation (potting) is an effectivemeans of protection from shock and vibrational environments, but thefollowing problems are commonly associated with this technique.

(a) High-temperature potting cure can damage components.

(b) Internal stresses during and after curing can damagecomponents.

(c) Potting materials tend to be good th•cmal insulators andcan prevent adequate component cooling.

l'•o (d) PCB repair can be very difficult or impossible.

An effective alternative to potting is to pack small silicone rubberparticles into a closed assembly containing those PCB assemblies forwhich protection is required. 1

Because of the problems associated with potting and a predeterminedpackaging concept which precluded the silicone rubber technique, analternative method %ws required for PCB shock and vibration protectionin a surface-to-air guided-missile environment.

2. VISG3ELASTIC VIBRATICN DAM4PING

"Viscoelastic damping materials are helping to solve a number ofspecial noise and vibration problems. Materials of this sort are beingused on -.,Arcraft, saw blades, skis, and even skyscrapers." 2

The particular viscoelastic polymer evaluated herein is made by theIndustrial Specialties Division, (ISD) of the 3M company. The dampingsystem consists oZ a self-adhesive layer of polymer bonded to a,reinforced-plastic constraining layer (fig. 1), and the polymer side isbonded directly to the structure requiring vibrational damping. Thepolymer is made up of randomly tangled chainlike molecules, and theintermolecular viscous friction during structural vibration dissipatessome of the structure's vibrational energy. 2

IR. Rich,4rd Palmisanc and Darrell W. Neily, Particulate SiliconRubber: An Effective, Removable Encapsulant for Electronic Packaging,Harry Diamond Laboratories, HDL-TR-1762 (August 1976).

2polymer Damps Vibration and Noise by Converting Dynamic Energy toHeat, Product Engineering (December 1977), pp 27-28.

i, t >

777

SELF-ADHESIVEVISCOELASTIC POLYMER

(C(1.524 mm

REINFORCED PLASTICCONSTRAINING LAYER-

lrigu~re W~5saoeladtiC damping system.f

-3. PCI3 APPLICAT~ION OF VISCOELASTIC DMP6INGV

Since the solaer-tip ride_ of a high~-component-density PCB3 assemblypresents a- nonuifif6rin sufa,ýe, unlike- the application examples describedin the Product Brigineeririj &-ticvle,z it was assumed that attaching thedamping mate-rial to t 'he- VCB assembly would be difficult. A goodmechanical hond Is regniured to in~sure-energy transfer from the vibratingPC3 to, the fam~ping material, yet PCB3 repArability is a highly desirablefeature which could Ibe jeopardii-zed by a bonding technique which would be

The above factovs f cc~ibihed with an absence of experience with theviscoel~stic materWa, led to the concalusion that a two-stage productevaluation would- be the most econdmioal appcoach. Stage one wouldconsist of simple,.- eaniiy r~apeatable quantitative -measurements of thematrlalsb dampifig capability, if -the damping- capability provedadequate, stagse -ti~ vo~ld conslst of the design and evaluation of anacceptable method for- attaching theý damping systdm to a PCB assembly.An acce6ptable nothad would preferably possess al1 the followingfeatures:

(a) efficient energy tranzmission from PCB to dampinginaterial,

=@oymr Damips Vibration and Noise by Converting Dynamic Energy toHleat., Produat Enginesr~lym (Dacsmber- 1977)e pp 27-28-.

(b) no introduction of' circuit performance degradation,

(c) elimination or minimization of the possibility of PCBdamage during damping system attachment,

'(d) possibility of component replacement, and

(e) small cost with respect to PCB.

'>Features (a) through (c) are absolute requirements, while -failure topossess (d) and/or (e) would have to be considered in light of theoverall system requirements. If, for example, a method possessed allfeatures except (d), it might be possible to add this feature byincreasing the cost. Such a trade-off could be evaluated by comparingthe increased first-time assembly cost with the potential savingrealized when a "throw-away" assembly is converted to a reparable item.

3.1 Stage One-Z'valuation of Damping Capability

A simple model :fig. 2) consisting of 0.065-in.-;thick(1.651-mm) epoxy-fibergl ass laminate (NEMA Grade G-10) and uniformlydistributed steel weights was used for initial evaluation. Board si~ze

and mounting configuration were based on the intended application, butno attempt was made to faithfully model any specific components orlayout.

,I -! IiI**,o • *

fOC

11 -VIA-I I

M MC14 k •-4 V.•to

01OAA0 I W1I$ lIS OAWMIO IIAIIRAL.L S~ll 200 V..0 ,IIklS.0 D.0) 1 0.(103 *.CI$

D 4•A.0Y LOA A}&CIIAOl A -' l I

II I0.16 -AC" L I-V - I

Figure 2. Pca model.

7Z

3M Company's ISD No. 112 is a self-adhesive viscoelasticpolymer which peaks in damping capability in the temperature range of 60to 100 F (15 to 38 C). It was supplied by 3M in a 0.060-in. (1.524-mm)layer bonded to a 0.019-in. (0.483-mm) reinforced plastic constraininglayer. Attachment to the model PCB was accomplished by merely peelingoff the protective paper cover and pressing the polymer layer againstthe smooth siei of the board. Accelerometers were positioned as shownin fIgure 2.

Sinusoidal vibration tests were run using a 7.5-min logarithmicsweep from 10 to 2500 Hzz Four runz ware made in which the var.iableswere input excitation level (I or 2 g) and vibration attenuation (dampedor undamped). Table I shows the results of these runs, called tests 1through 4. Higher-order harmonics were encountered, but peak 0,acceleration levels were insignificant with respect to those encounteredat the model's primary natural frequency. For tests I and 2 (1-ginput), peak g values with damping averaged 13 percent of those withoutdamping. For tests 3 and 4 (2-g input), peak values with damping 3averaged 16 percent of those without lamping. Though the additionalwei;'ht of the damping system alone would lower the model's naturalfrequency, che increased stiffness due to the damping material plus theconstraining layer yielded oa net 25-Hz increase in model naturalfrequency.

I The above results clearly indicated that the viscoelasticdamping material could greatly reduce acceleration ampliZication atsystem resonances, and thus stage one of the product evaluation waspositively concluded.

3.2 Stage Two--Damping Layer Attachment

Although the 3M Company's application engineere had no direct1 t experience with PCB vibration damping via viscoelastic polymers, they

felt that the material was sufficiently compliant to allow thesolder-tip side of a PCB to be pressed against it hard enough to allowadequate contact between the self-adhesive damping material and the flatportion of the PCB. This may be true if component density is not toohigh, but for the intended application, component density on some boardswas very high (average of 28 leads/in. 2 ). The force required to press ,the solder tips (0.060-in. maximum length) into the damping materialcould lead to board and component damage unless the board could berigidly supported, but an economical support method for a PCB denselypopulated with discrete components could not be found.

"8

In 0 n 0, 0 0t InO 0O00 111 e0..O:D ý14 C1 NC 11 to O0 10 ?1 1000

r00 A 14 An 'AI 0011 I'UI'

v% 0CNN 10 0 7 0 4. N o '

)C- Ln %n oI o 0 0 Io 0 .1%0 0000 4 . 0 In mn 4 .4 0 P.14 CO.711

U)N '0 C4 1 N N r -% 11 %'n m

Ln -0 %D

0H

Ln 0 tA a 0 0 tn 0 00o tn ~jc7 0004 0 . 0 in In Ny tn inin 6 N vA M aO

(4 N - N N %D04. %D M M %D 0 ID O m

A A

r L a %n 0 0 a% Clt' cc. 0.cc a77

0 r.n I 0 0 0l 00 Ca 0 A0 C.C,' C47 4.N N 0.7

U x 0Ir 0 CO V n OD 0 tf O InC MAO

12 In %0 r- D

II4 04)-

M FO

!ihe next method considered for damping layer attachment was toprovide solder-tip clearance in the damping layer by drilling a 0.10-in.(2.54-nm) diameter hole at each solder-tip location. In a productionsituation where PCB's are drilled with numerically controlled equipment,the existing drill tape could be used to drill the clearance holes inthe damping material. Before this method was pursued further, a testwas run to determtine the effect cn damping capability which might resultwhen a significant amount of the energy-dissipating material wasremoved. Ilie same model used in tests 1 through 4 was used, but 25percent of the damping system was removed by rand-n drilling of 0.10-in.diameter holes. Table I shows test results for I- and 2-g inputs astests 5 and 6; the only significant difference from tests 2 and 4 is theincrease in natural frequency which is due to weight decrease withoutsignificant change in stiffness.

The final tests included in this report were conducted on afunctional PCB assembly (fig. 3). The PCB drill tape was used to drill0.10-in. diameter solder-tip clearance holes in the damping system, andthen the material was cut to match the PCB outline. CutoL-s wereincluded to allow rigid mounting against the four 0.25-in. (6.35-sm)diameter mounting standoffs. Polyurethane conformal coating had beenapplied to the PCB assembly, and the solder-tip side was cleaned withtrichlorethane before the damping system wae poressed in place. This PCBwas much less crowded (13 leads/in.2) than the remaining boards in thesystem, bit it was the only sample available for testing. Accelerometerlocations are shown in figure 3; table I shows the test results as tests7 throngh 10. For accelorometer location No. 12, peak accelerationswith damping were only 14 percent of peaks without damping for both the1- and 2-g inputs. Acceleration attenuation for other locations and forhigher-order resonances was not as good on a percentage basis, but forall these capc-, the peak levels with damping were no greater than 7.0 gand no greater than 3.5 times the input g level.

The machinability of the damping material was a mi.,or problemfor the above test--the material is extremely gummy and tended to buildup on the drill. This minor trouble was magnified when damping layerswere drilled for the more densely populated PCB assemblies, and it was 4necessary to use a backup material which would clean the drill aftereach hole. The resultant damping layers were unacceptable because smallparticles of the backup material were throw., outward around each holeand forced between the protective paper cover and the self-adhesivedamping material. Various efforts to remove the embedded particlesfailed. When a wooden backup was used, the sawdust particles led toinsufficient bonding between the PCB and damping system. When aluminumwas rmed as a drilling backup, the conductive bits were a potentialsource of electrical shorts.

10

I[

7 - 33/4

A I

TYPICAL TYPICALMOUNTING ACCELEROMETER4 HOLE

4 •5 1/4

DIMENSIONS IN INCHES(in,)25.4-(mm)

rPC ASSEMBLY + CONFORMAL COATING 0.064 Ib. 0,029 kg)

PCO AS'SEMBLY + CONFORMAL COATING + PCRFORATED DAMPING MATERIAL0.086 lb. (0.039 kg)

Figure 3. Functional PCB assembly (top view). I)i oltonwic asnt e be tie oudbet snwih.•sacrFor a production PCB application of the viscoelastic damping

system, an acceptable attachment method must be devised. A possiblesolution which has not yet been tried would be to sandwich a1 spacerbetween the PCB and the damping material. This spacer would be0.065-in. thick epoxy-fiberglass laminate with 0.10-in. diametersolder-tip clearance holes drilled via the PCB drill tape. A plain(undrilled) layer of damping material (constraining layer plusviscoelastic) could be attached to this board, and then this subassemblycould be bonded to the solder-tip side of the PCB using polyurethaneconformal coating. The self-adhesive damping system could be peeled off

J ito permit PCB repair and then replaced after the repair.

Though this technique appears to add 0.065 in. to the totalassembly thickness, it would be possible to pare down this increase oreven decrease the previous thickness of 0.144 in. (0.65-in. PCB +0.060-in. polymer + 0.019-in. plastic constraining layor). Since thePCB stiffness would be greatly increased by the 0.065-in. spacer, PCBthickness could be cut back to 0.035 or possibly 0.020 in. Since the0.060-in. polymer thickness was originally intended to permit solder-tipimbedding, this dimension could be reduced significantly (tests showedno significant decrease in damping when 25 percent of the polymer wasremoved by drilling) . The thickness of the constraining layer could begreatly reduced by substitution of metal for the reinforced plastic (the

-.1 nonmetallic constraining layer would no longer be required since itwould be insulated from the solder-tips by the undrilled dampingpolymer). In cases requiring electronic shielding between PCB

I 11

11

assembliles, the metal constraining layer would provide this function,and thus a further reduction in overall assembly thickness could berealized.

'{ ~4. CONCLUSIONS AND &RCOMMENDATIONS •Tests to date have shown that a properly designed and attached K

viscoelastic damping system can reduce acceleration amplification levelsin printed-circuit-board assemblies by as much as 87 percent. Althoughreductions of this maquitude were not attained under all conditions oftest, all tests showed significant ir'provement with the viscoelasticmaterial applied.

The testi',,j to date has only been performed on limited samples andunder ambier' temperature conditions. The test results are sufficientlypromising tý warrant further tests of the material. It is recommendedthat futuxe evaluation of the following areas be made.

(1) Damping characteristics should be ascertained for the-,!temperature range from -25 to +1.60 F (-31 to +71 C). (3M Company

engineers indicated that a lamination of three separate polymers wouldbe required to provide the most effective damping over this range, butsuggested that the ISD No. 112 polymer might prove adequate if somedegradation could be tolerated at the temperature extremes.)

(2) Chemical inertness should be determined to establish shelflife before and after application and to identify any tendency toSdecompose or initiate corrosion.

(3) The material's ability to resist moisture absorptionshould be determined.

(4) Electrical inertness regarding the dielectric strength,dielectric constant, resistivity, and dissipv'tion factor of the materialshould be ascertained to assure absence of edverse effects on circuitoperation.

12