IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. IM-25, NO. 4, DECEMBER 1976 533

REFERENCES

[1] G. Schuster, "A high-resolution electrodynamic ac-to-dc power transfer instrument," IEEE Trans. Instrum. Meas., vol. IM-23, pp. 330-333, Dec. 1974.

[2] H. J. Schräder, "Leistungs-Frequenzwandler auf thermischer Grundlage," Ζ. Instrum., vol. 71, pp. 330-334, Dec. 1963.

[3] W. Albach, "Multiplikationsschaltungen zur Wechselstrom-leistungsmessung," Wiss. Abh. der PTB, vol. 21, pt. 1, no. 389, p. 74, 1969.

[4] P. L. Richman, "Electrical measuring systems," U.S. Patent 3 633 166, Jan. 4,1972.

[5] G. Schuster, "A precision stabilized sine-wave source for ac power measurements," IEEE Trans. Instrum. Meas., vol. IM-22, pp. 391-394, Dec. 1973.

Automating Wide-Band AC/DC Transfer Measurements

R A Y M O N D G E R R AND F R E D L. K A T Z M A N N , MEMBER, IEEE

Abstract—A semiautomatic rms ac/dc transfer standard that permits a 30-fold increase in the rate at which transfer measure-ments can be made is described. The precision of measurement is ±50 ppm in the automatic mode and ±10 ppm in the manual mode. A novel contactless overload protection circuit is included which allows waveforms of any crest factor up to the breakdown voltages of the components to be measured. The instrument is suitable for programmable operation making it adaptable to systems opera-tion.

INTRODUCTION

FOR M A N Y Y E A R S t h e a c c e p t e d m e t h o d for calibrating ac sources and voltmeters has been through

the use of vacuum thermocouples as ac /dc t ransfer s t an -dards . T h e y are m o u n t e d in several types of conta iners along with ranging resistors and frequency-compensating elements , all of t h e m deriving from t h e pioneering work of F . L. He rmach of t he Nat iona l Bureau of S t a n d a r d s [l]-[3]. A schematic of a device of this kind is shown in Fig. 1.

Tak ing a Hermach-s ty le thermocouple , Ba l lan t ine 1394A, as typical, the procedure in using them is as follows: T h e interconnect ions are shown in Fig. 2. Well - regulated power lines and a stable ambient tempera ture are required. A reference frequency, usually dc or 1 kHz, is applied and the dc o u t p u t of t he the rmocouple is recorded. T h e re-cording is obtained by reading the indication of a D V M or in t he case of nul l - indicat ing bridges a n d po ten t iomete r s by the position of t h e controls . T h e frequency of in te res t is t hen appl ied and t h e ampl i t ude adjusted unt i l t h e dc o u t p u t of t he thermocouple equals t h a t p roduced by t h e reference. T h e reference signal is again appl ied a n d t h e system checked for drift and consistency. T o cal ibra te a source, t he unkown level is appl ied first, and t h e known reference adjusted to produce t h e identical dc from t h e thermocouple . A skilled opera tor can achieve two to four readings an hour.

T h e slowness of t he process resul ts from t h e care nec-essary in making readings precise within 50 p p m or bet ter .

Manuscript received June 28,1976. The authors are with the Ballantine Laboratories, Inc., Boonton, NJ

07005.

ι ι

TC I

J2

o J Fig. 1. Basic coaxial thermal converter.

V O L T M E T E R

1 3 9 4

T H E R M O C O U P L E -

P R O B E

D E V I C E B E I N G

C A L I B R A T E D

T E R M I N A T I O N

Fig. 2. Interconnections for using coaxial thermal converter.

As balance points are reached, controls are t u rned slowly to avoid generat ing t he rma l E M F ' s or t rans ients . After a large change, t h e sys tem has to be given t ime to reach t he rma l equi l ibr ium. A wrong a t t enua to r bu t t on pushed can resul t in t r ans ien t s requir ing several minu tes t o dis-appear , or worse, can des t roy t h e expensive and fragile thermocouple .

W h e n finished, t he opera tor has a record of accuracy relat ive to a reference frequency. Th i s record m u s t t h e n be corrected for the reference frequency ac/dc error unless dc has been used as t h e reference.

THERMOCOUPLE PROTECTION

T o preven t damage to t h e thermocouple , protect ion schemes are sometimes used. See Fig. 3. T h e most common protection circuits involve peak detectors t ha t monitor the voltage across t he thermocouple . These detectors have serious shortcomings. They place a relay in the signal path.

534 IEEE T R A N S A C T I O N S O N I N S T R U M E N T A T I O N A N D M E A S U R E M E N T , D E C E M B E R 1976

- P E A K D E T E C T O R - - Z E N E R D I O D E S -

O P T I O N A L R A N G I N G D I S C O N N E C T

. R E S I S T O R L

o r f ι w y y ^ = W

1

το R E L A Y D R I V E S

R A N G I N G R E S I S T O R

D I S C O N N E C T '

Fig. 3. Protection schemes.

They shun t the thermocouple . T h e shunt ing impedance is sometimes a function of ampli tude, sometimes a function of the condition of the detector , and always a function of frequency. The detectors also severely limit the crest factor of nonsinusoidal waveforms. If t ru ly accura te and repea-table readings are to be made , a t r ia l r un is first t aken . If this is satisfactory, the overload detector is d isconnected from the thermocouple and the measu remen t s are re-peated. Another common approach is to use Zener diodes. These have the advantage of protect ing wi thout t h e use of relays in the signal pa th . Otherwise they suffer from all the disadvantages of the peak detector approach. T h e best approach is a form of contactless detector tha t operates on the t empera tu re of the heater . Th i s is t he approach used in the ins t rument abou t to be described. T h e r e is no ab -solutely foolproof protection. If a thermocouple should be destroyed, it can be replaced in this i n s t rumen t wi thou t requiring recalibrat ion.

BALLANTINE 1600A

T h e ins t rument described in th is paper is a semiauto-ma ted ac/dc transfer s t anda rd t h a t allows an unski l led operator to make two measu remen t s per minu te , wi th a precision of 50 p p m in au tomat ic mode and 10 p p m in manua l mode with somewhat more t ime. T h e o u t p u t is a dc voltage equal to the rms value of t he ac input . T h e cal-ibration is traceable to N B S and the recertification period is one to three years. Measurements are made in 12 octave ranges covering 250 mV to 1 kV. T h e frequency range using the internal transfer head is 100 MHz. However, provision is made for using external thermal-conver ter probes and with a device such as t h e Bal lant ine 1396A t h e r m a l con-verter, t ransfer measuremen t s can be m a d e u p to 1 GHz. The accuracy is tha t of the external thermal converter, t he precision is t h a t of t he Bal lant ine 1600A, t h a t is, 50 p p m in au to balance, 10 p p m in manua l .

Each measuremen t consists of two dis t inct phases . Phase one, "ac mode , " measures t he rms value of t he un-known and stores its value in memory. P h a s e two, "dc mode , " injects a dc voltage, t h rough t h e same p a t h t h a t equals the rms of t he unknown, t h u s t h e n a m e ac /dc transfer . Figs. 4 and 5 show simplified block d iagrams of t he two phases . In each, t he circuit is a null-seeking am-

E R R O R D E T E C T O R

Fig. 4. AC mode block diagram.

W O R S T C A S E

L O O P G A I N 109 d B

Fig. 5. DC mode block diagram.

plifier. T h e error signal is t h e s u m of t h e dc signal gener-a ted by the the rmocouple and t h e null ing voltage devel-oped across t h e resistor. In ac mode , t h e amplified error signal opera tes an er ror -detec tor compara tor . T h e com-para to r level in t u r n controls t h e direct ion of count of a 16-bit, R/2R, D/A converter whose dc level is appl ied to the summing junction where the resistor and plus terminal of t h e thermocouple o u t p u t join. W h e n a nul l is reached and sustained for 15 s, a transfer relay operates, set t ing up the circuit for phase two, and inhibit ing the counter. Wi th t he counter inhibi ted, i t serves as a memory for t he rms voltage of t he unkown. Th i s s tored voltage is now the ref-erence inpu t for t he new feedback configuration shown in Fig. 5. T h e inpu t is disconnected from the external source and connected to t he in te rna l dc amplifier. W h e n th is l inear system is balanced, a t t h e end of abou t 15 s, t he dc a t t he i npu t is equal t o t he rms of t h e unknown, and is available a t t he o u t p u t jack. No tewor thy in th is a r range-m e n t is t he fact t h a t t h e longest t ime period in which critical ad jus tments are m a d e is 15 s, a n d t h e prob lems associated with long-term stabili ty are bypassed. A switch is provided t h a t allows t h e i n s t r u m e n t to automat ica l ly repeat the measurement every 30 to 45 s. A second ac mode is provided wherein the nulling voltage is derived manually from an internal dc supply and potentiometers. This mode is t he mos t accura te and precise; repeatabi l i ty within 10 p p m is possible.

T h e numbers associated with the ac mode are as follows. T h e error de tec tor window is ±0 .7 V. T h e gain from the compara to r to t h e chopper i n p u t is 126 d B (2 X 10 6 ) , so t h a t t h e dead zone a t t he chopper i n p u t is 366 nV. T h e

GERR AND KATZMAN: WIDE-BAND AC/DC TRANSFER MEASUREMENTS 535

output from the thermocouple is nominally 7 mV full scale and allowing for 25-percent overload and the square- law characterist ic we get 11-mV m a x i m u m signal so t h a t t he resolution is ± 3 1 p p m . T h e o u t p u t of t he D / A converter and scaling resistor are adjusted to produce 180 nV per count so t ha t the dead zone or backlash corresponds to ± 2 counts out of 2 1 6 or ± 3 0 p p m . T h e combined uncer ta in ty is therefore ± 4 3 p p m .

The numbers for the dc mode are not as straightforward. Looking a t Fig. 5 t he re is t h e same 126-dB gain as in t h e ac mode, an addi t ional 56 d B (on t h e low-voltage range) to obtain the o u t p u t voltage V, and negative gain from V to small u, t he r e tu rn voltage. Th i s negat ive gain varies with range and with ampl i t ude within t h e range as can be seen from the following simplified derivat ion. T h e dc ou tpu t of the thermocouple υ is propor t ional to power dissipated in t he thermocouple heater . In t e rms of t h e ou tpu t voltage V

V = K ( R R V + R T ) * R T

where RR is t he series ranging resis tance and R T t h e ther -mocouple hea ter resistance. T h e gain t h e n is

dv _ 2KRt

dV~(RR + RT)* '

Within any one range the gain varies by 2:1. For t h e low-voltage range, the worst case is t he high end of t h e 32-V range which t rans la tes in to a gain of —73 d B . For t he low end of the 32-V range, t he gain would be ~~67 d B . On t h e other hand , in the 0.5-V range m a x i m u m , t he gain is —31 dB . Taking the worst case, for example , t h e smallest loop gain is 109 d B (281 Χ 10 3 ) . Referr ing again to Fig. 5,

υ _ 280 000

Ε 280 001 , therefore, υ equals Ε wi thin 3.6 p p m .

Adding this to the uncer ta in ty of t h e ac mode we obta in 47 p p m . T h e high-voltage range has similar similar fig-ures.

In the manua l mode t h e dc bucking voltage for t h e ac mode is obta ined from a s table dc supply and high-reso-lution potentiometers. I t is possible to obtain a dc reference as good as 5 p p m and an overall t ransfer be t te r t h a n 10 p p m .

ATTENUATORS

T h e previous discussion covered t h e basic sys tem pa-rameters involved in handl ing t h e dc o u t p u t of t h e ther -mocouple. T h e hea r t of any t ransfer s t andard , however, is the a t tenuator- thermocouple assembly where the actual conversion takes place. T h e geometry of th is assembly largely de te rmines t he frequency response. T h e rigidity of its construct ion and t h e qual i ty of t h e componen t s de-te rmine its s tabi l i ty wi th t ime , and consistency from un i t to uni t . T h e control of t e m p e r a t u r e effect is also a m a t t e r of t h e mechanical construct ion.

T h e r e are two separa te a t t enua to r assemblies. T h e low-voltage a t t enua to r covers t h e range from 250 m V to

Fig. 6. Low-voltage attenuator.

Fig. 7. High-voltage attenuator.

32 V in 7 octave steps and is housed in a massive clamshell a luminum casting shown in Fig. 6. Each ranging resistor is conta ined within a coaxial t ube . T h e switching is done wi th reed relays which ma in ta in t h e coaxial s t ruc ture of t he signal p a t h and, since they are opera ted by closing a contac t to ground, lend themselves to p rogrammed oper-at ion. T h e inpu ts and ou tpu t s are th rough rigid bus bars . All resistors are Corning t in-oxide, V 2 -W type with t em-pe ra tu re coefficients of ±100 p p m . Th i s construct ion satisfies all t he r equ i rements for s tabi l i ty and reproduc-ibility. I t has t he d isadvantage of placing a fairly large capacity, about 20 p F , across the thermocouple. Since the input is not a terminated structure, this capacity resonates a t about 97 M H z with t he inductance of the lead from the Tee t h a t feeds t he input . T h e four lowest voltage ranges, t h a t are also widest in bandwidth , require adjustments to control t he frequency and Q of th is resonance. All com-ponen ts , wi th t he exception of t h e relay coils, are stock commercial i tems. T h e r m a l E M F ' s are handled by using identical mater ia ls as far as possible, by balancing series junct ions where necessary, and by main ta in ing uniform t empera tu r e s t h roughou t t h e s t ruc tures .

T h e high-voltage a t t enua to r shown in Fig. 7 covers t he range from 32 to 1000 V. T h e major factor in th is compo-nen t is the ability to handle 1 kV. This is accomplished by using resistor s tr ings to d i s t r ibu te t he power and voltage gradients. Since real power can be dissipated, the resistors are meta l film with 25-ppm tempera tu re coefficients. T h e large resis tance values and s t ray capacit ies produce a falling-frequency character is t ic . Th i s is compensa ted by capacitively bypassing a port ion of the ranging resistance. T h e capacity is made of a brass tube t ha t fits over a portion of t h e ranging resistor and is centered a round it by layers of mylar foil. T h e tube is connected to the high-voltage end of t h e str ing. Adjus tment is by posit ioning the t u b e along

536 IEEE T R A N S A C T I O N S ON I N S T R U M E N T A T I O N A N D M E A S U R E M E N T , D E C E M B E R 1976

L .

RANGE

VOLT8

Ο - - - y

H , G A H T ^ T

SECTION - r 1 1

1 1 ι 1

> LOW VOLT A T T E N . SECTION

I OVERLOAO ^ f X A

V DC O U T P U T

M E T E R

DC AMPL.1 S F I L T E R CIRCUIT

L _ .

- MANUAL BALANCE V

COARSE FINE

Q Q

_ d > à .

I

2 * L A N C E ^ ~ ^ T R A N S F E R |

TRANSFER-RESET]

F A S T NORMAL

, A U T 0 . TRANSFER

CIRCUIT

- J

U TRANSF

I R F L Δ Y

I

Fig. 8. Simplified schematic of Ballantine 1600A system.

the length of the resistor. After ad jus tmen t t h e tubes are cemented in place.

T h e thermocouple itself is housed in a massive a lumi-n u m casting t h a t is bolted to the low-voltage a t t enua to r housing. T h e thermocouple is of t he U H F type and is thermal ly isolated by a cushion of u re thane foam. T h e thermal t ime cons tant of the i n s t rumen t is 4 h a n d since a reading takes 30 s, t empe ra tu r e s tabi l i ty is no t a serious problem. Still, for best precision the ins t rument should be allowed to stabilize and several readings should be t aken a t a part icular set t ing to bring the thermocouple a n d re-sistors to the rmal equi l ibr ium.

Associated with the thermocouple is an infrared detector for the overload protect ion circuit. T h e detector is a flat silicon photovoltaic cell mounted against the thermocouple glass and aligned with the fi lament. Between the ther -mocouple and the photocell there is an electrostatic shield of 1-mil thick copper with 80-percent transmissivity. T h e the rmal mass of th is shield is too small to affect t h e effi-ciency of the detector sensibly, while it is adequa t e as an R F shield. T h e system opera tes a t dull red hea t or below. T h e ou tpu t of the photovol taic cell is amplified by a two-stage high-gain dc amplifier followed by a comparator. T h e comparator triggers an SCR t h a t opens a power-switching transistor , located in the ground re tu rn of the reed-relay drive circuits, opening all the relays in the signal path . T h e relays will not reset unless t he manua l reset b u t t o n is pushed. No overload protect ion system is 100-percent ef-fective since relays take a finite t ime to opera te . Th i s sys-tem has been tes ted to 100 t imes overload. T h e relays opened in 2.5 ms, though they are specified a t 5 ms. There are two advantages to this overload protection system tha t cannot be overemphasized. One, because the system is nonloading, it has no effect on accuracy a t any frequency or level. Two, because it does not depend on a voltage level,

the re is no restr ict ion a t all on crest factor (barring breakdown voltages of components ) or wave form.

T h e overall system organizat ion is shown in Fig. 8. T h e thermocouple a t t enua to r s , range switch, overload pro-tect ion circuit, and relay power supply are contained in a single subassembly. Th i s t ransfer assembly can be used and calibrated independently of the rest of the system. T h e remaining circuitry, with t he exception of large compo-nen t s such as t ransformers , is organized in a group of plug-in printed circuit boards. T h e mode switch S i a front panel , 3-pole switch, sets u p the source for the dc memory from the reference supply in manua l mode or " c o u n t s " o u t p u t for au tomat ic mode; also, it g rounds t he error de-tector for m a n u a l opera t ion or connects it to t he dc am-plifier for au tomat ic operation; and it t u rns control of the ac /dc shift of the ma in t ransfer relay from the manua l t ransfer reset S 4 to a 15-s delay generator t h a t is pa r t of the low voltage amplifier assembly. T h e dc/ac shift is al-ways possible through the transfer reset bu t ton . T h e main t ransfer relay, a 12-pole double- throw relay, sets u p the loop-stabilizing networks and the mete r circuit, and switches the inpu t from external ac to in ternal dc, as well as performing several o ther secondary control functions. Control of the ac loop is th rough a set of gates on the clock oscillator board. T h e error detec tor de te rmines whether up-enabl ing or down-enabl ing signals are sent to the counter . At null , t h e counter is inhibi ted. T h e up- and down-enabl ing signals are also sent to a t rans is tor reset switch on the low-voltage amplifier assembly, and the 15-s delay generator t h a t controls the automat ic ac/dc transfer is r e tu rned to 0 each t ime a count ing sequence occurs. In the dc mode, the delay generator is locked to 0 through the main t ransfer relay. In "au torecyc le" operat ion this dc lockout of the transfer relay is overridden by timing signals from a special source so t h a t the cycle repea ts every 30 to

GERR AND KATZMAN: WIDE-BAND AC/DC TRANSFER MEASUREMENTS 537

Fig. 9. Overall view of Ballantine 1600A.

A C S I G N A L Ί

S O U R C E

T H E R M O — C O N V E R T E R

I 3 9 4 A /

X 3 9 6 A

TUT

8 9 - I O I 2 6 - I F ( 3 - P I N C O N N E C T O R ) ( L E N G T H 3 ' F T )

O P T I O N A L D V M

C O M M A N D P R O G R A M I N G

C A B L E

0 =

D V M 3 6 0 0 A

8 9 - I O l 2 e - I F ( L E N G T H 3 ' F T )

8 9 - I 0 I 2 9 - I F ( L E N G T H 3 ' F T )

Jl J 2

- 8 9 - 1 0 2 5 6 - I F

Fig. 10. Interconnections for external thermal converter.

J I 6 J I 5

1 6 0 0 A

45 s. There is also a N O R M A L - F A S T switch t h a t reduces the 15-s delay to 8 s. This delay is provided mainly to allow operation with relatively unstable input signals. An overall view of t he i n s t rumen t is shown in Fig. 9.

Finally, t he ac /dc t ransfer s t a n d a r d Model 1600A does no t have to be used wi th i ts own a t t e n u a t o r a n d t he rmo-couple. T h e thermocouple ou tpu t s are b rough t th rough a jack in the rear panel . Any 0-11-mV dc signal can be in-jected and the main frame of t h e i n s t rumen t will ac t as a null-balancing indicator . In par t icular , an external ther -mocouple can be connected as shown in Fig. 10. T h e au-tomatic or manua l modes function in the normal way. T h e range switch is set to t h e app rop r i a t e posit ion. T h e over-load protect ion feature is inoperat ive. W h e n t h e ac /dc transfer switch is made , t he ac signal source is replaced manual ly by t h e o u t p u t of J l a n d J 2 . T h e dc balance

proceeds in t h e normal way. W i t h th is a r rangement , t he frequency range is ex tended to 1 GHz. T h e accuracies are those of the converter plus t he uncer ta in ty of the transfer as descr ibed previously.

REFERENCES

[1] J. J. Duckworth, N. Schnog, and A. B. Mueller, "System providing a d.c. voltage equal to the r.m.s. value of an unkown a.c. voltage," U.S. Patent 3 518 525, June 30,1970.

[2] F. L. Hermàch, "Thermal converters as ac-dc transfer standards for current and voltage measurements at audio frequencies," J. Res. Nat. Bur. Stand., vol. 48, no. 2, pp. 121-138, Feb. 1952.

[3] F. L. Hermach, "Ac-dc transfer instruments for current and voltage measurements," IRE Trans. Instrum., vol. 1-7, pp. 235-240, Dec. 1958.

[4] E. S. Williams, "Practical aspects of the use of ac-dc transfer in-struments," Nat. Bur. Stand. Tech. Note 257.