70 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASuREMENT, FEBRUARY 1971

UNITS TENS HUNDREDS |THOUSANDSREASOUT XREAOUT READOUT READOUT

PUL SEINPUTFROM SCM TT UNTS TENS MUiRO THOUSANDSTRANSDUCER TRIGGER GATE COUNTER COUNTER GOUNIER COUNTERUNIT _0 _- I) (-10) _ 10) -10)

COUNT SWITCH1 ANTI -r--NBJNCEJ'> R T1 CIRCUJITRESET SWITCH ~~ANTI-

RESET LINE

X IRGUITIFig. 3. Diagram of counter unit.

measurements must be performed over a standard length pathat a standar(d constant speed. With the effects of these variablesminimized, the magnitude of the resulting count reflects therelative roughness of the road surface with good repeatability.

Typically, if three or more test runs are made and the counttotals are averaged, it has been found that the individual counttotals generally vary less than 10 percent from the average.Thus, the equipment appears to perform its intended functionof providing a means of making quantitative comparisons ofrelative road roughness quite satisfactorily.

Since the relative motion of the rear axle of a vehicle is highlyFig. 4. Circuit boards used in counter unit, dependent on the condition of the suspension, it is evident that

for long-term repeatability of measurements it would be neces-sary to have a standlard test path available for periodic chiecks ofequipment performance. With such a check path available, itshould be possible to determine multiplying factors to normalizethe data obtained over an extended period of time. With long-term repeatability of measurements assured, the equipmentwould appear to have potential application in highway surfacestudies where investigations of differences or deterioration inroad surfaces might continue for extended periods of time, asopposed to the relatively short periods of time involved inmaking rumble strip comparisons.

Amplifier Whose Amplification Is the Exponential of anExternal Parameter

Fig. 5. Complete counter unit. RAUL JOSE OTERO

Abstract-A method is described to achieve the current amplifica-The counter unit is housed in a small aluminum cabinet as tion of an amplifier equal to el, x being the angle of rotation of an

shown in Fig. 5. The unit is placed on the seat beside the driver, axis. The systematic error is lower than 1 percent for lxl < 0.5.or other operator, and connected by means of a cable and This device, used for the analog recording of the neutron monitor'connector to the transducer unit in the trunk of the vehicle. counting rate, solves the equation N(po) = N(p) - e.Power for both units is provided by a 12-V storage battery.

PERFORMANCE Manuscript received April 18,1969. was presented at the 19th IEEE

Semana de la Ingenieria 1l1ctrica y Electr6nica, B,ienos Aires, Argentina, Novem-The output of the equipment is a count number, which is ber Il-IS, 1968. under titile Ampliicador de Ganancia Controlada" in the

related to vehicle speed and length of test path in addition to the The auithor may be reached c/o Juan A. del Georgio. Surmiento SoD, BuenosroughnessAof the test path. Thus, for meaningful results, the A Argentina.roughness othtetpt.Tu,frmaigureus,he I IQS Y Instruction Man-ual No. 7. IQSY Secretariat, London. 1964.

SHORT PAPERS 71

INTRODUCTION RL

The usefulness of the analog recording of the neutron monitor I0(NM) counting rate is greatly increased if a correction is madeaccording to N(po) = N(p) el",P where6 ~~~~~~~~~~A0'RN(po) = counting rate reduced to the reference atmospheric Fig. 1. Operational current amplifier.

pressure po; in this case po = 760 mmHg;N(p) = counting rate measured by the NM at atmospheric

pressure p;= barometric coefficient; in this case, it has a value

of 8.8 X 10-3/mmHg;p = -p O<s 4t20 mmHg in Buenos Aires.

Since the sum of the currents fed by the rate meter is pro- f- cportional to N(p), in order to obtain and record N(po) it is A 5necessary to intercalate an amplifier with the amplification Ri = R between A and B

Ro = R between B and CA = MVpo)/N(p) I1 f AP 0O,IRi Ro R?.A=1AR = k Ap - increment that, with opposite signs, is added to Ri

and Ro for Ap s 0. Physically, it is the variation in resistancewhere x = Iu *Ap. of a linear potentiometer actuated by the barometer.

Fig. 2. Adapted operational amplifier showing how the pressure information is

METIIOD introduced.

The amplifier used is an operational one2 of the type shown 1 + x/2in Fig. 1, where Ai 1 -x/2

Ai = is/ii = Ri/Ro. (2) 2 3 n

As shown in Fig. 2, Ri = R + AR; Ro = R- AR. 2+2 + xi + 22 + + 2n-I + (6)Where which is compared to the expression we need:

R = k-Ap (3) x x~~~~~2 x3 aR=k~~~~~~~Ap (3)= -+ -+ +(7)A R + A\R _1 + AR/Rt 1 + x/2 i r=1+1! +2! +3! + +n!+A- ± K-1AKK_1 x2 (4)R - AR 1- AR/R 1 -x/2 Expressions (6) and (7) coincide up to the second-degree term:

the systematic error will thus beand therefore x = 2AR/R,

R-x RRKAp (Ai -A x3 +AR-=2- 2 (5) 3

Developing (4) in MacLaurin series, + 2 1 n! +

-.9

47K 70KI 6I7

100 2N697 ~,136 K

L~~~~~~~~~oi UK> to ox+9V1 251 2 1 471(

100( 10002/ 0

3638 ~~ ~ ~ ~ ~~~~~3

+9v

Fig. 3. Complete amplifier.

2 C. P. Gilbert, The Design end Use of the Electronic Analogue Computers.London: Chapman and Hall, 1964, pp. 11-16.

72 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, FEBRUARY 1971

taking E, as an upper limit for E, are used to automatically select a working element in each group3 4 on the LSI slice. A simple language has been devised to facilitate

I 1 3 1 1 4 X X the use of this library of routines.= 3! x 2 - 4!! 12 6

and the relative error INTRODUCTION

3/ 4 The test set described in this paper has been designed and builtEr = (Ai - A)/A < (x /12 + x /6)/ex. (8) mainly to perform measurements on LSI structures that contain

In our case, the interval of operation for the instrument is redundant circuit elements. Some of its capabilities are dataAp = 4 20 mmHg and x = Mt Ap + 0.176. The error reaches a storage, arithmetic calculation, and logical decision makingmaximum for the upper end and it is Er < 0.05 percent. On from data information obtained from a probed silicon slice. Theaccount of this value, the error in the method may be neglected. set generates masks for the final metalization process in which aFrom (8) we can see that the difference between (6) and (7) discretionary wiring pattern selects the required circuit element.

is less than 1 percent for lxi < 0.5. HARDWARE SYSTEMAMPLIFIER REALIZATION The computer that controls the system is a PDP8/S processor

Fig. 3 shows the circuit diagram for the amplifier. It consists having a 4000 12-bit word memory. Input and output to theof two stages. The first stage carries out the pressure correction; computer is accomplished through a teletype with a pal er tapeits amplification, for Ap = 0, is equal to one. The second stage reader/punch.has a very small passband and amplifies the difference between Most of the test units that are interfaced to the computerthe first-stage output current and a fixed current controlled by are commercially available and of standard design. In certainthe adjustable resistance RD, thereby providing an expanded cases they were modified to allow the units to interface morescale; its amplification is 16.6 times. conveniently with the central processor. These units consistThe analogical output of the barometer has a coefficient of an A/D converter, voltages and current measuring modules,

k = AR/Ap = 258 ohms/mmHg. From (5) programmable power supplies, a crossbar switch, and a multi-prober, see Fig. 1. The A/D converter is a standard Digital

L.J?/Ap = k = R H/2 Equipment unit, operating on the successive approximationand principle and capable of measuring from 0 V to -10 V with

a resolution of 10 mV.R = 2k/,u = (2 X 258)/(8.8 X 10-3) = 58.8 kQ. In order to use this for current and voltage measurements,

two modules using operational amplifiers and feedback resistorsk was selected to have a high value of R in order to improve the were designed and built. The unit can be used to measure voltagetemperature performance of the circuit. ranges from 10 mV to 100 V positive or negative and currentThe passband of the amplifier is 3.10-3 Hz. Thus, only the from 10 gA to 100 mA in either direction. The feedback resistors

slow variations will be recorded. for range selection are connected by reed switches. These areThe measured amplifier drift was 0.1 percent/°C over a driven from a diode matrix that decodes a simple computer

10-30°C range, for the center of the scale. We did not try to word to a more complex arrangement of contact closures. Thisreduce this value because the device operates in the NM room, hardware feature simplifies the software program required forwhich is temperature controlled. selecting measuring ranges.

There are three power supplies, two of which are program-mable while the third requires manual adjustment, but can beturned on or off by computer control. The programmable unitscan be programmed to a constant current ranging from 10,A

Design of a Test Set for Measuring Large-Scale-Integra- in

D g a s tifor Mysteasr Large-Sale-Integra to 100 mA or a constant voltage from 1 V to 100 V positive ortion Systems negative by means of BCD words from the computer. The

interface required for programming the power supplies consists ofRICHARD A. MORRISON AND JOHN A. GUGGENHEIMER I/O selector boards, registers to store the programming data,

and level converters for each line to give the correct interfaceAbstract-The ability to manufacture large-scale-integration voltages to the units. The other supply is a constant-current

(LSI) systems necessitated our building a special type of test set. constant-voltage type and must be programmed manuallyThe set is capable of controlling test equipment to produce mea- from the front panel. It can be turned on and off by an excitingsurements, storing the results, performing arithmetic operations circuit consisting of a gated ring oscillator, which is capacitivelyon the results, and then carrying out logical decisions. coupled to a detector in the power supply. This method permitsThe set contains a PDP8/S computer to which is interfaced a de isolation of the power supply allowing either side of the

teletype unit, an A/D converter, various power supplies, voltage supply to be connected to ground.and current measuring modules, a prober, and a matrix for inter- The crossbar switch forms a 10 X 20 matrix with Kelvinconnecting the test equipment to the prober. All of these interfaced contacts and is used to connect the device under test to theunits are under computer control. various power supplies, measuring systems, loads, and ground.Most of the interfaced units are of standard design, however, A total of 16 vertical columns are connected to the device while

because of the incompatibility of logic levels and coding, a fairly ten horizontal rows and four vertical columns are connectedcomplex DTL interface medium had to be constructed. to the measuring units, Fig. 2. Programming the erossbar

Software associated with this test set consists of programs that requires three computer words. The interface circuitry consistsperform the functions of peripheral control, electronic function of two I/O selection boards, registers, and relay driver units.block testing, and data analysis. The results of the data analysis The multiprobe unit for probing silicon slices is a MIicrotech

model 8120. It was modified to allow computer control of theindexing magnitude and direction in the X and Y planes,

Defence Research Board throulgh its Defence InPdustMriaal RePsePa°rch Program. byt.wstg punching1presented at the 1969 EEMTIC and IM Symposium. of information cards. A special mask generator was added to itThe authors are wjith the R&D Division, Northern Electric Company Ltd., n hswsdsge n ul ntelb.I ossso

Ottawa, Ont., Canada.''an hswsdsgean ul ntelb.Ieoitsfa