EVAL-ADE7756EB

–8–

REV. PrB 01/01

PRELIMINARYTECHNICALDATA

Measuring CT Phase Errors using the ADE7756

The ADE7756 itself can be used to measure a CT (and

external components) phase error during calibration. The

assumption is that the ADE7756 has no internal phase

error (PHCAL = 00 hex) and the error due to external

components is small (<0.5°). The procedure is based on a

two point measurement, at PF=1 and PF = 0.5 (lag). The

PF is set up using the test bench source and this source

must be very accurate. The ADE7756 should be config-

ured for energy measurement mode.

An energy measurement is first made with PF=1 (mea-

surement A). A second energy measurement should be

made at PF=0.5 (measurement B). The frequency output

CF can be used for this measurement. Using the formula

shown below the phase error is easily calculated:

















⋅

−

=

°

−

3

2

2

tan

)

(

1

A

A

B

Error

Phase

For example, using the frequency output CF to measure

power, a frequency of 3.66621Hz is recorded for a PF=1.

The PF is then set to 0.5 lag and a measurement of

1.83817Hz is obtained. Using the formula above the

phase error on Channel 1 is calculated as:

°

+

=

















⋅

−

=

°

−

091

.

0

3

2

66621

.

3

2

66621

.

3

83817

.

1

tan

)

(

1

Error

Phase

The formula will also give the correct sign for the phase

error. In this example the phase error is calculated as

+0.091° at the input to the Channel 1 of ADE7756. This

means that the CT has introduced a phase lead of 0.091°.

Therefore the phase difference at thr input to Channel 1 is

now 59.89° lag instead of 60° lag. Determining whether

the error is a lead or lag can also be figured intuitively

from the frequency output. Figure 11 shows how the

output frequency varies with phase (cos{

φ}). Since the

output frequency B (1.83817Hz) at the PF=0.5 lag setting

in the example is actually greater than A/2 (1.833105Hz),

this means the phase error between Channel 1 and Chan-

nel 2 was actually less than 60°. This means there was

additional lead in Channel 1 due to the CT.

60

PF=1

PF=0.5

PF<0.5

PF>0.5

CF (Hz)

PF=0

Phase lag

360

Frequency B > A/2

Phase difference < 60 lag

Figure 10—CF Frequency Vs Phase(PF)

Using the Phase Calibration to correct

small (<0.5°) external phase errors

From the previous example it is seen that the CT intro-

duced a phase lead in Channel 1 of 0.091° . Therefore

instead of 60° phase difference between Channel 1 and

Channel 2, it is actually 59.89°. In order to bring the

phase difference back to 60°, the phase compensation

circuit in Channel 2 is used to introduce an extra lead of

0.091°. This is achieved by reducing the amount of time

delay in Channel 2.

The maximum time delay adjustment in Channel 2 is

±143µs with a CLKIN of 3.579545MHz. The PHCAL

register is a signed 2's complement 6 bit register. There-

fore each LSB is equivalent to 4.47µs. In this example the

line frequency is 50Hz. This means each LSB is equiva-

lent to (360° x 4.47µs x 50) = 0.08°. To introduce a lead

of 0.091° the delay in Channel 2 must be reduced. This is

achieved by writing -1 (FFh) or +0.08° to the PHCAL

register.

Figure 11—Writing to the PHCAL register to correct

phase error

Correcting large external phase errors

In this example the phase correction range at 50Hz is only

approximatey ±2.5°. However it is best to only use the

PHCAL register for small phase corrections, i.e., <0.5°.

If larger corrections are required the larger part of the

correction can be made using external passive component.

For example the resistors in the anti-alias filter can be

modified to shift the corner frequency of the filter so as to

introduce more or less lag. The lag through the anti-alias

filters with 1k

Ω and 33nF is 0.56° at 50Hz. Fine adjust

can be made with the PHCAL register. Note that typically

CT phase shift will not vary significantly from part to

part. If a CT phase shift is 1°, then the part to part

variation should only be about ±0.1°. Therefore the bulk

of the phase shift (1°) can be canceled with fixed compo-

nent values at design. The remaining small adjustments

can be made in production using the PHCAL register.