Meter Testing - Traditional

1
Meter Testing - Traditional New Techniques
GE Meter School

• Presented by
• T Jeffrey Wolters
• GE Meter, Somersworth, NH
• The following people designed, developed or
  provided the substance for the material
  presented
• Dave Elmore, Don Bullock (ret), Les Rosenau, Mike
  Coit, Warren Germer, Joe Provost

2
Some K Definitions

• Kh - Watt-hours per (equivalent) disk revolution
• Ke - Energy value per pulse (used commonly for
  PI)
• K - Register programming watt-hour constant (for
  EV, Kh/12 K)
• Kt - Test constant Watt-hours (or VArH) per
  pulse
• Kr - Dial Multiplier
• Inside the kV, K values have no real meaning or
  bearing on energy calculations. All energy values
  are supplied from the DSP in engineering units.

3
GE kV Meter Test LED
Test LED
4
GE kV Meter Functional Diagram

• DSP blinks the Optocom LED every Kt watt-hours
  (or VArh)
• Through programming, one can set the calibration
  pulse rate (Kt) to just about any practical
  value. This can
• allow one to adjust testing speed (within the
  confines of other, more restrictive parameters)
• standardize pulse values regardless of meter
  type.
• Remember the kV DSP outputs (and the register
  stores) actual energy values rather than pulses
  which can be converted to energy values.

5
Kt Value Selections

• The kV meter offers a great deal of flexibility
  in specifying Kt (test pulse) and PI (Pulse
  Initiator) values.
• To obtain the most accurate results, it is
  important to understand more about how the kV
  calculates its values.
• There is a number stored in the kVs firmware
  which is responsible for making the math work
  based on meter class.
• Rev3 below Rev 04 above
• Meter Class kV divisor kV divisor
• 20 0.0006 0.0005
• 200 0.006 0.005
• 320 0.009 0.0075

6
Kt Value Selections

• Rev3 below Rev 04 above
• Meter Class kV divisor kV divisor
• 20 0.0006 0.0005
• 200 0.006 0.005
• 320 0.009 0.0075
• The use of Kt (and PI) values that are evenly
  divisible by these numbers will prevent small
  errors from being introduced due to truncation of
  repeating or long decimal values.
• Most traditional Kt (and PI) value selections
  provide error-free operation, or contain
  insignificant errors.

7
Kt Value Selections

• To calculate the magnitude of any possible error,
  use the formula above.
• Example Use a Kt value of 1.0 on a CL 20 meter,
  Firmware 3
• 1-((Integer(1/0.0006))/(1/0.0006))x100
• 1-(1666/1666.666)x100
• 0.04 error

8
Kt Value Selections

• Now Use a Kt value of 1.0 on a CL 20 meter,
  Firmware 4
• 1-((Integer(1/0.0005))/(1/0.0005))x100
• 1-(2000/2000)x100
• 0.00 error

9
Some Factors Affecting Measurement Uncertainty

• Total meter performance is a function of many
  variables.
• The methods we use to verify the meters
  performance are also subject to many variables.
• In general, any variable which affects the meter
  and the standard differently will lead to
  UNCERTAINTY.

10
Some Factors Affecting Measurement Uncertainty

• In digital meters, the more obvious factors
  affecting uncertainty include
• Repeatability limits of the test equipment being
  used
• Random systematic errors including response
  time and characteristics of the meter calibration
  LED circuit and the test equipment optotransistor
  circuit used in the pickup device
• Digitizing error (remember we reduced this by
  careful selection of sampling rate)
• Inherent measurement differences between filtered
  and unfiltered methods of measuring AC power
• Temperature, frequency, stray magnetic fields,
  current and voltage non-linearities...

11
Measurement Uncertainty Example

• Inherent measurement differences between filtered
  and unfiltered methods of measuring AC power

• Take two meters having equal accuracy
• The first, a heavily filtered meter (e.g.,
  RadianTM standard). It integrates average power
  with respect to time
• The second, an unfiltered meter (e.g., GE kV
  meter). It integrates instantaneous power with
  respect to time.

12
Measurement Uncertainty Example
Inherent measurement differences between filtered
and unfiltered methods of measuring AC power

• Compare the meters power calculation during the
  time interval from a to b in the figure.
• The unfiltered meter will record energy
  proportional to the horizontally shaded area plus
  the cross-shaded area.
• The filtered meter will record energy
  proportional to the cross-shaded area.

13
Measurement Uncertainty Example
Inherent measurement differences between filtered
and unfiltered methods of measuring AC power

• In this example, the Instantaneous power
  measurement is obviously larger than the Average
  power measurement. For other small time
  differences (a and b), Average power could be
  larger or the same.
• As test time increases, the effect of this source
  of uncertainty disappears. Also, if you could
  test over an exact integral number of cycles,
  this uncertainty would not exist.

14
GE kV Flash Calibration

• Flash Calibration Objective
• Calibrate voltage gain, voltage phase, current
  gain and current phase
• Accomplish multiple measurements simultaneously
• Minimize calibration time
• Minimize number of conditions necessary to assure
  overall meter accuracy

15
GE kV Flash Calibration

• How Flash Calibration Works
• Meter is put into Flash Cal Mode via Optocom
• Meter blinks Optocom LED and starts
  accumulation
• After specified time, meter stops accumulation
  and blinks LED
• Accumulated quantities are read and compared to
  standard meters
• New correction factors are calculated and
  programmed into the meter as required
• Rerun until all results are acceptable

16
GE kV Flash Calibration

• How Flash Calibration Works (cont)
• Factory has independent, per-phase equipment
• Meter Flash Calibrated at FL and LAG
• Meter verified using both Flash Calibrate mode
  AND pulse mode
• Total Flash Cal time is just a few seconds
• How can we do that? ... Remember, if you choose
  an amount of time that is an exact integral
  number of power cycles, there is no uncertainty
  error between instantaneous and average power
  measurements.

17
9S vs 8S Testing Conditions

• Why doesnt the 9S kV meter seem to test
  correctly on my calibration panel set up for an
  8S meter?
• The socket terminal connections are the same.
• The actual (in)SERVICE phasors presented will be
  read by either meter. In fact, since the 9S meter
  is a Blondel solution, it will properly meter
  unbalanced conditions, where the 8S will not.
• The problem is that most test panels do not (by
  default) apply the voltages the way they would be
  presented in service. They treat the 8S as a two
  voltage element device, applying one voltage A-B
  and the other C-N. For the 8S, this is good
  enough to test, but for the 9S kV, there is no
  real A-N voltage reference since the A-B voltage
  is subjected to a voltage divider between kVs
  A-N and B-N voltage sensing resistors.
• In service, this is not a problem because the
  line to neutral volts are all independently
  regulated by the distribution transformers.

18
9S vs 8S Testing Conditions

• Some test panels can be configured to present a
  more realistic 4 wire delta input to the meter
  under test.
• This can also be used on 8S meters.

19
kV Test Voltages

• Typically, apply Test Voltage (TV nameplate).
  120v.
• Other voltage(s) may be used. Wide range meter is
  120-480v Line to Line (not Line to Neutral).
• WHEN TESTING, DO NOT APPLY 480 L-N.
• IN POLYPHASE, THIS WILL EXCEED THE 575V LIMIT ON
  THE POWER SUPPLY
• WITH REVENUE GUARD, THIS WILL EXCEED THE 575V
  LIMIT ON THE POWER SUPPLY AND THE REVENUE GUARD
  OPTION BOARD.
• Be aware of test board power supply limitations
• some boards were not designed with newer meter
  switch-mode power supplies in mind.
• Test board error magnitudes typically increase
  with test voltage increase.