1 ? fn7166 caution: these devices are sensitive to electrostatic discharge; follow proper ic handling procedures. 1-888-intersil or 321-724-7143 | intersil (and design) is a registered trademark of intersil americas inc. copyright ? intersil americas inc. 2003. all rights reserved. elantec is a registered trademark of elantec semiconductor, inc. all other trademarks mentioned are the property of their respective owners. el4421, el4422, EL4441, el4442, el4443, el4444 multiplexed-input video amplifiers the el44xx family of video multiplexed-amplifiers offers a very quick 8ns switching time and low glitch along with very low video distortion. the amplifiers have good gain accuracy even when driving low-impedance loads. to save power, the amplifiers do not require heavy loading to remain stable. the el4421 and el4422 are two-input multiplexed amplifiers. the -inputs of the input stages are wired together and the device can be used as a pin-compatible upgrade from the max453. the EL4441 and el4442 have four inputs, also with common feedback. these may be used as upgrades of the max454. the el4443 and el4444 are also 4-input multiplexed amplifiers, but both positive and negative inputs are wired separately. a wide variety of gain- and phase-switching circuits can be built using independent feedback paths for each channel. the el4421, EL4441, and el4443 are internally compensated for unity-gain operation. the el4422, el4442, and el4444 are compensated for gains of +2 or more, especially useful for driving back-matched cables. the amplifiers have an operatio nal temperature of -40c to +85c and are packaged in plastic 8- and 14-pin dip and 8- and 14-pin so. the el44xx multiplexed-amplifier family is fabricated with elantec's proprietary complementary bipolar process which gives excellent signal symmetry and is very rugged. pinouts features unity or + 2-gain bandwidth of 80mhz 70db off-channel isolation at 4mhz directly drives high-impedance or 75 ? loads 0.02% and 0.02 differential gain and phase errors 8ns switching time < 100mv switching glitch 0.2% loaded gain error compatible with 3v to 15v supplies 160mw maximum dissipa tion at 5v supplies el4421/el4422 (8-pin pdip, so) top view EL4441/el4442 (14-pin pdip, so) top view el4443/el4444 (14-pin pdip, so) top view manufactured under u.s. patent no. 5,352,987 ordering information part nimber temp. range package pkg. no. el4421cn -40c to +85c 8-pin pdip mdp0031 el4421cs -40c to +85c 8-pin so mdp0027 el4422cn -40c to +85c 8-pin pdip mdp0031 el4422cs -40c to +85c 8-pin so mdp0027 EL4441cn -40c to +85c 14-pin pdip mdp0031 EL4441cs -40c to +85c 14-pin so mdp0027 el4442cn -40c to +85c 14-pin pdip mdp0031 el4442cs -40c to +85c 14-pin so mdp0027 el4443cn -40c to +85c 14-pin pdip mdp0031 el4443cs -40c to +85c 14-pin so mdp0027 el4444cn -40c to +85c 14-pin pdip mdp0031 el4444cs -40c to +85c 14-pin so mdp0027 data sheet january 1996, rev. c o b s o l e t e p r o d u c t n o r e c o m m e n d e d r e p l a c e m e n t c o n t a c t o u r t e c h n i c a l s u p p o r t c e n t e r a t 1 - 8 8 8 - i n t e r s i l o r w w w . i n t e r s i l . c o m / t s c
2 notes: 1. the '21, '41, and '43 devices are connected for uni ty-gain operation with 75 ? load and an input span of 1v. the '22, '42, and '44 devices are connected for a gain of +2 with a 150 ? load and a 1v input span with r f = r g = 270 ? . 2. the '21 and '41 devices are connected for unity gain with a 3v input span while the output swing is measured. 3. cmir is assured by passing the cmrr test at input voltage extremes. absolute maximum ratings (t a = 25c) v+ positive supply voltage . . . . . . . . . . . . . . . . . . . . . . 16.5v v s v+ to v- supply voltage . . . . . . . . . . . . . . . . . . . . . . . .33v v in voltage at any input or feedback . . . . . . . . . . . . v+ to v- v in difference between pairs of inputs or feedback . . . . . .6v v logic voltage at a0 or a1 . . . . . . . . . . . . . . . . . . . . . . -4v to 6v i in current into any input, feedback, or logic pin . . . . . 4ma i out output current. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30ma p d maximum power dissipation . . . . . . . . . . . . . see curves caution: stresses above those listed in ?a bsolute maximum ratings? may cause permanent damage to the device. this is a stress o nly rating and operation of the device at these or any other conditions above those indicated in the operational sections of this specification is not implied. important note: all parameters having min/max specifications are guaranteed. typical values are for information purposes only. u nless otherwise noted, all tests are at the specified temperature and are pulsed tests, therefore: t j = t c = t a open-loop dc electrical specifications power supplies at 5v, t a = 25c, r l = 500 ? , unless otherwise specified parameter description min typ max units v os input offset voltage '21, '41, and '43 -9 3 9 mv '22, '42, and '44 -7 2 7 i b input bias current, positive inputs only of the '21, '22, '41, '42, and all inputs of the '43 and '44 -12 -5 0 a i fb input bias currents of common feedback '21 and '22 -24 -10 0 a '41 and '42 -48 -20 0 a i os input offset currents of the '43 and '44 60 350 na e g gain error (note 1) '21 and '41 and '43 0.2 0.6 % '22, '42 and '44 0.1 0.6 v/v a vol open-loop gain (note 1) el4443 350 500 v/v el4444 500 750 v/v v in input signal range, el4421 and EL4441 (note 2) 2.5 3 v cmrr common-mode rejection ratio, el4443 and el4444 70 90 db psrr power supply rejection ratio v s from 5v to 15v 60 70 db cmir common-mode input range el4443 and el4444 (note 3) 2.5 3 v v out output swing 2.5 3.5 v i sc output short-circuit current 40 80 ma f t unselected channel feedthrough attenuation (note 1) '21, '41, '43 70 80 db '22, '42, '44 55 64 db i logic input current at a0 and a1 with input = 0v and 5v -16 -8 0 a v logic logic valid high and low input levels 0.8 2.0 v i s supply current el4421 and el4422 11 14 ma EL4441, el4442, el4443, and el4444 13 16 el4421, el4422, EL4441, el4442, el4443, el4444
3 closed-loop ac electrical specifications power supplies at 5v. t a = 25c, for el4421, EL4441, and el4443 a v = +1 and r l = 500 ? , for el4422, el4442, and el4444 a v = +2 and r l = 150 ? with r f =r g = 270 ? and c f = 3pf; for all c l = 15pf parameter description min typ max units bw - 3db -3db small-signal bandwidth el4421, '41, '43 80 mhz el4422, '42, '44 65 mhz bw 0.1db 0.1db flatness bandwidth 10 mhz peaking frequency response peaking 0.5 db sr slewrate, v out between -2.5v and +2.5v, v s = 12v el4421, EL4441, el4443 150 200 v/sec el4422, el4442, el4444 180 240 v/sec v n input-referred noise voltage density el4421, EL4441, el4443 18 nv/ hz el4422, el4442, el4444 14 nv/ hz d g differential gain error, v offset between -0.7v and +0.7v el4421, ?41, ?43 (v s = 12v) 0.01 % el4421, ?41, ?43 (v s = 5v) 0.10 % el4422, ?42, ?44 (v s = 12v) 0.02 % el4422, ?42, ?44 (v s = 5v) 0.11 % d ? differential phase error, v offset between -0.7v and +0.7v el4421, ?41, ?43 (v s = 12v) 0.01 el4421, ?41, ?43 (v s = 5v) 0.1 el4422, ?42, ?44 (v s = 12v) 0.02 el4422, ?42, ?44 (v s = 5v) 0.15 t mux multiplex delay time, logic threshold to 50% signal change el4421, '22 8 nsec EL4441, '42, '43, '44 12 nsec v glitch peak multiplex glitch el4421, '22 70 mv EL4441, '42, '43, '44 100 mv iso channel off isolation at 3.58mhz (see text) el4421, EL4441, el4443 76 db el4422, el4442, el4444 63 db el4421, el4422, EL4441, el4442, el4443, el4444
4 typical performance curves el4421, EL4441, and el4443 frequency re sponse for various gains el4421, EL4441, and el4443 large-signal response v s = 12v, r l = 500 ? el4422, el4442, and el4444 frequency response for various gains el4421, EL4441, and el4443 small-signal transient response v s = 5v, r l = 500 ? el4421, EL4441, and el4443 frequency response for various loads v s = 5v, a v = +1 el4422, el4442, and el4444 frequency response for various loads v s = 5v, a v = +2 frequency response for various loads v s = 15v, a v = + 1 el4422, el4442, and el4444 frequency response for various loads v s = 15v, a v = +2 el4421, el4422, EL4441, el4442, el4443, el4444
5 typical performance curves (continued) el4443 open-loop gain and phase vs. frequency el4444 open-loop gain and phase vs. frequency el4421, EL4441, and el4443 -3db bandwidth, slewrate, and peaking vs. supply voltage el4422, el4442, and el4444 -3db bandwidth, slewrate, and peaking vs. supply voltage el4421, EL4441, and el4443 bandwidth, slewrate, and peaking vs. temperature, a v = +1, r l =500 ? el4422, el4442, and el4444 bandwidth, slewrate, and peaking vs. temperature, a v = +2, r l = 150 ? , r i = r g = 270 ? , c f = 3pf el4421, EL4441, and el4443 -3db bandwidth and gain error vs. load resistance input noise vs. frequency el4421, el4422, EL4441, el4442, el4443, el4444
6 typical performance curves (continued) el4421, EL4441, and el4443 differential gain and phas e errors, vs. input offset, a v = +1, r l = 500 ? , f = 3.58mhz el4422, el4442, and el4444 differential gain and phase error vs. input offset; a v = +2, r l = 150 ? , f = 3.58mhz el4421, EL4441, and el4443 differential gain and phase error vs. load resistance; a v = +1, f = 3.58mhz, v offset = 0 0.714v el4443 and el4444 open-loop gain vs. load resistance change in v os , a v , and i b with supply voltage change in v os , i b , and a v vs. temperature switching waveforms switching from grounded input to uncorrelated sinewave and back channel-to-channel switching glitch el4421, el4422, EL4441, el4442, el4443, el4444
7 typical performance curves (continued) el4422, el4442, and el4444 unselected channel feedthrough vs. frequency el4443 and el4444 input and output range vs. supply voltage (output unloaded) el4421, EL4441, and el4443 unselected channel feedthrough vs. frequency supply current vs. supply voltage supply current vs. temperature 8-pin package power dissipation vs. ambient temperature 14-pin package power dissipation vs. ambient temperature el4421, el4422, EL4441, el4442, el4443, el4444
8 applications information general description the el44xx family of video mux-amps are composed of two or four input stages whose inputs are selected and control an output stage. one of the inputs is active at a time and the circuit behaves as a traditional voltage-feedback op-amp for that input, rejecting signals present at the unselected inputs. selection is controlled by one or two logic inputs. the el4421, el4422, EL4441, and el4442 have all -inputs wired in parallel, allowing a single feedback network to set the gain of all inputs. these devices are wired for positive gains. the el4443 and el4444, on the other hand, have all +inputs and -inputs brought out s eparately so that the input stage can be wired for independent gains and gain polarities with separate feedback networks. the el4421, EL4441, and el4443 are compensated for unity-gain stability, while the el4422, el4442, and el4444 are compensated for a fed-back gain of +2, ideal for driving back-terminated cables or maintaining bandwidth at higher fed-back gains. switching characteristics the logic inputs work with standa rd ttl levels of 0.8v or less for a logic 0 and 2.0v or more for a logic 1, making them compatible for ttl and cmos drivers. the ground pin is the logic threshold biasing reference. the simplified input circuitry is shown in figure 1 below. the ground pin draws a maximum dc current of 6a, and may be biased anywhere between (v-) +2.5v and (v+) -3.5v. the logic inputs may range from (v-)+2.5v to v+, and are additionally required to be no more negative than v(gnd pin)-4v and no more positive than v(gnd pin)+6v. for example, within these constraints, we can power the el44xx's from +5v and +12v without a negative supply by using these connections. the logic input(s) and ground pin are shifted 2.5v above system ground to correctly bias the mux-amp. of course, all the signal inputs and output will have to be shifted 2.5v above system ground to ensure proper signal path biasing. a final caution: the ground pin is also connected to the ic's substrate and frequency compensation components. the ground pin must be returned to system ground by a short wire or nearby bypass capacitor. in figure 2, the 22k ? resistors also serve to isolate the bypassed ground pin from the +5v supply noise. signal amplitudes signal input and output voltage s must be between (v-)+2.5v and (v+)-2.5v to ensure linearity. additionally, the differential voltage on any input stage must be limited to 6v to prevent damage. in unity-gain connections, any input could have 3v applied and the output would be at 3v, putting us at our 6v differential limit. higher-gain circuit applications divide the output voltage and allow for larger outputs. for instance, at a gain of +2 the maximum input is again 3v and the output swing is 6v. the el4443 or el4444 can be wired for inverting gain with even more amplitude possible. the output and positive inputs respond to overloading amplitudes correctly; that is, they simply clamp and remain monotonic with increasing +input overdrive. a condition exists, however, where the -input of an active stage is overdriven by large outputs. this occurs mainly in unity-gain connections, and only happens for negative inputs. the overloaded input cannot control the feedback loop correctly and the output can become non-monotonic. a typical scenario has the circuit running on 5v supplies, connected for unity gain, and the input is the maximum 3v. negative input extremes can cause the output to jump from -3v to figure 1. simplified logic input circuitry figure 2. using the el44xx mux amps with +5v and +12v supplies el4421, el4422, EL4441, el4442, el4443, el4444
9 around -2.3v. this will never happen if the input is restricted to 2.5v, which is the guaranteed maximum input compliance with 5v supplies, and is not a problem with greater supply voltages. connecting the feedback network with a divider will prevent the overloaded output voltage from being large enough to overload the -input and monotonic behavior is assured. in any event, keeping signals within guaranteed compliance limits will assure freedom from overload problems. the input and output ranges ar e substantially constant with temperature. power supplies the mux-amps work well on any supplies from 3v to 15v. the supplies may be of different voltages as long as the requirements of the gnd pin are observed (see the switching characteristics se ction for a discussion). the supplies should be bypassed close to the device with short leads. 4.7f tantalum capacitors are very good, and no smaller bypasses need be placed in parallel. capacitors as small as 0.01f can be used if small load currents flow. single-polarity supplies, such as +12v with +5v can be used as described in the switching characteristics section. the inputs and outputs will have to have their levels shifted above ground to accommodate the lack of negative supply. the dissipation of the mux-amps increases with power supply voltage, and this must be compatible with the package chosen. this is a clos e estimate for the dissipation of a circuit: p d = 2v s i s ,max + (v s ?v o ) v o /r pa r where i s , max is the maximum supply current v s is the supply voltage (assumed equal) v o is the output voltage r pa r is the parallel of all resistors loading the output for instance, the el4422 draws a maximum of 14ma and we might require a 2v peak output into 150 ? and a 270 ? +270 ? feedback divider. the r pa r is 117 ? . the dissipation with 5v supplies is 191mw. the maximum supply voltage that the device can run on for a given p d and the other parameter is v s , max = (p d + v o 2 /r pa r )/2i s + v o /r pa r ) the maximum dissipation a package support is p d , max = (t d , max-t a , max)/r th where t d , max is the maximum die temperature, 150c for reliability, less to retain optimum electrical performance t a , max is the ambient temperature, 70 for commercial and 85c for industrial range r th is the thermal resistanc e of the mounted package, obtained from data sheet dissipation curves the most difficult case is the so-8 package. with a maximum die temperature of 150c and a maximum ambient temperature of 85, th e 65 temperature rise and package thermal resistance of 170/w gives a maximum dissipation of 382mw. this allows a maximum supply voltage of 9.2v for the el4422 operated in our example. if the el4421 were driving a light load (r pa r ), it could operate on 15v supplies at a 70 maximum ambient. the EL4441 through el4444 can operate on 12v supplies in the so package, and all parts can be powered by 15v supplies in dip packages. output loading the output stage of the mux-amp is very powerful, and can source 80ma and sink 120ma. of course, this is too much current to sustain and the part will eventually be destroyed by excessive dissipation or by metal traces on the die opening. the metal traces are completely reliable while delivering the 30ma continuous output given in the absolute maximum ratings table in this data sheet, or higher purely transient currents. gain or gain accuracy degrades only 10% from no load to 100 ? load. heavy resistive loading will degrade frequency response and video distortion only a bit, becoming noticeably worse for loads < 100 ? . capacitive loads will cause peaking in the frequency response. if capaciti ve loads must be driven, a small-valued series resistor can be used to isolate it. 12 ? to 51 ? should suffice. a 22 ? series resistor will limit peaking to 2.5db with even a 220pf load. input connections the input transistors can be driven from resistive and capacitive sources but are capable of oscillation when presented with an inductive input. it takes about 80nh of series inductance to make the inputs actually oscillate, equivalent to four inches of unshielded wiring or about 6 of unterminated input transmission line. the oscillation has a characteristic frequency of 500mhz. often simply placing one's finger (via a metal probe) or an oscilloscope probe on the input will kill the oscillation. normal high-frequency construction obviates any such problems, where the input source is reasonably close to the mux-amp input. if this is not possible, one can insert series resistors of around 51 ? to de-q the inputs. el4421, el4422, EL4441, el4442, el4443, el4444
10 all intersil u.s. products are manufactured, asse mbled and tested utilizin g iso9000 quality systems. intersil corporation?s quality certifications c an be viewed at www.intersil.com/design/quality intersil products are sold by description only. intersil corporation reserves the right to make changes in circuit design, soft ware and/or specifications at any time without notice. accordingly, the reader is cautioned to verify that data sheets are current before placing orders. information furnishe d by intersil is believed to be accurate and reliable. however, no responsibility is assumed by intersil or its subsidiaries for its use; nor for any infringements of paten ts or other rights of third parties which may result from its use. no license is granted by implication or otherwise under any patent or patent rights of intersil or its subsidiari es. for information regarding intersil corporation and its products, see www.intersil.com feedback connections a feedback divider is used to increase circuit gain, and some precautions should be observed. the first is that parasitic capacitance at the -input w ill add phase lag to the feedback path and increase frequency response peaking or even cause oscillation. one solution is to choose feedback resistors whose parallel value is low. the pole frequency of the feedback network should be maintained above at least 200mhz. for a 3pf parasitic, th is requires that the feedback divider have less than 265 ? impedance, equivalent to two 510 ? resistors when a gain of +2 is desired. alternatively, a small capacitor across r f can be used to create more of a frequency-compensated divider. the value of the capacitor should match the parasitic capacitance at the -input. it is also practical to place small capacitors across both the feedback resistors (whose values maintain the desired gain) to swamp out parasitics. for instance, two 10pf capacitors across equal divider resistors will dominate parasitic effects and allow a higher divider resistance. the other major concern about the divider concerns unselected-channel crosstal k. the differential input impedance of each input stage is around 200k ? . the unselected input's signal sources thus drive current through that input impedance into the feedback divider, inducing an unwanted output. the gain from unselected input to output, the crosstalk attenuation, if r f /r in . in unity-gain connection the feedback resistor is 0 ? and very little crosstalk is induced. for a gain of +2, the crosstalk is about -60db. feedthrough attenuation the channels have different crosstalk levels with different inputs. here is the typical attenuation for all combinations of inputs for the mux-amps at 3.58mhz: switching glitches the output of the mux-amps produces a small ?glitch' voltage in response to a logic input change. a peak amplitude of only about 90mv occurs, and the trans ient settles out in 20ns. the glitch does not change amplitude with different gain settings. with the four-input multiplexers, when two logic inputs are simultaneously changed, the g litch amplitude doubles. the increase can be a avoided by keeping transitions at least 6ns apart. this can be accomplished by inserting one gate delay in one of the two logic inputs when they are truly synchronous. feedthrough of EL4441 and el4443 at 3.58mhz select inputs, a1 a0 in1 in2 in3 in4 00 selected -77db -90db -92db 01 -80db selected -77db -90db 10 -101db -76db selected -66db 11 -96db -84db -66db selected feedthrough of el4421 at 3.58mhz channel select input, a0 in1 in2 0 selected -88db 1 -93db selected el4421, el4422, EL4441, el4442, el4443, el4444
|
