1963 , Volume v.15 n.3 , Issue Nov-1963
HEWLETT-PACKARD JOURNAL T E C H N I C A L I N F O R M A T I O N F R O M T H E - d p - l A B O R A i O R I E S Vol. 15, No. 3 PUBLISHED BY THE HEWLETT-PACKARD COMPANY, 1501 PAGE MILL ROAD, PALO ALTO, CALIFORNIA NOVEMBER, 1963 A New Multi -Purpose Digital Voltmeter 1 HE accuracy and speed with which digital volt meters make dc voltage measurements have made these instruments extremely useful in a variety of applications. One of their attractive properties is that the digital readout has greater resolution than a meter and is less likely to be misread, par ticularly by unskilled SEE ALSO: personnel. Also, data in Extremely flat response detector, p. 8 digital form may be processed or stored without loss of accuracy, a feature which has made digital voltmeters the basic element in automatic data acquisition sys tems. The high precision, i.e. resolution, attainable with a digital display is of particular interest to many users. Precision of readability by itself, however, is not a guarantee of measurement ac curacy. There is an analog device behind every digital voltmeter display, and the voltmeter can be no more accurate than its associated analog circuitry. With this fact in mind, a new digital voltmeter has been designed whose accuracy has been con servatively specified to be within Â±0.05% of reading Â±1 count. This accuracy holds for long periods of time under a wide range of environ mental conditions. Actually, the calibration of the new voltmeter is held at the factory within Â±0.01% of reading Â± 1 count, the maximum precision obtainable with â€”648.7 --64.87 6.487 â€” +10.5 â€”+10.49 10.487 Decimal Point Position Polarity or Overrange Fig. 1. Front panel plug-ins adapt Model 3440A Digital Volt meter to manual or automatic ranging for bench use or remote ranging for systems use. Unit shown in center foreground is working with -hp- Model 562A Printer to record voltage/tem perature characteristics of circuit operating in environmental chamber. New voltmeter measures dc voltages to an accuracy better than 0.05% of reading Â±1 count. P R I N T E D I N Fig. 2. Printed tape sample shows volt meter output recorded by -hp- Model 562A Printer. Number 9 in polarity column indicates overrange; blank means plus. Numeral in decimal point column is negative exponent of XlO multiplier. Numbers shown at right in terpret recorded values. C O P Y R I G H T U . S . A . © Copr. 1949-1998 Hewlett-Packard Co. 1 9 6 3 H E W L E T T - P A C K A R D C O include a basic manual ranging unit, providing ranges of 10, 100, and 1000 volts full scale, and an automatic ranging unit which can be remotely controlled. Plug-ins under development include a high gain preamplifier, allowing 100 millivolts full scale on the most sensitive range, and a multi-meter plug-in which will measure current and resistance in addition to volt age. OverRange Indicator Fig. 3. Block diagram of -hp- 3440A Digital Voltmeter. the 4-digit readout. The formal spe cification of accuracy is relaxed by a factor of five to 0.05% to allow for aging of the range attenuator and /.ener-diode reference and also to permit use of the voltmeter in a variety of environmental condi tions. Any drift associated with the analog-to-digital conversion system does not enter into the accuracy specification since it can be quickly checked and calibrated out at the front panel by use of the precision internal reference voltage. One source of inaccuracies, of particular concern with high-resolu tion voltmeters, is the impedance of the voltage source being measured. Since the voltmeter's input imped ance has been designed to be con stant regardless of range or sample rate, the error caused by the 10.2 megohm dc input impedance of the Model 3440A Voltmeter is predict able at approximately 0.01% per 1000 ohms in circuits with imped ances higher than 1000 ohms. The voltmeter input impedance has neg ligible effect on the voltage of a source that has an impedance of less than 1000 ohms. PLUG-IN ADAPTABILITY The voltmeter is designed for use with front-panel plug-ins, thus per mitting the basic instrument to be adapted to a large variety of appli cations. Presently available plug-ins Fig. rectangular Model 3440 A Digital Voltmeter uses closely-spaced rectangular Nixie- Readout to achieve numerical display with readable span. Readout system has display storage which retains previous reading while new measurement cycle is in process, insuring steady, flickerless display that changes only when input voltage changes. Front panel calibrate button furnishes quick operational check of voltmeter accuracy. © Copr. 1949-1998 Hewlett-Packard Co. READOUT SYSTEM The 4-digit display uses closespaced, rectangular NixieÂ® tubes in addition to polarity, decimal point, and function indicators. A display storage feature, as on the â€” hpâ€” solidstate counters, retains the reading from the previous sample while a new measurement is being made; the displayed numerals therefore remain steady, changing only if the new reading differs from the pre vious one. The measurement (sample) rate may be varied by a front panel con trol from greater than 5 per second to less than 1 per 5 seconds. The voltmeter makes accurate measure ments quickly, achieving the rated 0.05% accuracy within 450 milli seconds after a step change of input voltage. A HOLD position on the SAMPLE RATE control retains any given measurement on display in definitely. The HOLD position also enables a measurement to be ini tiated by external control. The voltmeter makes measure ments to 5% above full scale with full accuracy on any range, thus ob taining 5-digit resolution just above the decade range switching points. In effect, this amounts to a range overlap. An overrange measurement is indicated as such by the illumi nated OVERRANGE indicator on the front panel. Full overload protection to 1100 volts is provided on all ranges. The input circuit is floating and can ence between these two pulses is a direct analog of the difference be tween the unknown signal voltage and signal ground. The time difference is converted to digital form by the counting of clock pulses. The first-occurring comparator coincidence pulse opens a count gate, enabling the counters DESIGN APPROACH The Model 3440A Digital Volt to operate, and the second pulse meter uses a voltage-to-time-interval closes the gate to terminate the conversion system as outlined in count. The first comparator pulse Fig. 3. The time interval is evalu also starts a Colpitts oscillator ated by digital counter techniques which provides the clock pulses. By appropriate choice of ramp to obtain the digital display. The unknown voltage, appropri slope (400 v/sec) and clock pulse ately attenuated, is applied to a repetition rate (400 kc) the total comparator to which a linearly-de count displayed corresponds to in creasing ramp is also applied. When put millivolts. Range switching the ramp voltage becomes equal to operates an input attenuator and the unknown voltage, a coincidence places the decimal point so that the pulse is generated, as shown by the display reads directly in volts. Input polarity is determined by a timing diagram in Fig. 5. The ramp voltage is also applied to a zero circuit that detects which compara reference comparator which has sig tor pulse occurred last and displays nal ground for its other input. a plus or minus sign accordingly. Again, when the ramp voltage be When readings are taken above full comes equal to signal ground, a scale, an overflow pulse from the coincidence pulse is generated by decade counters (the 10,000th pulse this comparator. The time differ counted) operates an overrange bin- measure voltage sources that are up to Â±400 volts off ground. The instrument is completely transistorized with no electricalmechanical moving parts other than reed relays for attenuator switching and two non-signal relays in the automatic ranging plug-in. Fig. 5. Timing diagram of 3440A Digi tal Voltmeter. Solid lines show opera tion when input voltage is positive and dotted lines show operation with nega tive input voltage. Polarity-sensing circuit determines whether input coin cidence pulse or zero-reference coinci dence pulse occurs last, and switches plus-minus Nixie indicator accordingly. ary which in turn illuminates the OVERRANGE indicator. THE VOLTAGE COMPARATORS The all-solid-state voltage com parators of the new digital volt meter are of special interest. The basic design requirements for the comparator circuits were: sensing level considerably less than one mil- VOLTMETER PLUG-INS & Â© MODEL 3441A MANUAL SELECTOR PLUG-IN The two plug-in units presently avail able for the 3440A Digital Voltmeter are the 3441A Manual Range Selector and the 3442A Automatic Range Selec tor. In addition to range selection, both plug-ins provide the decimal point in dication for the voltmeter and gather the necessary polarity, overrange, and decimal information for the digital re corder input. Choice of range with the Manual Range Selector is made from the plugin front panel by a switch which selects the 10 volt, 100 volt, or 1000 volt range. Since the new digital voltmeter has an overrange indicator and cannot be dam aged by overloads to 1100 volts on any range, the correct range for an un known voltage is found easily. The Automatic Range Selector plugin uses signals and control voltages from the digital voltmeter to place the instrument on the proper voltage range automatically. If the selected range is too low, the OVERRANGE signal immedi ately causes an automatic uprange. If the range is too high (no count regis tered in the leading significant digit), the plug-in downranges. A built-in hys teresis, however, prevents downranging unless the voltage drops to less than 90% of full scale of the lower range. This prevents slight input voltage per turbations or attenuator tolerance over lap from causing erratic range shifting at the decade range change points. © Copr. 1949-1998 Hewlett-Packard Co. MODEL 3442A AUTOMATIC RANGE SELECTOR PLUG-IN As an aid in general purpose meas urements, the sample rate is automatic ally increased to the maximum rate when a range change is initiated. This speed-up lasts for less than one second but insures that the voltmeter quickly switches to the correct range and ac curately displays the input voltage without delay. The Automatic Range Selector also has manual range selection to allow use of the new digital voltmeter's overrange capabilities. CR4, Back- biasing Diode Extreme care has been exercised in selection of the components that determine the ramp slope to ensure ramp stability with respect to time Fig. 6. Input volt and temperature. The temperature age comparator circuit. Zero refer coefficient of ramp reference zener ence comparator diode CR6 matches that of the basecircuit is similar except that circuit emitter junction of ramp generator ground is used as input transistor Ql . Likewise, ramp input voltage. charging resistor R5 has a positive temperature coefficient which can cels the negative temperature coeffi cient of precision polystyrene ramp capacitor C3. THE GATED OSCILLATOR livolt (i.e., circuit noise and equiva lent time jitter approximately 100 microvolts); high input impedance (in excess of 10,000 megohms); low leakage currents (less than one nanoampere); and good tempera ture stability (less than one millivolt drift in sensing level from 0Â°C to 55Â°C). These needs were met by an un usual voltage comparator. As shown in Fig. 6, a matched pair of diodes (CR1 and CR2) with common cath odes is ac-coupled to a gain-stable current amplifier which in turn is coupled to a high-level voltage com parator. The anode of diode CR1 is connected to the source of the ramp voltage while the other anode (CR2) is connected to the unknown input signal voltage. A constant current is supplied to the common cathodes. The ramp, originating at a poten tial greater than full scale input voltage, initially forward-biases diode CR1, thereby reverse-biasing diode CR2. When the linearly decreasing ramp voltage approaches the input voltage, diode CR2 begins to con duct. This current change is capacitively-coupled to the current ampli fier and converted to an equivalent voltage. When this voltage reaches a fixed level, which was chosen to correspond to a current in CR2 equal to that being drawn by ramp diode CR1, it triggers the. high-level voltage comparator, a bistable cir cuit. The diode pair immediately is reverse-biased through diode CR3 so that no further energy is removed from the input circuit. Also, a small amount of charge is transferred back through CR4 into input capaci tor Cl to replace the charge (about 500 pico-coulombs) which was re moved during the act of compari son. This removes any loading of the input attenuator circuit by the comparator and eliminates offsets caused by source impedance varia tions. RAMP LINEARITY One of the fundamental accuracy limits of any digital system is the resolution of the least significant digit. Since the time interval being measured in the new digital volt meter is between two pulses which occur at arbitrary points in time, clock pulse ambiguities can exist around both the first and second comparator pulses. To reduce this effect, the clock pulse oscillator is turned on by the first pulse so that the clock pulses are synchronized with respect to the first comparator pulse. This is made possible by holding the Colpitts LC oscillator with tank capacitors fully charged but with no current in the tank inductance. The current path through the inductance is closed by saturation of a series transistor, thereby starting the oscillator im mediately at full amplitude from a predicted state. The initial calibration accuracy of 0.01% of reading Â±one count includes allowances for the input attenuator, comparator drift, and ramp linearity. (Oscillator fre quency drift and ramp slope varia tions do not limit rated instrument accuracy since these may be cor rected simultaneously by front panel calibration against the inter nal reference.) To permit this accu racy, the ramp must be extremely linear. The ramp is generated by a boot strap circuit which has high loop gain and feedback to the internal ramp amplifier. The feedback cir cuit ensures that the ramp capacitor charging current is constant for the duration of the ramp and that Fig. 7. Ramp generator circuit achieves variations in transistor parameters exceptional linearity with feedback amplifier having input impedance of have no effect. nearly 1,000 megohms. © Copr. 1949-1998 Hewlett-Packard Co. RESOLUTION AND ACCURACY The numerical resolution on the lowest range (10 v) is 1 millivolt. Actually, the instrument may be calibrated to within a fraction of a millivolt by observing the voltage levels at which the least significant digit flickers to the next higher or lower number. The flicker, or un certainty, turns out to be about 0.3 microsecond of ramp or 100 micro volts of the measured voltage. The specification of accuracy al lows for aging and for the tempera ture coefficients of the reference zener diode and the input range re sistors. The nine-volt reference di ode has a maximum temperature coefficient of Â±0.001% per degree C over the instrument operational temperature range. A precision volt age divider across the diode is ad justed at the factory to provide a â€” 8.000 volt input reference for cal ibrating the voltmeter. The input range resistors are ad justed at the factory to better than 0.005% accuracy. These have a max imum temperature coefficient of Â±5 ppm per degree and a long term stability of Â±100 ppm per year. By proper readjustment of the trim mers associated with the reference diode and range resistors, the in strument may be reset to its original factory accuracy of Â±0.01% of read ing Â±1 count in the event of any detectable long term resistance drift. DIGITAL OR ANALOG RECORDING The new voltmeter is designed to drive directly the -hp- Model 562A Digital Recorder or Model 580A Digital-Analog Converter. Each of the four digits, together with polar ity, decimal, and overrange infor mation, is represented by four-line, binary-coded decimal voltages in the 1-2-2* -4 weighted code, avail able at a rear-panel connector. The voltmeter-recorder combina tion operates at a sampling rate de termined by the voltmeter SAMPLE RATE control, or by an external Fig. 8. Specified accuracy limits of Model 3440 A Dig ital Voltmeter show maximum possible errors. Digital error arises from finite resolution of digi tal readout and vanishes when ever digitized value coincides exactly with ac tual input voltage. trigger when the SAMPLE RATE con trol is in the HOLD position. Printer action is initiated by a print com mand pulse from the sample rate multivibrator. When the 3442A Automatic Range Selector plug-in unit is being used, the print com mand pulse is held off for approxi mately 500 ms after the start of an automatic range change. This pre vents the recorder from printing er roneous voltage information while the voltmeter is settling down on a new range. REMOTE CONTROL A 36-pin remote control jack on the rear panel of the new voltmeter permits a range change on remote SPECIFICATIONS -fipMODEL 3440A DC DIGITAL VOLTMETER Voltage Range: 4-digit presentation of 9.999, 99.99 and 999.9 volts full scale; 5% overrange capability with indicator. Full overload protection on all ranges. Accuracy: Â±0.05% of reading Â±1 digit with line voltage variations of Â±10% from nominal throughout temperature range between +15Â°C and +40Â°C; Â±0.1% of reading Â±1 digit for tem perature range of 0Â°C to +55Â°C. Sample Rate: Greater than 5 samples per second to less than 1 sample per 5 sec onds with display storage between sam ples. HOLD position displays single measurement indefinitely or permits ex ternal initiation of samples by applica tion of +10 volt pulse. Range Selection: WITH 3441A: Manual. WITH 3442A: Manual, automatic, and programmed. Range Change Speed â€” Automatic: achieves accurate reading within 500 ms after new voltage is ap plied. Programmed: changes range with in 40 ms. © Copr. 1949-1998 Hewlett-Packard Co. Possible Digital Error L (Â±1 Count Ambiguity) 4 0 6 0 8 0 (1 OF FULL SCALE) command when the voltmeter is used with the 3442A Automatic Range Selector plug-in unit. The desired range may be selected either by a contact closure or by a transis tor switch, enabling the new volt meter to be used in digital data acquisition systems. ACKNOWLEDGMENTS The design group for the new voltmeter included Donald E. Barkley, Charles W. Near, and project leader David S. Cochran. Paul G. Baird and Peter Kertesz developed the plug-ins and the mechanical de sign was by Tor Larsen. â€”David S. Cochran and Charles W. Near Input Impedance: 10.2 megohms (dc) on all ranges. Input Filter: AC REJECTION: 30 db at 60 cps in creasing at 12 db per octave. RESPONSE TIME: Less than 450 ms to a step input. Polarity: Automatic indication. DC Isolation: Signal pair may be operated up to 400 volts dc from chassis ground. Electrical Readout: 6 columns consisting of 4 digits, polarity, and decimal posi tion; 4-line BCD with l-2-2*-4 weighting. "0" is â€”24 volts and "1" is â€”1 volt; 120 kÂ£i output impedance. Print Command: +25 volt peak pulse at completion of each sample except dur ing automatic range change (including short settling time). 100 ohm source impedance ac-coupled. Power: 115 or 230 volts Â±10%, 50 to 1000 cps; approximately 20 watts. Weight (including 3441A Manual plug-in): 19 Ibs. net; shipping weight 24 Ibs. Price: 3440A Digital Voltmeter (requires plug-in): $1,160.00. 3441A Manual Selector Unit: $40.00. 3442A Automatic Range Selector: $135.00. Prices f.o.b. factory Data subject to change without notice FLAT-RESPONSE DETECTOR (continued from p. 8) tors. Reflection coefficient measure ments and such measurements as coupling and directivity of direc tional couplers can readily be made on a swept-frequency basis and dis played on an oscilloscope for instant analysis of the effects of changes and adjustments. One Model 423A is used to level the sweep oscillator (with its internal amplifier) and a second Model 423A is used to detect the signal to be measured. FREQUENCY (GO Fig. 4. Frequency response of two typical Model 423 A Crystal Detectors. Ordinate represents RF power re quired to produce 100 millwolts output. Detector is a result of HewlettPackard's ability to integrate semi conductor technology and micro wave technique in solving measure ment problems. The crystal diode itself was developed by â€”hpâ€” to meet performance goals unattainable with presently available diodes. A point contact microwave diode alone is an extremely poor match to 50 ohm transmission line. The us ual technique for reducing the VSWR of broadband crystal detec tor mounts is to place shunt and series resistors of about 70 ohms each ahead of the diode. By contrast this problem is solved in the new detector by placing a 50-ohm resistive film on the dielec tric cyclinder which constitutes the case of the diode. Thus, the physical separation between rectifying con tact and shunt resistor is minimized. The 50-ohm film resistor is a good match to coaxial line, and the crys tal diode is essentially a high im pedance shunting the resistor. The RF bypass capacitor across the output is an integral part of the crystal capsule. The bypass capaci tance is purposely kept small (about 10 pf) so that the rise and fall times of the detected envelopes of fast RF pulses would not be unnecessarily degraded. If the RF bypassing is in sufficient in a given situation, an external low pass filter can be used. Another factor involved in the design of a crystal detector for level PULSE RESPONSE ling applications is the output re sistance of the detector. The RC time constant formed by the output resistance and the bypass plus out put cable capacitances limits the response to fast power fluctuations. The output resistance of the Model 423 A is less than 15,000 ohms. The crystal capsule can be readilyreplaced in the field. The capsule includes the 50-ohm film resistor and RF bypass capacitor as well as the crystal diode itself. The critical components are supplied as a unit, the mount playing only a minor role in determining frequency re sponse and VSWR. The Model 4 23 A complements the new line of â€” hp- sweep oscilla The Model 423A is an ideal de tector for displaying the envelopes of fast RF pulses on an oscilloscope. The technique is to shunt the out put of the detector with a resistor (usually 50 to 1000 ohms) to speed up the response to pulses. The ob ject is to make the RC time con stant (shunt resistance times the sum of RF bypass, cable and oscil loscope input capacitances) short compared with the pulse-envelope buildup and decay times. Of course, output voltage is considerably re duced by heavy loading; conse quently, a compromise has to be effected between output voltage and response time. For example, with short cables, an output load resist ance of 50 ohms and a peak pulse INPUT POWER (DBM) Fig. 5. Square-law response of typical Model 423A Detector with and without load. Load indicated is -hp- 11 523 A matched video load. © Copr. 1949-1998 Hewlett-Packard Co. 4 e FREQUENCY (GO FREQUENCY (GO Fig. 6. Frequency response of one Model 423 A when another Model 423 A was used to level the sweeper. Second trace has Model 423 A' s reversed to em phasize response differences between units. This randomly-selected pair is matched within 0.1 db. Vertical scale is 50 mv/div. power of 50 mw, pulse buildup and decay times down to about 5 nano seconds can be measured with oscil loscopes having sensitivities of 50 mv/cm or better. The Model 423A is normally sup plied with negative output polarity and without a video load. Positive SPECIFICATIONS -hpMODEL 423A CRYSTAL DETECTOR Frequency Range: 10 Me to 12.4 Gc. Frequency Response: Within Â±0.5 db from lOMc to 12.4 Gc as read on an -hpModel 416 Ratio Meter or -hp- Model 415 SWR Meter calibrated for square law detectors. SWR: <1.2 from 10 Me to 4.5 Gc. <1.35 from 4.5 to 7 Gc. <1.5 from 7 to 12.4 Gc. Sensitivity: HIGH LEVEL: <0.35 mw produces 100 mv output. LOW LEVEL: >0.4 mv dc/^w cw. Output Impedance: 15 k maximum, shunted by 10 pf. Detector Element: Supplied. Maximum Input: 100 mw. Noise: <200 Â¡i\i pk-pk, with cw power ap plied to produce 100 mv output. Output Polarity: Negative is standard; see Option 03 below. Input Connector: Type N male. Output Connector: BNC female. Dimensions: % in. (2.2 cm) diameter, 2'/2 in. (6.4 cm) long. Weight: Net, 4 oz. (110 gm). Price: $125.00. Option 02: Furnished with -hp- 11523A video load for optimum square law char acteristics, < Â± 0.5 db variation from square law from low level up to 50 mv dc output. (75 k min. load, cw input.) 11523A length, 2ft6 in. (6.5 cm). Add $20.00. Option 03: Positive polarity output. No extra charge. (Replacement crystal: -hp- 00423-801, $75.00.) Prices f.o.b. factory Data subject to change without notice Fig. 7. Measured VSWR of two typical -tip- Model 423A Crystal Detectors. output polarity can be supplied as option 03. If square-law behavior is needed over the widest possible dy namic range, then option 02, which includes a selected video load resis tor, should be specified. ACKNOWLEDGMENT The writer wishes to acknowledge the assistance of Harold E. Hiner, who made many valuable contribu tions to this project. â€”Russell B. Riley APPENDIX Tangential Sensitivity of the Model 423A Detector While not designed for maximum tangential sensitivity (the 50-ohm RF resistor actually absorbs most of the incident power), there may still be applications requiring a wellmatched, broadband detector where a knowledge of the tangential sensi tivity of the new Model 423A De tector would be useful. At low levels without bias the crystal diode gen erates no excess noise. Its noise is simply that of a resistor equal in value to the output resistance of the diode. Thus the signal-to-noise ratio at the output of the video amplifier is given by Signal power available from source 1 \ Noise power available F from source kTBF where y is open-circuit voltage sen sitivity, P is RF power, Rd is crystal diode output resistance, k is Boltzmann's constant, T is absolute tem perature, B is bandwidth of the video amplifier and F is the noise © Copr. 1949-1998 Hewlett-Packard Co. figure (expressed as a ratio) of the video amplifier when connected to a source with resistance equal to Rd. The RF power P1 required for a signal-to-noise ratio of unity is then P1 = Y l/kTBRdF With a typical Model 423A (y = 0.5 mv/^w = 500 volts/watt, Rd = 5000 ohms) and a video amplifier with a bandwidth of 1 megacycle and noise figure of 3 db (F = 2), we have P, = 5(50 ^(1.38xlO-=')(300)(10Â«)(5000)(2) s 2.5 x 10-8 watts = 2.5 x 10-" milliwatts s â€”46 (ibni Tangential sensitivity, which is often taken to be an RF power level 4 db more than that required for a signal-to-noise ratio of unity, is then â€” 42 dbm. A bandwidth of one meg acycle has been assumed for this example. Much weaker signals can be detected with narrow-band am plifiers of the kind used in VSWR and reflectometer measurements. * Since the output impedance of the 423A is relatively low compared with other point-contact silicon diodes, bias is usu ally not required. A NEW COAXIAL CRYSTAL DETECTOR WITH EXTREMELY FLAT FREQUENCY RESPONSE NEW coaxial crystal detector has been developed by the -4ipâ€” labora tories for such applications as level ing microwave sources and meas uring reflection coefficients, as well as other uses where improved stand ing wave ratio and flatness of fre quency response are needed. The state of the art has also been ad vanced with respect to sensitivity for broadband detectors. The frequency range of the new detector is 10 Me to 12.4 Gc (kMc) with less than Â±0.5 db variation in sensitivity over the entire range. The sensitivity is a slowly varying function of frequency with little "fine structure," as shown in Fig. 4. Over narrower bandwidths (for ex ample, over 4 to 8 Gc, or over X-band) the performance is typi cally much better than Â±0.5 db. Thus, any two crystal detectors can serve as a "matched pair" for reflectometer and other similar appli cations. The standing wave ratio is less than 1.5 up to 12.4 Gc and is even less at lower frequencies (see specifications). Fig 1. New -hp- Model 423A Crystal Detector has so constant a frequency response over the 1240:1 range from 10 megacycles to 12.4 gigacycles that any two of the Detectors constitute a well-matched pair. Illustra tion also shows optional video load which is selected for optimum Detector square-law response. The crystal detector is also avail able with an external load resistor (-hp- 1 1 523 A) selected for maxi mum dynamic range with good square-law response (Fig. 5). When highest sensitivity is needed, the de tector is simply operated without the load resistor. Without the load, the sensitivity is nearly three times better than previously available broadband crystal detectors. The detector is also unusual in that a noise output specification has been added to the others normally listed for broadband crystal detec tors. Low noise is important, for example, when a crystal detector is used in a closed-loop system to level the power output of an RF source, since a noisy crystal causes the lev eled RF amplitude to be "noisy!" The â€” hpâ€” production noise test sim- ulates operation of the detector as a leveling device. The noise voltage is approximately proportional to output voltage. CONSTRUCTION The advance in performance rep resented by the Model 423A Crystal (continued inside on p. 6) FREQUENCY (GC) Fig. 3 (a). Reflection coefficient of typi cal Model 423A Detector. Measure ment was made using setup of Fig. 2. Maximum value indicated is 7.5%. FREQUENCY (GC) Fig. 2. Two Model 423 A Crystal Detectors used as a matched pair with an -hp- Model 692A 2-4 Gc Sweeper and an -hp- Model 777D Coaxial Dual Directional Coupler to form a swept reflectometer. © Copr. 1949-1998 Hewlett-Packard Co. Fig. 3(b). Reflection coefficient of broadband detector typical of the state of the art prior to development of Model 423 A. Maximum value indicated exceeds 20%.
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project