TEMPERATURE CONSIDERATIONS FOR SCR CONTROLS

Transcription

1 AN Applicatin Nte PAYNE ENGINEERING TEMPERATURE CONSIDERATIONS FOR SCR CONTROLS q = (h c + h r ) A (T s - T amb )

2 TEMPERATURE CONSIDERATIONS FOR SCR CONTROLS Thyristr cntrls - mre cmmnly called SCR cntrl - are being widely used fr cntrl f electric heat and electric mtrs. Fr the mst part, they are replacing electrmechanical cntrls such as cntactrs, auttransfrmers, and starters because f their vastly imprved perfrmance and reliability. Hwever, there are basic rules which must nt be verlked in the applicatin f SCR cntrls: 1... Psitive, nn-destructive ver-vltage prtectin must be prvided. Lad fault ver-current prtectin must be prvided Heat transfer f the nminal 2 watts per amp per phase (generated by all silicn pwer semicnductrs) must be prvided. The fllwing addresses the last requirement in sme detail. HEAT GENERATOR It is nt widely understd that all silicn semicnductrs have a frward vltage drp in the cnductin directin which, fr pwer silicn, typically ranges frm 1 vlt t mre than 2 vlts. This vltage drp multiplied by the amunt f RMS current flw thrugh the devices results in watts - smetimes kilwatts - f heat at the semicnductr munting pint. Watts = V drp x I If this heat is cncentrated at the semicnductr and nt allwed t dissipate, the temperature f the device will very rapidly rise t levels far greater than an SCR s 125 C maximum rating* (fr a dide, this temperature is typically 200 C) causing instant destructin. The prper internatinal nmenclature fr this frward cnductin drp is: Dides: VFM = Maximum peak frward vltage rated current SCR s VTM = Maximum n-state rated current Referring t the typical AC switch r DC bridge arrangements f varius slid state pwer circuits (see Payne Engineering Applicatin Nte 8-18), it is quite evident that the equipment designer must prvide a heat transfer mechanism t dissipate literally thusands f watts n any given cntrl assembly. A 600HP, three phase mtr starter, fr example, wuld generate ver 600 watts f heats t be dissipated, while a much simpler kw heat seal cntrl wuld need t rid itself f nly 0 watts r s. *Typically, 125 C is a true maximum SCR temperature with allwable current at 0 amperes.

3 HEAT ECHANGER DESIGN 26A-12 Ppularly knwn as heatsinks, semicnductr heat exchangers are made in a variety f shapes and sizes, bth air cled and liquid cled. Fr purpses f this paper, nly aluminum air cled heatsinks will be cnsidered in detail. The future fr wide, ecnmical usage f slid state pwer cntrls requires a lightweight but cmpact and reliable design with MBTF (mean time between failures) measured in years. This requirement immediately precludes fan cling f clsely spaced fin heatsink designs which can fill up and be clgged with dirt and st in a few years. The Mdel 26A Series f aluminum heatsinks has been painstakingly develped since 1961 t prvide natural cnvectin and radiatin cling f pwer semicnductrs frm 10 amps/phase t 600 amps/ phase, utilizing isthermal crss-sectins with a design gal f 50 C maximum rise under actual cntrl perating cnditins. TABLE 1 26A-16 MODEL # SHAPE AMPERES MA C/watt SP WGT lb/ft 26A-12 channel beam A-16 6 channel beam A-11 6 channel beam A-10 flat puck A flat puck A-11 Extensin f the 26A-6 and the 26A-10 designs t accmmdate the range frm 600A t 1200A, utilizing nly the standard 1/2 (110mm) 100 cfm muffin fans, was dne t ensure that the fan replacements culd be easily fund anywhere in the wrld. The wide, natural cnvectin fin spacing has prven t be cmpletely resistant t dirt buildup since their first service in the early 1970 s. Fr the relatively lw temperature vertical surface type heat exchanger f the abve, the heat transferred by cnvectin is substantially larger than that transferred by radiatin. The ttal can be calculated by the fllwing: 26A-10 Eq. 1.0 q = (h c + h r ) A (T s - T amb ) q = watts h c = cnvectin heat transfer cefficient (watt/cm 2 C) h r = radiatin heat transfer cefficient (watt/cm 2 C) A = heat transfer surface area (cm 2 ) 26A-6

4 CONVECTION Experimental studies f cnvectin heat transfer cefficients frm vertical surfaces tk place in the early years f the 20th century, building up a reasnably reliable base f data. Fig. 1 describes in sme detail the lcal upward air velcity and air temperature gradients in frnt f a 0cm lng heated vertical plate as measured by Schmidt 2. Fig. 2 is an interfermeter phtgraph by Kennard, illustrating the natural cnvectin air stream appraching a vertical plate 0 C htter than rm ambient air Temperature Velcity Lcal Velcity, cm/sec Lcal Temperature, deg C Fig y = Millimeters The relatively cmplex equatins and data acquired by these early researchers can be simplified fr ur lw temperature laminar flw t the fllwing: Eq. 1.1 h c = K F [ (T - T ).25 S amb L ] K F = fin cnstant L = vertical length f fin (cm) T S = fin surface temperature C Further experiments by Payne Engineering in the late 1960 s indicated a very high sensitivity f the fin cnstant K F, t any frm f blckage which culd create turbulent flw disturbances t the heatsink. It was fund that these disturbances seriusly degrade the cling ability f the heatsink and further that additinal fins and/r heatsink mass did nt alleviate the situatin. The mst effective fin designs were determined t be thse that allwed the highest velcities, appraching 1 meter/sec per Fig. 1, ver the greatest wetted area. Fig. 2 - Interfermeter phtgraph demnstrating laminar airflw.

5 RADIATION The heat emitting capability f a black bdy depends n its temperature nly, and the Stefan- Bltzmann Law 1 defines the radiant heat utput f a perfectly black bdy as q = 5.67 x 10-8 T s watts/m 2. Frm this relatinship, the radiant heat transfer cefficient can be stated as: Eq. 1.2 h r =.22 x e(1-f) [ T + T + 27 S amb ] e = surface emissivity F = fin shield factr Fr multiple fin heatsinks, with clsely spaced fins radiating int each ther, calculatin f shielding factr F is unreliable at best. Experimental tests, including tests in an evacuated chamber, indicate that cnvectin heat transfer cntributes apprximately 85% and radiatin 15% f the ttal heat dissipated ver a semicnductr ampere range frm 10A t 50A. EFFICIENCY = EUROPE 5 = USA = Mdel 26A natural cnvectin The effectiveness f heatsink designs is nrmally expressed in C temperature rise f the semicnductr case fr a given wattage input. Rearranging Eq. 1.0 allws ne t create a thermal resistance cefficient in C/watt fr any heatsink design: 1.0 C/watt 26A-12 26A-11 26A-16 26A-10 Eq. 1. = 1 C (h = c + h r ) A watt 26A C/watt lb 10 lbs.5 kg.5 kg 50 C rise vs. heatsink weight (Al) () lb s. kg. = EUROPE 5 = USA = Mdel 26A natural cnvectin 26A-12 26A-16 26A-11 26A-10 f rce d a ir 26A Thermal 50 C rise vs. Heatsink Vlume in cm A = heatsink wetted area (cm 2 ). Fig. details the abve vs. heatsink vlume and weight fr cmmercially available heatsinks in Eurpe and the USA. The equipment designer can increase wetted area A with nly decreasing returns in rder t lwer his thermal resistance. T achieve values less than.10 C/watt fr cling requirements f 600 amp and larger cntrls requires frced air r liquid cling. The 26A-6 design 100 cfm muffin fans can achieve a value f.075, half the best natural cnvectin value shwn. 5

6 ENCLOSURE CONSIDERATIONS The classical electrical enclsure ver the years has nrmally been a clsed, dusttight bx whse design has been primarily dictated by the sensitivity f electrmechanical cntrls t dust, dirt and crrsin. Heat and/r temperature buildup were rarely serius cnsideratins with such cntrls. Hwever, the advent f slid state cntrls fr bth lgic and pwer handling duty mandates a different design apprach t electrical enclsures. 1.. Dust, dirt and crrsin are nt serius cncerns fr a prperly designed slid state pwer cntrl. Temperature extremes, hwever, can be catastrphic fr any semicnductr device as well as its assciated electrnic cmpnents. The 50 C ambient temperature, which is a generally accepted internatinal standard, is a very demanding requirement when ne adds a 50 C temperature rise t the semicnductr case and is left with nly a 25 C wrking range t the 125 C zer current rating f nrmal SCR s. Of curse, fr a silicn dide this is n prblem, since mst dide designs are flat-rated t at least 10 C and their zer current des nt ccur until 200 C. Obviusly, the mre dides ne can substitute fr SCRs, the mre reliable will be ne s cntrl. In additin, many electrnic cmpnents have nly a 1000 hur life at a typical specificatin temperature f 65 C. Therefre, it is clear that the end user must carefully cnsider his electrical enclsure fr prper internal cling - usually by cnvectin ventilatin t the utside ambient. Lw pwer slid state cntrls f less than 80 amps per phase r three phase mtr starters up t 50 HP can be successfully installed in cnventinal sealed dust free enclsures with derating and/r with an islated heatsink enclsure design. Hwever, nte Fig. where a typical sealed enclsure f substantial size (0 x 20 x 10 in. r 76 x 51 x 25 mm) has been tested fr it s internal ambient temperature fr varius pwer cntrl utputs. Even a relatively benign 60 amp three phase external lad has caused the inside ambient t exceed 50 C with nly a 22 C utside enclsure ambient. 110 Temp C 65 Outside ambient temp = 22 C Watts Amps/Phase (Mtr Lad) Fig.. Inside temperature f dust-tight NEMA 12 steel enclsure (0 x 20 x 10 in.) vs. semicnductr watt lsses. 6

7 The same enclsure with ventilatin screens* n the bttm and each tp side wuld then have nly a small increase f it s inside ambient temperature simply because utside ambient air is allwed t flw freely - using the buyancy frces f natural cnvectin - up thrugh the bttm enclsure screen past a vertically munted slid state heatsink and ut the tp vents. Finally, ne shuld nt be cmplacent abut simply using a larger sealed enclsure and/r emplying internal fans t cl the slid state cntrl. Fans add even mre air frictin surce energy t the enclsure ambient and may nly delay the inevitable ver-temperature cnditin when the silicn frward drp wattage is nt prperly dissipated t the surrunding ambient. *Typically, 10 in 2 (65cm 2 ) f inlet area (and equal utlet area) fr every 50 amp phase is usually sufficient in mst industrial plant applicatins. References McAdams, W.H. Heat Transmissin, McGraw-Hill, 195 Schmidt, E. Zeitung des Kälte-Ind. N. 5, pg. 21, 1928 Kennard, R.B. Nat. Bur. Stds. J. Research N. 8, pg. 787, 192 Phillips, N.V. Semicnductr Data Handbk, 1972 EG&G/Wakefield Cat. 100/0M,