71M6515H
Energy Meter IC
DATA SHEET
MARCH 2008
Registers Used for Calibration
CAL_IA (0x24), CAL_VA (0x25), CAL_IB (0x26), CAL_VB (0x27), CAL_IC (0x28), CAL_VC (0x29):
These registers adjust the gain for the current and voltage measurements of each phase for the purpose of calibration. The
calibration factors have to be stored by the host and written to the registers of the 71M6515H after power-up. The allowed
range is (215 – 1) to –(215 – 1). The default value of 16384 equals unity gain.
If a voltage measurement of phase C is higher than expected, CAL_VC has to be adjusted to:
CAL_VC = 16384 / (1 + error)
Error must be expressed as a fraction, not a percentage value.
If the percent error is +3.5%, the relative error is 0.035, and the calibration factor becomes:
CAL_VC = 16384 / (1 + 0.035) = 15829.952,
which is rounded up to15830.
KVAR (0x2F)
This register holds the relative gain of the VAR calculation with respect to the Watt calculation. The value should always be
6444.
PHADJ_A (0x2A), PHADJ_B (0x2B), PHAD_C (0x2C)
These registers hold the phase correction factors for channels A, B, and C. The values are used by the CE to compensate
for phase errors induced by current transformers. The allowed range is (215-1) to -(215-1). See the Calibration Procedure
section for applicable values.
If the CE is operated in Rogowski Coil mode, no phase compensation should be required. The default value is not zero and
should need to be changed only slightly, if at all. See the Calibration section for details.
QUANT_W (0x36), QUANT_VAR (0x37), QUANT_I (0x38)
These registers hold DC values that are added to each calculated product in order to compensate for internal quantization
(a very small amount) and external noise. These values are normally set to zero. The LSB values for these variable are
listed in Table 7.
Variable
LSB Value
Unit
W
QUANT_W
QUANT_VAR
QUANT_I
(VMAX IMAX/ In_8)* 1.04173*10-9
(VMAX IMAX / In_8)* 1.04173*10-9
(IMAX2 / In_82)* 5.08656*10-13
VAR
A (rms)
Table 7: LSB Values for QUANT Variables
Nonlinearity is most noticeable at low currents, and can result from input noise and truncation. Nonlinearities can be
eliminated using the QUANT_W register. The error can be seen as the presence of a virtual constant noise current that
becomes dominant at small load currents.
The value to be used for QUANT_W can be determined by the following formula:
error
V ⋅ I ⋅ In _ 8
100
QUANT _W = −
VMAX ⋅ IMAX ⋅ LSB
Where error = observed error at a given voltage (V) and current (I),
VMAX = voltage scaling factor, as described in section Scaling of Measured Values
IMAX = current scaling factor, as described in section Scaling of Measured Values
LSB = QUANT LSB value = 1.04173*10-9W
Page: 34 of 57
© 2005-2008 TERIDIAN Semiconductor Corporation
V1.4
TERIDIAN [ TERIDIAN SEMICONDUCTOR CORPORATION ]
