How measurement system solvesaccuracy loss measurement tradeoffby Christoph Denk, Sales Director, EPRO Gallspach GmbH, Gallspach, Austria

Abstract

International standards and quality policy of manufacturers have increased the needs ofmeasurements for transformers and reactors. Many measurements have been raised fromtype tests to routine test, making a measurement system with improved data managementmandatory. Measurement equipment manufacturers have to design systems that takeadvantage of most of the technology improvement, from measurement electronic to ITsystem. Not only the measurement is important, but also the complete chain, datamanagement, reporting and accuracy assessment of the measurement have to beautomatically computed for the end user.

Tr a d i t i o n a l L o s s M e a s u r e m e n tapproaches

Loss measurement classically is facing thecontroversial requirements of measuringiron losses vs. measuring copper losses.Measuring of iron losses implies highvoltage (for example the nominal voltageof the transformer) in an open circuit andlow current values. Measuring of copperlosses in a short circuited cycle implieshigh current (for example the nominalcurrent of the transformer) and lowvoltage.

The traditional approaches of lossmeasuring have been dealing with thesecontroversial test requirements byseparate test infrastructures. This meansfor the test engineers the process oftesting was including a high demand ofcabling time for providing the correct setupfor each kind of test, using the right kind ofmeasuring equipment for each single testfor providing the best possible measuringaccuracy.

The increasing production capacities ofmodern transformer plants and theincreasing customer’s demand on routinetests putting a lot of pressure on the testlabs. More

Transformers have to be tested and moredistinct tests have to be accomplished atshorter available time resources.

Out of this reasons new approaches fortransformer testing have to be found.Integrated systems offer drasticallydecreased testing times and thereforereduced testing costs beside to majorother advantages in the workflow of thestate of the art test labs.

Advantages of integrated systems

Digital processing of measurement values,as already mentioned, offers a variety ofpossibilities to the measurement processitself. Synchronic and automatic readingand processing of data eliminates readingerrors what caused uncertainties anderrors with old measurement instruments.

Digital processing techniques are highlyimportant for the digital sampling ofmeasurements. The importance ofharmonics contents of the measureds i g n a l h a s t o b e d e t e r m i n e d .Electromagnetic noise what can be foundin industrial equipment has to be filtered.

The most important advantage by digitaltechniques is the possibility of automaticcompensation of measurement devices.Only in case of highly accurate and highlystable standard transformers are used formeasurement of voltage and current anautomatic compensation of amplitude andfrequency can be done easy and reliable.The requirements and technical abilities ofthe standard transformers wi l l behighlighted later in this article.

The overall accuracy of measurementshas to be assessed constantly. Theresponsible person constantly has tocollect the accuracy data of differentmeasuring systems and devices. Plus to

this we have to consider the relationship ofdifferent uncertainties and the relations ofthis on the overall measurement results.Integrated measuring systems drasticallyreduce the work and the re lateduncertainty that constant assessmentbrings with. With integrated systemscomplete accuracy data is availableconstantly; this includes each part of thewhole system. An integrated measurementsystem has to point out directly, at eachmeasurement, the values of measurementuncertainty of the related measurement.

State of the art IT management is anotherbig argument for fully integrated systems.Both the manufacturer of the system aswell as the user of the system easily takesadvantages of building up databases oftest routines and test objects on a dailybase. While the user of the system canbuild up test object databases anddatabases of different test routines, themanufacturer of the system is able to dolong term statistical calculation for verifyingthe accuracy data; this has also its mainimportance for calibration and calibrationintervals.

Calibration and verification

Any measurement system is only accurateduring its calibration period, any systemhas to be recalibrated or verif iedper iodical ly. The structure of themeasuring system is having tremendous

influences on the calibration periods. Infact calibration periods always are definedby the quality management system of thespecific user of the system but in generalnational authorities are giving at leastinformal indications for different types ofdevices. Normally passive (inductive)components as standard transformers areused to have longer calibration periods (forexample 8 years or even longer). Thishelps the customer in terms of economictest bay management while the devicescause less calibration cost and less labdowntime.

Electronic system parts have to becalibrated more often, informal indicationsspeak of yearly or bi-yearly basis.Manufacturers of integrated test systemshave to foresee this calibration work. Theparts for calibration should be size andweight reduced. Also the process has tobe considered. In case of availability ofspares during the calibration perioddowntimes can be reduced to zero. Theusage of front-end devices ensures theability of a simple change for calibrationpurposes.

For best measurement quality it isp re fe rab le to use au tomat ic se l fverification measurement devices. A selfverification will be performed prior to eachmeasurement process and this is building

confidence in measurement results as wellas good traceability of all measurements.

It is obvious to mention that all parts ofintegrated measurement systems have tobe calibrated related to the commonstandards as well as the entire completesystem. Traceability to national standardsfor voltage, current or power is a must.

This rigid way dealing with calibration andverification is the only effective way onlywith this rigorous trading about calibrationor verification can provide a high level ofaccuracy in the loss measurements. Ameasurement system foreseen for flexiblecalibration is the only way to avoid sometradeoff between the calibration period ofthe system and the production throughput.If calibration process is not a bottleneck fortesting resources, calibration will not bedelayed for product ion throughputreasons.

Accuracy of loss measurements

The evaluation of the accuracy of powermeasurements for the losses has beendescribed in the standards since a longtime, and is still of the first importance.

The basic calculation of the powermeasurement is given in [1] and iscomposed of the different measurementuncertainties of each measured quantity.

As described in the standards, theaccuracy of measurement is of firstimportance for the voltage and current, butthe most important factor for powermeasurement is the phase angle or powerfactor measurement.

The measured power losses a reexpressed as

[1]

For the composite relative error theformula gives:

Modified, this equation shows that whenthe phase angle is small due to theinductive nature of the equipment undertest, the power measurement error, whichvaries with the inverse of the power factor,will increase.

If somebody first analyses the first terms ofcurrent and ratio, another term comes intoaccount as the loss measurement usuallydeals with high voltages or high currents.This term is the ratio of the divider usedduring the measurements. Reducing theuncertainty of voltage and currentmeasurements is mandatory. But, asexplained in the standards;the ratio of themeasurement t ransformer can becompensated in the calculations of powerto improve their accuracy. In order tocompensate these ratios, the ratiodeviation must be known exactly with thebest accuracy and last but not least, mustbe very stable, not depending fromt e m p e r a t u r e o r h u m i d i t y o relectromagnetic noise for example.

The phase angle between current andvoltage or the power factor must bemeasured accurately to reduce theuncertainty of the power.

The Phase or power factor measurementerrors are impacting the active powermeasurement very fast, applying theequations from [1] directly leads to curveslike the Figure below.

The typical curve of the Figure wasobtained by assuming current and voltageat ±200ppm and a phase error ofmaximum 0.4 min.

The different error sources for the phaseare coming from the measurementtransformers and eventually, the front endelectronic and the digitizing systemsynchronization.

As described before in this article, thephase error coming from the measurementtransformers and the front end electroniccan be compensated by the measurementsystem if they are well calibrated andstable over environment conditions andtime.

I n a d i g i t a l p o w e r m e t e r , t h esynchronization of the sampling systemmust be of the first importance tosecurethe lowest jitter between thesamples. The latest digi ta l signalprocessing techniques have to be used toprovide the best determination of thephase shift between current and phase.

In practice, the phase errors are verydifficult to quantify from the theory. Someverification setups have to be measuredover long periods of time and the phaseerror can be determined experimentallyusing statistical methods. In this case, thesame current or voltage will be observedusing identical measurement devices andan important amount of results will be

acquired. Analyzing these results and theirdispersion will increase the confidence inthe determination of the phase error.

These experimental results then need tobe calibrated with equipmentthat arethemselves calibrated to the nationalnormal (like PTB for example).

Parameters that degrade the accuracy

Effect of electromagnetic noise

The most actual point of interest for thelosses measurement is the matter ofe l e c t r o m a g n e t i ca l l y n o i se . M a n ylaboratories are using electronic powersupplies for their tests in order to getflexible power conditions like higherfrequencies.

These electronic power supplies areswitching power supplies and are emittinga wide spectrum of electromagneticallynoise. This noise has already been studiedin many articles mainly for partialdischarge measurements.

The electromagnetically noise at highfrequencies can be coupled on themeasurement devices themselves or onthe connecting cables. A modular systemwith front-end and digitizing as near asp o s s i b l e f r o m t h e m e a s u r e m e n t

transformers reduces the coupling ofnoise.

Electronic power sources generally usecapacitor banks in order to stabilize thecontrol system or provide a tank ofreactive power. The control system andthe capacitor bank can generate higherharmonics content in the power circulatingin the device under test.

The standards do not require yet afrequency analysis, but the increasingcontent in harmonics gives importantinformation about the measurementconditions.

Effect of range switching

An important aspect to reduce themeasurement uncertainty is the stability ofthe measurement setup. Changes in themeasurement setup directly impact theuncertainty of the measurements.

Measurement devices including rangeswitching with relays influences theimpedance in the measurement circuitsand then brings an additional uncertaintyto the measurements. Even the bestmeasurement front end will be inaccurateif the setup is changed during themeasurements.

Important but forgotten points toaccuracy

As soon as a measurement discussionstarts, it ends quickly to accuracyassessment. During these discussions, themetrological point of view is immediatelyon the scene, but some points are oftenforgotten, even if they can affectdramat ica l ly the accuracy of anymeasurement, and in the worst case makethe measurement unusable.

Effect of the system user

The use of digital measurement deviceshas reduced the uncertainties due to the

user against the old measurement deviceswith arrows. The “reading” errors havebeen eliminated.

There is still an error that can come fromthe user and that affect the measurement.The user can still make mistakes duringthe test, especially if the user is notexperienced or trained to these types ofmeasurements.

The use of the latest man machineinterfaces with touch screen or otherdisplays guides the user during the tests inorder to avoid any failure during themeasurements. The use of intuitivegraphics is of first importance to terminatethrough the requirements of the standards,without forgetting any measurement oraction.

Integration of automatic power sourcecontrol can also avoid testing at badcurrent or voltage levels. Issuing anautomatic report from secured test resultdatabase is also a security that differenttest results will not be mixed at the end.

Effect of equipment quality

As seen above, the measuremente q u i p m e n t d i r e c t l y i m p a c t s t h emeasurement uncertainty of the power, notonly with the initial accuracy, but mainlyfrom their stability.

In case of usage of inductive parts the enduser is gaining a variety of advantagesover the life cycle of any measuringsystem. In compares to electronic orcapacitive dividers inductive parts arestable over decades and under outerinfluences. This is shown in the unofficialmeaning o f major meteoro log ica lauthorities, for example in Germany.Where inductive measuring instrumentsare scheduled for recalibration each 16years only there is the broad meaning ofcalibration for capacitive parts every singleyear or at least all 2 years.

Inductive standard transformers offer thehighest level of accuracy in voltage/currentmeasurement and also in phase angle.Rea l l y s ta te o f the a r t s tandardtransformers should offer accuraciesbetween +/-0,005 and +/- 0,01% and aphase angle uncertainty of not more than0,5-1,0min depending on the height of therated voltage/current.

Obviously accuracy is only one parameterwhat has to be achieved for stable andexact measuring, the other point isstability. Stability in fact has two meanings,y o u c a n d i s t i n g u i s h b e t w e e nenvironmental stability and range as wellas time stability. Both aspects of stabilityare perfectly covered by inductivestandard transformers.

Environmental stability means that themeasuring output has to be on the samelevel for same measurements even if outerconditions as temperature, humidity of airor altitude strongly varying. To be broughton a practical example this means themeasuring device has to show hardlysame measuring results weather it isoperated in Mumbai where humidity is veryhigh or operated in Arizona / USA wherehumidity is low and temperatures areextremely high, or if it is operated in Quito /Ecuador what is hardly 3.000m above sealevel. Inductivity always calculates on

same physical values and formulas.Theoretically there are compensationfactors for all these values but in case ofinductive devices these factors easily canbe neglected, there are no empiricalvalues for this.

Time stability is guaranteed at inductivestandard transformers because of theconstruction/design parameters. Thelifetime of inductive standard transformersis calculated in decades it is proventechnology with no variations over time.Range stability means the development ofthe measuring uncertainty over a widemeasuring range. High quality standardtransformers are defined by accuracy. It iseasy to reach a very high measuringaccuracy at the rated voltage/current. Thecritical point and question in the daily workin the test lab is the range where thestability is guaranteed. High qualitystandard transformers are offering a failureof less than 0,1% over a wide measuringrange (i.e. between 20 and 120% ofrated). Even below 20% of rated forexample at a minimum of 1% of rated thefailure should be constant. In this case thefailure will be more than 0,1% but stabilityconditions should be the same in all othercategories.

If these conditions are completely reaches,and only in this case it is possible that themeasurement system compensates thetransducers ratio and phase displacement.In this case the end user receives a

tremendous advantage. Stability andperfect accuracy of a measuring systemover a wide range of measuring values. Ina practical point of view this meansflexibility in the test lab.

Effect of temperature measurement

Maybe one of the most neglected parts ofthe loss measurement, maybe because itis not directly related to the power lossmeasurement , is the temperaturemeasurement. Every standard gives amethod to correct the measured losses,depending on the materials composing thetransformer or the reactor. Where everyexpert will discuss about some tens of apercent in the loss measurement accuracy,the same expert will take an approximatedmeasurement of the temperature to correctthe loss according to the standards.

The following example shows howimportant are the accuracy of temperaturemeasurement in the final results of theloss measurements.

Considering the Load Losses in W asdefined in the standards, the temperaturecorrection applied to these losses iscalculated using the following formula:

Where

Pj(Ir, T): Ohmic losses at Ir and T.

Pa (Ir, T): Additional losses at Ir and T.

Tk: Temperature coefficient of EUT windings.

Tr: Assigned temperature, in °C.

T: Temperature measurement, in °C.

Ir: EUT assigned current, in A.

PLL: Load losses, in W.

Taking a small academic example tocalculate the consequences of a badtemperature measurement uncertaintygives the following:

With:

Tr = 75 °C, Assigned temperature, in °C.

Ir = 10A, EUT assigned current, in A.

Tk = 235 (Copper), Temperature coefficient ofEUT windings.

RTr = 0.715Ω ± 0ppm EUT windings resistanceat Tr

Measurements Calculations

P TRT

PLL (W)

90W±291ppm

70°C± 0°C

0.727Ω± 0ppm

88.55W± 435ppm

70°C± 0.3°C

0.727Ω± 984ppm

88.55W± 926ppm

70°C± 1°C

0.727Ω± 3279ppm

88.55W± 2757ppm

70°C± 5°C

0.727Ω± 16390ppm

88.55W± 13620ppm

This example has shown the importanceof the temperature measurement in thecomputed results. It becomes obviousfrom these results that a measurementsystem must include an accuratemeasurement of the temperature,s y n c h r o n i z e d w i t h t h e p o w e rmeasurement. This ensures that duringt h e a v e r a g i n g t i m e o f t h e l o s smeasurements, the measurements area l w a y s c o r r e c t e d w i t h t h e r i g h ttemperature.

Conclusion

This Article shows that modern measurement centers are complex systems. The task ofmeasuring transformers is not an easy one and brings various tradeoffs and with it. Modernmeasurement centers are able to minimize the tradeoffs and the total uncertainty of themeasurement process. Therefore some abilities should be guaranteed, the architecture ofthe measurement system can help minimizing tradeoffs and uncertainties. The Architectureof a combination of electronics and automation in combination with well proven inductivemeasuring devices brings advantages of two worlds together. The end user has to focus oneasy handling, security and accuracy. All these points are important and are having a biginfluence on the process and therefore on the productivity of a lab.