DD214460A1 - METHOD OF AVERAGE VALUE AND CURVE SYMMETRY MEASUREMENT - Google Patents

Abstract

The method for the digital measurement of the quaternary and arithmetic mean and the curve symmetry of periodic alternating signals of any waveform, even with DC component is in the AC technology for higher frequencies with arbitrarily high resolution with reasonable measurement and computing time for real-time operation applicable. The invention is essentially based on several parallel-running level-controlled Zeitmessvorgaenge, preferably by counting high-precision reference signals and subsequent offsetting according to special algorithms.

Description

Title of the invention

Mean value method and curve symmetry measurement

The invention relates to the further development of the method according to PS.208 865 for digital

Measurement of the quadratic and the arithmetic mean value as well as the curve symmetry of periodic alternating signals of arbitrary waveform with DC component * The main field of application is the automatic electronic measuring technique, in particular the automatic function test in printed circuit board production «

Characteristic of the known technical readings

For a long time (W · Witt, D. Fränkel »Technisches Messen, atm, 1976, No. 11, page 341 ff) Bind method for the digital determination of the arithmetic and / or root mean square (rms value) of periodic time functions known in which the amplitude Time function in equidistant temporal steps Sample values are taken from which the arithmetic and / or root mean square value is calculated according to the respective definition formula after analog-to-digital conversion and digital intermediate storage in a fast digital computer or the linear or squared memory values to a malt value of the integral of linear or squared time function over a period T be further processed, for example by means of the known Simpson rule or the trapezoidal rule.

Furthermore, methods for the exclusive determination of the root mean square value are known (DE-OS 28 06 695), in which a period of the amplitude-time function is divided in 3J equal time intervals At = T / U, the amplitude-time function at the times η · At - η · Τ / Ν sample values f (ηΤ / Ν) are taken and the effective value EW by the algorithm

(EW) ² _β jyf ² (} i)

is determined

There are also methods for the exclusive determination of the root mean square value known (DD-WP 140 923, 146 659> 151 510 »151 511) f in which the signal to be measured first a squared analog frequency conversion or analog pulse number conversion is performed with subsequent radizierender pulse count and the output of the effective value takes place numerically.

Methods for exclusively determining the arithmetic mean value have also been known for a long time _$, in which the integral is formed over several periods of the signal to be measured in a first step with the aid of an integrating amplifier, and in a second step the storage capacitor is discharged with constant current, at the same time the discharge time is measured by a digital counting method _e The most well-known representative of this group is the so-called two-flank * "or dual-slope converter (R ₉ Gößler: Fast AD» converter _s Elektronik Ί975 »Heft 11, Se 86) * With these Although, as a rule, only the interfering mains AC voltage which is superimposed on the useful signal is eliminated, the application of the known method to signals of any frequency is possible in principle.

A decisive disadvantage of the first-mentioned methods is that their technical implementation requires a high-speed, high-resolution analog-to-digital converter which must have an ultra-fast sample and hold circuit if the arithmetic and / or quad * A further disadvantage is that many time-consuming arithmetic operations are required to determine the approximated integral value, so that the evaluation can basically only be carried out after buffering all the sample values and one requires a lot of time «A time-constant measurement (real-time processing) with a change in the measured quantity is therefore not possible

The decisive disadvantage of all other methods is that they either allow only the calculation of the root mean square value or only the determination of the arithmetic mean value, so that write-in complete Be ** the signal function are generally apply two different measurement methods _e where the last liable

mentioned two-flank method for determining the arithmetic mean of Hachteil that the technical realization of the signal-controlled Aufintegrationezeit requires a lot of effort and the application remains limited due to the finite speed of currently available operational amplifier in the KHz area ·

The invention according to the main patent pursues the aim of avoiding the disadvantages of the known methods for digital RMS measurement with AD conversion, especially in strongly deviating from the sinusoidal signals while allowing high resolution accuracy at higher signal frequency and short measurement and computing time · In the main patent has already been proposed a method that allows in real-time operation digital RMS measurement of periodic alternating signals of arbitrary Abszissensymmetrischer waveform, even with DC for higher fundamental frequencies, with a super fast sample and hold circuit and complex AIMJmsetzer be avoided · The measurement is essentially attributed to a level-controlled timing ,

The invention according to the main patent provides that the positive peak value A, the negative peak value B, the period T and the highest representative spectral frequency component of the signal fmax are digitally determined and stored according to known measuring methods, that a number of 2 (Ν-1 ) +1, where N έ 0.75.T.fmax is selected, different amplitude levels which result as follows:

HI! Level ^Ώ _ A + B

O A 2 B _{+ A} - B ^p n + A B A - B +

positive levels ρ '= +, Wu η = 1., 2 .. · Ν-1 negative levels p ^ = A- ^ J - A _ ^ _ B ^ n "_H ^₂

are compared with the signal instantaneous value during one or more periods, wherein the 'time intervals t _s t during the signal instantaneous value within a period are more positive than the respective positive P or »negative than. the respective negative level P "is digitally measured and stored and that the effective value EVY of the total function by means of known fiechenverfahren according to the following algorithm

(EW) ² = / -A * "^ ² 8 = 1 n = 1

is formed.

The disadvantage of the method proposed in the main patent is that it is limited to waveforms that may have any curve, but must be abscissa-symmetric. "

For this purpose, a further embodiment of the method according to the main patent has already been proposed in which in a first series of measuring steps the determination of the effective value (EWW) of the pure alternating signals freed from direct component by capacitive separation takes place in such a way that depending on the positive peak value A "and negative Peak * value Έ a number N ^/ positive level £ * and a number M ^{7 of} negative levels p ~, which are as follows

η = 0, 1, ···, N-I

is chosen, the RMS value through the algorithm

(EWW)

0.5

N-1

M - 1

ο, 5

T -1.

is formed, in a second series of measuring steps, the measurement of the DC signal is carried out according to the main patent, wherein in the case of a negative peak B i = only a number of IT positive level levels, which result as follows

n a O, 1,

, N-1

and in the case of a negative peak value B <0, a number Έ of positive levels p _n and a number M of negative levels p ", which result as follows

n β 0, 1,. ♦., N-1 dl β 1, 2,. · ·, M

and the RMS value of the overall signal depending on the sign of the negative peak B by one of the two algorithms

(EW) ² =

0.5 ψ-

For

IN THE

(EW) ² = £

0.5 ^

0.5 τρ

for B <0

is formed and in a final evaluation step the arithmetic mean AM by the algorithm

(AM) ² = (EW) ² - (ERA /) ²

is formed «

In order to determine suitable parameters N and M for the number of levels, the already proposed method provides that an auxiliary result of the pure change effective value (BWIy) is only taken into account with every second subpage having even indices and half the values of the characteristics H / 2 and M / 2 j is the relative deviation of his result (P) sum with all times determined (EWW)

EWW-

ν * SBvW '

is formed _s , whereby the originally selected characteristic Ef, M is increased »if this relative deviation exceeds a given residual error«

A disadvantage of the last proposed method embodiment is that the formation of the arithmetic mean of the determination of the RMS squares of the signal function and the DC component freed signal function · requires, so that at low frequencies the decay of transient must be awaited. Another disadvantage is that the method does not allow any statement regarding the symmetry properties of the signal function.

Object of the invention

The invention has the aim of avoiding the disadvantages of the known methods for digital detection, in particular of the arithmetic mean value.

Presentation of the essence

Based on the above object, the object is to further develop the previously proposed methods, so that in digital mode the digital arithmetic and quadratic average and the symmetry properties of periodic alternating signals of any waveform with DC component without separation of the DC component is enabled and a ultrafast Sample and hold operation and a complex conventional analog-to-digital conversion can be avoided, the measurement is essentially attributed to a level-controlled timing.

Erfindungsgeraäß the object is achieved in a further embodiment of the main patent in that according to the determined in the first measuring step positive peak X ~ and negative peak B * two levels groups as follows

«E + (X-B) | f |" η = 0, 1,, .. N ~ 1

= τ - (χ-π) | f§ "η = ο, ι, ... Ν-1

are formed, wherein in a Polge further or in a single parallel measuring step these levels associated time intervals t "_, t .." are determined such that for the first level group P ^ the time periods are selected, for which the alternating signal is more positive than a level P ^ _n and, for the second level group, the time periods for which the V / echo signal is more negative than a level P _2n or vice versa and in an evaluation step the following quantities are determined according to the following algorithms:

Arithmetic mean

"SIT

ΙΓ-1

N-1

n = 1

"IT

AM =

+ ΔΜ.

Asymmetry

N-1

OMJ) / 2

rp

n = 1

RMS value of the pure exchange variable

"10

Auafubrurig abeiapiel_

The invention will be explained below by way of example.

As a periodic signal function, as shown in the figure, a series of positive triangular pulses is selected. The digital determination of the arithmetic and quadratic mean value is possible with the currently known technical means for this failed function type only in the kHz range. It has been determined by known methods

, T β 10

τ = 4 ν

(f ₀ = 100 kHz

In the first attempt U = 16 is chosen.

Then follows according to the invention for the two level groups for

the levels in volts

P _1n = 4 (Tn / 16 "= | fn

P _2n = 4 - 4

η = 0, 1, 2 ... 15

d. H.

^P 10 = °

Ρ- Λ 11 ""

P ₁₂ . = 2 = 1.4142

P ₂₀ - 4

* 4 - 2 = 2.5858

^r 1 15

15 = 3.8730

P _{2 15} = 4 -

⁼ » ¹²⁷⁰

The waveform is now compared to these levels in two serial or parallel processes. In a first process, timing operations were canceled as soon as the signal level became more positive than a level of group P ₁ and in a second process as soon as the signal level became more negative than level Group P ₂ is ·

The results related to the signal period in both processes may be as follows

1_ Group 2

t _t0 / T = o ₅ 5 t _1rl / T = 0.375 \ o ^{/ T. β1} ^ n ^{/ T}

t ₁₂ / T = 0.3232 "t ₁₃ / T = 0.2835 t ₂₂ / T = 0.8232 t ^ / T = 0.7835

t ₁₄ / T = 0.25 t ₁₅ / T = 0.2205 t ₂₄ / T = 0.75 t ^ / T = O _s 72O5

t ₁₆ / T = 0.1938 t ₁₇ / T = 0.1693% / T = 0.6938 t _2y / T = 0.6693

t ₁₈ / T = 0.1465 t _1g / T = 0.125 t ₂₈ / T = 0.6464 t ^ / T = 0.625

t ,, _1O / T = O, 1O47 ^ ₁₁ / T = 0.0854 t _g ^^ = 0

_{1t15 2i14t 2t15} / T = 0.5159

jii = 2.4414

n = 1 n = 1

It follows according to the invention for the arithmetic mean in volts

^ =? * Έ7ΤΖ Οιψΐ- + 2.4414 - 9.9413

= 2 - 0.9687 = 1.0313V

2 V · JAMs 0.9687 V

Pur the symmetry deviation follows according to the invention S = - O.

and for the RMS value of the pure alternating variable ² ^ fS £ + 2,4414] - (1, O313) ² 2,6914 - 1,0636 = It6278 V ²

EWW = 1.2758 V

The exact results are

AM = 1V AIZi ₀ = 2V JAM = ~ 1V s - 0.5EV = 1.291V

The advantage of the method according to the invention becomes clear in the determination of the arithmetic mean AM, since there is no need for etching, so that only the four basic arithmetic operations are required for the evaluation.

Claims (1)

In fi nance s pru ch 1. A method for digital measurement of the arithmetic and the root mean square and the curve symmetry of periodic alternating signals of any Kurvenfortn, even with DC share according to the main patent (WP G01R / 240 896.5), characterized in that in dependence of the positive peak value A "and the negative peak value Έ two Level groups as follows ^P 1 η '2 η β B + (X - B) fn / tt = X - (XB) η s 0.1 · · · Ν-1 in a sequence of measuring steps or in a single parallel measuring process, the time intervals,, _Λ , t ₉ assigned to these levels are determined in such a way that for the first level group Έ * "the time interval I ₉ U sections for which the alternating signal is more positive than a level P ^ and for the second level group the time intervals for which the alternating signal is more negative than a level P ₂ iat or vice versa and in a final evaluation step the following quantities are determined: the arithmetic mean after AM = .AM ₀ + Δ AM = Χ + Ή. I-B ^t ip "^'t 20 n = 1 N-I ^ n -fjr- the symmetry deviation after S «AM and the RMS value of the pure change amounts N-1 rX-B, 2 10 ₊ n = 1 - (AM-F) ' For this a sheet of drawings.