What is the recommendation for protecting the VAP, VAN, IAP-IFP and IAN-IFN inputs of the ADE7816? Previously it has been recommended to use back-to-back diodes to ensure the maximum voltage across the inputs does not exceed +/-2V. Is this still the suggested approach? If not, please provide the recommended protection scheme. Thanks,
If the recommended anti-alias filter(1K or larger) is used the internal esd protection is enough to protect the part. Diodes are not required. The 2v max spec will be limited by the 1K and esd diode drops. You will still rail the modulator but it will recover after event is over.
Thanks for your answer, however would it be possible to elaborate on the details of the internal esd protection diodes; what rails they are connected to and the current handling capability. I would like to confirm that even with severe overloads on the CTs (say a monentary short on the primary) that such an event would not cause any permanent damage.
As a follow-up we did some overload testing of one of the standard 20A CTs we use. We fed the 20A CT with 110A @ 60Hz and with the standard anti-aliasing filter of 1K2 and 22nF we measured 1.72V RMS and 5.11V Pk-Pk across the capacitor (effectively the input of the ADE7816).
When we shorted out the capacitor (emulating the ESD diodes conducting) and measured the voltage across the 1K2 resistor we saw 1.66V RMS and 5V Pk-Pk. This represents 1.4mA RMS or 4.2mA Pk-Pk through the ESD diodes.
Is this acceptable, such that it won't damage the ADE7816?
I would not expect 110 amps to last very long in a 20 amp Circuit the breaker would trip. If this fault were to happen over and over there could be an issue. If you change the anti alias filter to 10K and 2.2nf this would put you in a safer range.
Thanks for the reply Dave. Based on your comments I think I will move to a filter of 3.16k and 10n. This is a compromise between overload protection and keeping the R as low as possible to avoid excessive drop due to the ADE7816 input impedance and bias offsets etc.
The values selected give a corner frequency of 5kHz. Is that still the recommendation for standard CTs? I note that the 1k/22nF values often cited have a corner frequency of approx 7KHz.
Hi Dave, any comment re the filter values and cut-off frequency?
The ADE7816 measurement bandwidth is 2K. Ideally you want a corner high enough to be flat in the dc to 2K range. At the same time we want enough attenuation at the sample rate to reduce any signals folded back into the band of interest. 5K is a good compromise.