rev. a information furnished by analog devices is believed to be accurate and reliable. however, no responsibility is assumed by analog devices for its use, nor for any infringements of patents 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 analog devices. a high speed, precision sample-and-hold amplifier ad585 one technology way, p.o. box 9106, norwood, ma 02062-9106, u.s.a. tel: 617/329-4700 fax: 617/326-8703 features 3.0 m s acquisition time to 6 0.01% max low droop rate: 1.0 mv/ms max sample/hold offset step: 3 mv max aperture jitter: 0.5 ns extended temperature range: C55 8 c to +125 8 c internal hold capacitor internal application resistors 6 12 v or 6 15 v operation available in surface mount applications data acquisition systems data distribution systems analog delay & storage peak amplitude measurements mil-std-883 compliant versions available product description the ad585 is a complete monolithic sample-and-hold circuit consisting of a high performance operational amplifier in series with an ultralow leakage analog switch and a fet input inte- grating amplifier. an internal holding capacitor and matched applications resistors have been provided for high precision and applications flexibility. the performance of the ad585 makes it ideal for high speed 10- and 12-bit data acquisition systems, where fast acquisition time, low sample-to-hold offset, and low droop are critical. the ad585 can acquire a signal to 0.01% in 3 m s maximum, and then hold that signal with a maximum sample-to-hold offset of 3 mv and less than 1 mv/ms droop, using the on-chip hold capacitor. if lower droop is required, it is possible to add a larger external hold capacitor. the high speed analog switch used in the ad585 exhibits aperture jitter of 0.5 ns, enabling the device to sample full scale (20 v peak-to-peak) signals at frequencies up to 78 khz with 12-bit precision. the ad585 can be used with any user-defined feedback net- work to provide any desired gain in the sample mode. on-chip precision thin-film resistors can be used to provide gains of +1, C1, or +2. output impedance in the hold mode is sufficiently low to maintain an accurate output signal even when driving the dynamic load presented by a successive-approximation a/d converter. however, the output is protected against damage from accidental short circuits. the control signal for the hold command can be either active high or active low. the differential hold signal is compatible with all logic families, if a suitable reference level is prov ided. an on-chip ttl reference level is provided for ttl compatibility. the ad585 is available in three performance grades. the jp grade is specified for the 0 c to +70 c commercial temperature range and packaged in a 20-pin plcc. the aq grade is speci- fied for the C25 c to +85 c industrial temperature range and is packaged in a 14-pin cerdip. the sq and se grades are speci- fied for the C55 c to +125 c military temperature range and are packaged in a 14-pin cerdip and 20-pin lcc. product highlights 1. the fast acquisition time (3 m s) and low aperture jitter (0.5 ns) make it the first choice for very high speed data acquisition systems. 2. the droop rate is only 1.0 mv/ms so that it may be used in slower high accuracy systems without the loss of accuracy. 3. the low charge transfer of the analog switch keeps sample-to hold offset below 3 mv with the on-chip 100 pf hold capaci- tor, eliminating the trade-off between acquisition time and s/h offset required with other shas. 4. the ad585 has internal pretrimmed application resistors for applications versatility. 5. the ad585 is complete with an internal hold capacitor for ease of use. capacitance can be added externally to reduce the droop rate when long hold times and high accuracy are required. 6. the ad585 is recommended for use with 10- and 12-bit successive-approximation a/d converters such as ad573, ad574a, ad674a, ad7572 and ad7672. 7. the ad585 is available in versions compliant with mil-std- 883. refer to the analog devices military products databook or current ad585/883b data sheet for detailed specifications. functional block diagram dip lcc/plcc package
ad585Cspecifications model ad585j ad585a ad585s min typ max min typ max min typ max units sample/hold characteristics acquisition time, 10 v step to 0.01% 333 m s 20 v step to 0.01% 5 5 5 m s aperture time, 20 v p-p input, hold 0 v 353535ns aperture jitter, 20 v p-p input, hold 0 v 0.5 0.5 0.5 ns settling time, 20 v p-p input, hold 0 v, to 0.01% 0.5 0.5 0.5 m s droop rate 111 mv/ms droop rate t min to t max doubles every 10 c double every 10 c doubles every 10 c charge transfer 0.3 0.3 0.3 pc sample-to-hold offset C3 3 C3 3 C3 3 mv feedthrough 20 v p-p, 10 khz input 0.5 0.5 0.5 mv transfer characteristics 1 open loop gain v out = 20 v p-p, r l = 2k 200,000 200,000 200,000 v/v application resistor mismatch 0.3 0.3 0.3 % common-mode rejection v cm = 10 v 80 80 80 db small signal gain bandwidth v out = 100 mv p-p 2.0 2.0 2.0 mhz full power bandwidth v out = 20 v p-p 160 160 160 khz slew rate v out = 20 v p-p 10 10 10 v/ m s output resistance (sample mode) i out = 10 ma 0.05 0.05 0.05 w output short circuit current 50 50 50 ma output short circuit duration indefinite indefinite indefinite analog input characteristics offset voltage 522 mv offset voltage, t min to t max 63 3 mv bias current 222 na bias current, t min to t max 5520 50 2 na input capacitance, f = 1 mhz 10 10 10 pf input resistance, sample or hold 20 v p-p input, a = +1 10 12 10 12 10 12 w digital input characteristics ttl reference output 1.2 1.4 1.6 1.2 1.4 1.6 1.2 1.4 1.6 v logic input high voltage t min to t max 2.0 2.0 2.0 v logic input low voltage t min to t max 0.8 0.8 0.7 v logic input current (either input) 50 50 50 m a power supply characteristics operating voltage range +5, C10.8 18 +5, C10.8 18 +5, C10.8 18 v supply current, r l = 610610610 ma power supply rejection, sample mode 70 70 70 db temperature range specified performance 0 +70 C25 +85 C55 +125 c package options 3, 4 cerdip (q-14) ad585aq ad585sq lcc (e-20a) ad585se plcc (p-20a) ad585jp (typical @ +25 8 c and v s = 6 12 v or 6 15 v, and c h = internal, a = +1, hold active unless otherwise noted) specifications subject to change without notice. specifications shown in boldface are tested on all production units at final electrical test. results from those tests are used to calculate outgoing quality levels. all min and max specifications are guaranteed, although only those shown in boldface are tested on all production units. notes 1 maximum input signal is the minimum supply minus a headroom voltage of 2.5 v. 2 not tested at C55 c. 3 e = leadless ceramic chip carrier; p = plastic leaded chip carrier; q = cerdip. 4 for ad585/8 83b specifications, refer to analog devices military products databook. rev. a C2C
ad585 rev. a C3C figure 2. acquisition time vs. hold capacitance (10 v step to 0.01%) absolute maximum ratings supplies (+v s , Cv s ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 v logic inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v s analog inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v s r in , r fb pins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v s storage temperature . . . . . . . . . . . . . . . . . . . C65 c to +150 c lead temperature (soldering) . . . . . . . . . . . . . . . . . . . 300 c output short circuit to ground . . . . . . . . . . . . . . . . indefinite ttl logic reference short circuit to ground . . . . . . . . . . . . . . . . . . . . . . . . . indefinite
ad585 rev. a C4C sampled data systems in sampled data systems there are a number of limiting factors in digitizing high frequency signals accurately. figure 9 shows pictorially the sample-and-hold errors that are the limiting fac- tors. in the following discussions of error sources the errors will be divided into the following groups: 1. sample-to-hold transi- tion, 2. hold mode and 3. hold-to-sample transition. figure 9. pictorial showing various s/h characteristics sample-to-hold transition the aperture delay time is the time required for the sample-and- hold amplifier to switch from sample to hold. since this is effec- tively a constant then it may be tuned out. if however, the aperture delay time is not accounted for then errors of the mag- nitude as shown in figure 10 will result. figure 10. aperture delay error vs. frequency to eliminate the aperture delay as an error source the sample- to-hold command may be advanced with respect to the input signal . once the aperture delay time has been eliminated as an error source then the aperture jitter which is the variation in aperture delay time from sample-to-sample remains. the aperture jitter is a true error source and must be considered. the aperture jitter is a result of noise within the switching network which modu- lates the phase of the hold command and is manifested in the variations in the value of the analog input that has been held. the aperture error which results from this jitter is directly re- lated to the dv/dt of the analog input. the error due to aperture jitter is easily calculated as shown be- low. the error calculation takes into account the desired accu- racy corresponding to the resolution of the n-bit a/d converter. f max = 2 ( n + 1) p ( aperture jitter ) for an application with a 10-bit a/d converter with a 10 v full scale to a 1/2 lsb error maximum. f max = 2 (10 + 1) p (0.5 10 9 ) f max = 310.8 khz . for an application with a 12-bit a/d converter with a 10 v full scale to a 1/2 lsb error maximum: f max = 2 (12 + 1) p (0.5 10 9 ) f max = 77.7 khz . figure 11 shows the entire range of errors induced by aperture jitter with respect to the input signal frequency. figure 11. aperture jitter error vs. frequency sample-to-hold offset is caused by the transfer of charge to the holding capacitor via the gate capacitance of the switch when switching into hold. since the gate capacitance couples the switch control voltage applied to the gate on to the hold capaci- tor, the resulting sample-to-hold offset is a function of the logic level . the logic inputs were designed for application flexibility and, therefore, a wide range of logic thresholds. this was achieved by using a differential input stage for hold and hold . figure 1 shows the change in the sample-to-hold offset voltage based upon an independently programmed reference voltage. since the input stage is a differential configuration, the offset voltage is a function of the control voltage range around the pro- grammed threshold voltage. the sample-to-hold offset can be reduced by adding capacitance to the internal 100 pf capacitor and by using hold instead of hold. this may be easily accomplished by adding an external capacitor between pins 7 and 8. the sample-to-hold offset is then governed by the relationship: s / h offset ( v ) = charge pc () c h total ( pf )
ad585 rev. a C5C for the ad585 in particular it becomes: s / h offset ( v ) = 0 . 3 pc 100 pf + c ext () the addition of an external hold capacitor also affects the acqui- sition time of the ad585. the change in acquisition time with respect to the c ext is shown graphically in figure 2. hold mode in the hold mode there are two important specifications that must be considered; feedthrough and the droop ra te. feedthrough errors appear as an attenuated version of the input at the output while in the hold mode. hold-mode feedthrough varies with fre- quency, increasing at higher frequencies. feedthrough is an im- portant specification when a sample and hold follows an analog multiplexer that switches among many different channels. hold-mode droop rate is the change in output voltage per unit of time while in the hold mode. hold-mode droop originates as leakage from the hold capacitor, of which the major leakage current contributors are switch leakage current and bias current. the rate of voltage change on the capacitor dv/dt is the ratio of the total leakage current i l to the hold capacitance c h . droop rate = dv out dt ( volts / sec ) = i l ( pa ) c h ( pf ) for the ad585 in particular; droop rate = 100 pa 100 pf + ( c ext ) additionally the leakage current doubles for every 10 c increase in temperature above 25 c; therefore, the hold-mode droop rate charact eristic will also double in the same fas hion. the hold-mode droop rate can be traded-off with acquisition time to provide the best combination of droop error and acquisition time. the tradeoff is easily accomplished by varying the value of c ext . since a sample and hold is used typically in combination with an a/d converter, then the total droop in the output voltage has to be less than 1/2 lsb during the period of a conversion. the maximum allowable signal change on the input of an a/d converter is: d v max = full -scale voltage 2 n + 1 () once the maximum d v is determined then the conversion time of the a/d converter (t conv ) is required to calculate the maxi- mum allowable dv/dt. dv dt max = d v max t conv the maximum dv max dt as shown by the previous equation is the limit not only at 25 c but at the maximum expected operat- ing temperature range. therefore, over the operating temper ature range the following criteria must be met (t operation C25 c) = d t. dv 25 c dt 2 d t c () 10 c dv max dt hold-to-sample transition the nyquist theorem states that a band-limited signal which is sampled at a rate at least twice the maximum signal frequency can be reconstructed without loss of information. this means that a sampled data system must sample, convert and acquire the next point at a rate at least twice the signal frequency. thus the maximum input frequency is equal to f max = 1 2 t acq + t conv + t ap () where t acq is the acquisition time of the sample-to-hold amplifier, t ap is the maximum aperture time (small enough to be ignored) and t conv is the conversion time of the a/d converter. data acquisition systems the fast acquisition time of the ad585 when used with a high speed a/d converter allows accurate digitization of high fre- quency signals and high throughput rates in multichannel data acquisition systems. the ad585 can be used with a number of different a/d converters to achieve high throughput rates. fig- ures 12 and 13 show the use of an ad585 with the ad578 and ad574a. figure 12. a/d conversion system, 117.6 khz throughput 58.8 khz max signal input figure 13. 12-bit a/d conversion system, 26.3 khz throughput rate, 13.1 khz max signal input
ad585 rev. a C6C printed in u.s.a. c851cC5C4/89 logic input the sample-and-hold logic control was designed for versatile logic interfacing. the hold and hold inputs may be used with both low and high level cmos, ttl and ecl logic sys- tems. logic threshold programmability was achieved by using a differential amplifier as the input stage for the digital inputs. a predictable logic threshold may be programmed by referencing either hold or hold to the appropriate threshold voltage. for example, if the internal 1.4 v reference is applied to hold an input signal to hold between +1.8 v and +v s will place the ad585 in the hold mode. the ad585 will go into the sample mode for this case when the input is between Cv s and +1.0 v. the range of references which may be applied is from (Cv s +4 v) to (+v s C3 v). optional capacitor selection if an additional capacitor is going to be used in conjunction with the internal 100 pf capacitor it must have a low dielectric absorption. dielectric absorption is just that; it is the charge absorbed into the dielectric that is not immediately added to or removed from the capacitor when rapidly charged or discharged. the c apacitor with dielectric absorption is modeled in f igure 14. figure 14. capacitor model with dielectric absorption if the capacitor is charged slowly, c da will eventually charge to the same value as c. but unfortunately, good dielectrics have very high resistances, so while c da may be small, r x is large and the time constant r x c da typically runs into the millisecond range. in fast charge, fast-discharge situations the effect of di- electric absorption resembles memory. in a data acquisition system where many channels with widely varying data are being sampled the effect is to have an ever changing offset which ap- pears as a very nonlinear sample-to-hold offset since the differ- ence between the voltage being measured and the voltage previ- ously measured determines the fraction by which the dielectric absorption figure is multiplied. it is impossible to readily correct for this error source. the only solution is to use a capacitor with dielectric absorption less than the maximum tolerable error. capacitor types such as polystyrene, polypropy- lene or teflon are recommended. grounding many data-acquisition components have two or more ground pins which are not connected together within the device. these grounds are usually referred to as the logic power return analog common (analog power return), and analog signal ground. these grounds must be tied together at one point, usually at the system power-supply ground. ideally, a single solid ground would be desirable. however, since current flows through the ground wires and etch stripes of the circuit cards, and since these paths have resistance and inductance, hundreds of millivolts can be generated between the system ground point and the ground pin of the ad585. separate ground returns should be provided to minimize the current flow in the path from sensitive points to the system ground point. in this way supply currents and logic-gate return currents are not summed into the same return path as analog signals where they would cause measurement errors. figure 15. basic grounding practice outline dimensions dimensions shown in inches and (mm). 14-pin cerdip (q-14) 20-terminal lcc (e-20a) 20-terminal plcc (p-20a)
|
