New frequency specific phase correction


The new release due out later this week includes this broad new functionality. I thought I’d take a minute and describe what it does:

First the problem:

IoTaWatt measures power by sampling the voltage and current signals associated with a circuit. The IoTaWatt is isolated from the high voltage and current through the use of transformers - VTs and CTs. Those transformers produce a good representation of the signals, but also introduce a time-shift that can effect the accuracy of the measurements. This is also called phase-shift.

With it’s very high sample rates, IoTaWatt has always had very good mechanism for correcting the net phase shift between the VT and CT. The problem has been identifying what the correction should be. Both VTs and CTs exhibit significantly different shift at 50Hz vs 60Hz. CTs also have different shift depending on the primary current they are measuring.

So determining the amount of shift to compensate, on a cycle-to-cycle basis, is the real problem. Historically, monitors seem to be designed for 50Hz or 60Hz, and with respect to CTs, compensation is usually fixed to the shift at an expected typical primary current.

The solution:

This has been, overwhelmingly, a measurement and data collection project. It has required collecting and measuring samples of all of the VTs and CTs that are in the tables. In the case of the CTs, an apparatus had to be built to measure the phase-shift across the spectrum of the CT current range - at both 50Hz and 60Hz.

Once that was done, the shift for each CT at various primary current ranges, at both 50Hz and 60Hz, was added to the device tables.

As each cycle is sampled, the frequency and primary current are used to lookup the prevailing phase-shift parameters and the resulting net compensation is applied. This is done about 40 times per second.

Going forward:

The mechanism for doing all this is sound, and using unity power-factor loads (more-or-less pure sine waves) the compensation can be reconciled and appears to work well. There is more work to be done to fine-tune the apparatus that measures the CTs, but in general the results have been encouraging and should yet further improve overall accuracy in situations where the line frequency or primary current did not correspond to the old scalar compensation parameters.

When the release is installed, all of this will happen automatically.