The present invention relates to measurement of insulation resistance, and other related properties of a specimen. The method and apparatus claimed is a refinement and improvement of apparatus and method disclosed in U.S. Patent Cooperation Treaty patent application Serial No. 81/01647, filed December 14,

1981, the invention of Peter H. Reynolds. Using such improvements, it is able to achieve high accuracy over a wider range of measured parameters.

Background of the Invention

The aforesaid Reynolds application made reference to the technical analysis of insulation testing entitled "Insulation Testing by D-C Methods" made in 1958 by E. B. Curdts which he revised and reprinted in 1964 in the Biddle Technical Publication 22T1. In that publication, Curdts showed that when d.c. voltage is applied, the current existing in the insulation of a capacitive specimen is always made up of three components, to wit: geometric capacitance current, !„; absorption current, i _a ? and conduction current, i _c • These currents are more fully explained in the referenced patent application. Conventional insulation testers were capable of only reading the algebraic sum of these currents such that I _τoτ = i _g + i _a + i _c .

In 22T1, E. B . Curdts described a method of calculating i _c from irr> _T « The derived equation is

OMg

<ilO>- ^x 3.16 χ« =

*1 ⁺ ilO ~ ²ⁱ 3.16 where i- _] _, 13.15 ^a nd iτ_g ^are three different values of Ir>oτ ^m easured at different times, based on a constant unit of time multiplied by their subscripts, i.e., 1, 3.16 and 10 minutes. The equation assumed that the unit time chosen was long enough such that !„ ■ the geometric capacitance charging current, has become negligible. In application of this equation, however, some limitations may be encountered of which the following two case examples are illustrative.

In case 1, the time unit was chosen too long for a given specimen. This is possible since nothing is known about the specimen prior to test. Under these conditions Irr> ₀ ^at ' 3.16 and 10 minute time units may be primarily composed of conduction current i _c which remains constant with time. The net result is that i- _j _ approximates 13.1 ^7lιrL ^- ^c ^ ¹ approximates i_ _Q' In placing these readings into the equation the denominator is a very small number having been derived by the subtraction of rather large numbers. As a result small errors in raw data can cause large variances in the calculated resultant

In case 2 the conduction current is, in fact, large relative to the other current components due to poor insulation characteristics. Again, the denominator will be small, making small error terms in the raw data dominate the calculated result, i_ •

Nature of the Present Invention As stated earlier, the total current in a c apacitive sample is made up of three components such that ^I TOT ^{= i} g ^{+ i} a ^{+ 1} c as represented in the schematic circuit diagram of

Fig. 1. The above equation can be approximated in the time domain by substitution for each current term, a formula representative of actual characteristics of each current type assuming a positive potential is applied to the specimen through a series resistance at time zero (t=0) by the following equation: I _TQT ( _{/ m} T _\ ) = _Es e-t/ ^' C Rs +, _E_ _s (,1.-e-t/C Rs) C__D T-n +, X. _Q s where Eg = the applied source voltage

Rg = total series resistance

C = the zero-frequency capacitance of the specimen t = time (seconds) = large time units (i.e. minutes) T>>4t

D = a proportionality factor on a per unit basis of applied voltage and the capacitance of the specimen. This depends on the type of insulation and its condition and temperature. n = a constant

It was further assumed that the series resistance was small in relation to the true insulation resistance-

If the analysis is restricted to many time constants (one time constant = C R _s ) , after t=0, then

I(t) can be further reduced to:

I _T0T (T) = 0 + Eg CDT ^_n + i _c = AT ^_n + i _c where A = E _S CD and T>>4t

At an arbitrary time when T=X

I _T0T (T=X) = AX ^"n + i _c (1)

Likewise at T=Y I _T0T (T=Y) = AY ^'n + i _c (2)

Equations (1) and (2) can be simultaneously solved for i,

_± _ Y ^"n (I _τoτ (T-X) )-X ^"n (I _τoτ (T=Y) ) (3) _____

Therefore if the value of n were known, and the value of j_T0T were measured at times T=X and T=Y, the value of i _c could be determined. To resolve n, ^s i<_f ^na l m y be differentiated to obtain l<τ _j0 τ ' such that

I ' = ^{d I} TOT ^(T) " _(A - ⁿ ₊ i )« -nAT ^{"(n+1) O} dt dt ^C

At the previously chosen time of T=X, the value of I.τ _j g _T '

±sx

I _T0T '(T=X)= -nAX ⁽ⁿ⁺¹⁾ (4)

While at time (T=Z)

I _τoτ ' (T=Z)= -nAZ ^"(n+1) (5)

Since the magnitude of Irrt _Q rr. ' (T) decreases with time, an arbitrary radio R may be defined such that.-

_R _ ⁽ = ^{) *} T ' ( Ψz-- ^" )

TOT Note. R<1.0

By substitute of equations (4) and (5)

R= (Z/X) ^_(n+1) Solving for n: n=(-logR/log(Z/X) )-1.0 (6) As a demonstration of how an instrument can be constructed by application of the derivation above,

OM

"

consider one set of parameters herein chosen to reflect one set of practical value which may be implemented electronically. These values do not define the only practical values.

Let X= 1 minute

Y= 10 minutes

R= 0.5

Z= a measured parameter. If one electronically differentiates the function

-TOT (T) and allows for a constant gain K to be produced by the differentiator, it can be mathematically shown that equation (6) remains unchanged. Let the value of I _TQ r ' (T) be stored by electronic means at one minute (T=X) and by using ^' a precision electronic divider, the stored value will be divided in half (R=0.5). Furthermore, let the instantaneous value of I _{T T} ' (T) be compared by electronic means to the value of 0.5 ITO ' (T=l minute) and record the time in minutes, via electronic means, until said signals are equal. (Solving for T=Z) . Once Z has been determined the value, of n can be accurately determined by: n= (-log 0.5/logZ)-1.0 and consequently i _c can also be determined accurately from _{i =} ι _τoτ (τ=ιo)-ιo ,- (,ι _T0T (τ=D)

Drawings Illustrating the Invention For a better understanding of the present invention, reference is made to the accompanying drawings in which:

Fig. 1 is the equivalent circuit of the best circuit illustrating the currents involved; and

Fig. 2 is a block diagram of a preferred system of the present invention.

5

Specific Decription of the Invention

Fig. 1 has been discussed above and in the parent

United States Patent Cooperation Treaty application.

Apparatus to perform the measurement in the manner _n described is shown in Fig. 2.

The microprocessor 13 is the central element of the instrument. Its functions include sequencing of the test, switching signal paths, controlling the elapsed timer clock, gathering data, performing all _C . necessary mathematical manipulations, making decisions relative to the data obtained, converting the data gathered to readable English format and controlling both an output device (printer) and input device (keyboard) . Together with the microprocessor 13, the _n memory 15 serves to store the list of actions and sequence of operations necessary to perform the test. Collectively this set of instructions is referred to herein as the instrument program. Another section of memory -15 acts as a storage area where the _. microprocessor 13 stores data during the test sequence for future use.

The basic test circuit is comprised of d.c. supply 1, a sense resistor (Rs) 2, and the Specimen 5 under test connected to the instrument via connections 4 and 0 6. The d.c. supply 1 is programmed by the microprocessor to the desired test voltage. The d.c. supply can also be switched in or out of the test

circuit by the microprocessor. When the d.c. supply 1 is programmed to the desired test voltage and attached to the load at the appropriate time, current flows to the specimen 5 through the sense resistor 2. It is the magnitude of this current flow with time which provides the data to solve for true insulation resistance, the object of the apparatus. To measure the value of this current, I o ' ^{a v} °ltage sense and level shift amplifier 3 is attached across the sense resistor. This amplifier has a high impedance input so that the effect of loading upon the sense resistor 2 is negligable with regard to circuit operation. By ohms law the voltage across the resistor is inversely proportional to the current flowing through the sense resistor 2. The voltage sense portion of device 3 measures this voltage and amplifies it with a gain which is set by the microprocessor 13 to provide the widest dynamic range and best resolution possible for the specimen being tested. Since neither side of the sense resistor 2 is at ground potential, the resultant amplified signal generated by the voltage sense portion of 3 is also not ground referenced. In fact, the signal may be at a potential of several thousands volts removed from ground. For operator safety considerations, and ease of construction of the rest of the instrument, a level shift is incorporated to move the differentially generated Iτoτ ^S;L 9 ^na l ^{to a} ground referenced I _T o ^w hi ^cn i ^{s the} second function of the voltage sense and level shift amplifier 3.

The algorithm based on the foregoing mathematical analysis and employed in the embodiment requires that the solution to the equations is based upon the value

of I o ' ( ^ttιe derivative of I _T QT^ • ^For this reason the signal path exiting amplifier 3 is split in two. One path of the Iτoτ ^s i-9 ^na - is f ^e< 3- into an analog to digital converter (A/D) 7 which is under microprocessor 13 control and can convert the magnitude of I o ^{to a} digital form (numerical value) for use within the microprocessor. The other path of the ITO ^s i9 ^na l is fed to a differentiator (d/dt) 8 which produces the desired differentiated signal I OT' * ^Tie 9 ^aLn °£ ^tΛe differentiator 8 described in the algorithm as the quantity K is controlled by the microprocessor 13 to ensure that the differentiator 8 has ample dynamic range and resolution to ensure high accuracy results. The I OT' signal path feeds both the sample/hold memory 9 and one side of the comparator 11. The algorithm describes the need to measure the elapsed time it takes the ITOT' siζf ^{1 3} -! magnitude to drop in magnitude by half. The sample/hold memory 9 is used to store, by analog means, the initial value of ITOT' " ^r ^ ^ιe t ³ ^" ² °f ^t * ^ιe sample/hold memory 9 is controlled by the microprocessor 13. When in the sample mode, the output of the sample/hold memory simply follows or tracks the input magnitude. When the microprocessor 13 switches the sample/hold memory 9 into the hold mode, the output value is held constant so that it no longer tracks the input signal. This is accomplished by trapping voltage on a capacitor by using a very high impedance buffer and a switch. The output of the sample/hold memory 9 retains or stores the magnitude of what the I _τ θτ' signal was at the instant in time when the sample/hold memory 9 was

instructed- to hold. The precision voltage divider 10 establishes an output of 0.5 times the input value and places this signal into the second side of comparator 11 for comparison with the instanteous value of 5 ^I τoτ' (stored). The time it takes for the I O ' signal to drop in magnitude by 50% as described in the algorithm is measured by an elapsed time counter 12. The microprocessor 13 gathers such elapsed time data from the elapsed time counter 12 via 10 signal path 16.

The microprocessor 13, and the associated memory 15 are programmed to compute n = (-log 0.5/Log T) - 1.0 and from this

¹⁵ i. = I ^~ 0 ^~ ( ^κ T=10 ' - lθ ^"n (l ^x T0T(T=D ')' c ι-ιo ^_n

This value may then be displayed on the display of display and controls 14, either as an analog or digital

₂₀ display or directly printed by a printer or any other convenient method of indicating the result. Display and controls 14 also incorporates controls to permit the operator to control the ranges and other settings of the apparatus. Since the value of i _c and Eg are

,._. both known, the microprocessor 13 can be programmed to calculate the true insulation resistance value. This value can be displayed on display 14 in place of or in addition to the value of i _c .

Since the apparatus can measure and store values

-_ of I OT' i ¹ - -- ^{s also} possible ^' to measure the value of apparent insulation resistance, which is the "insulation resistance" measured by conventional

megohmmeters such as the "Megger"® range of insulation testers made by the James G. Biddle Co. of Blue Bell, PA. Thus, it is possible for the operator to determine both the true value of insulation resistance and the apparent value as measured by the previous state of the art. This capability is useful to enable comparisons to be made with historical data.

It is also possible, when using this apparent insulation resistance mode, to pre-program the d.c. supply and to perform time dependent and step voltage or ramp voltage tests and to calculate polarization index as described in the previously references book 22T1. -

It will be understood by those skilled in the art that conventional data processing equipment is used in conventional ways in the practice of the present invention. The system of the invention is capable of many modifications within the scope of the claims and the method may be used with other systems.