The Sampling Direct Current Current Transformer
While a student at CERN I got the task of improving the design of the current transformers they were using to measure the magnet power supply currents.

These currents were carried in 4cm diameter copper bars and ranged up to 1000 Amps. The current measurement had to be made to an accuracy of 1 part in 105. Also the measurement system had to be isolated from the bar: i.e. done without electrical connection.

Current measurement systems like these all use the magnetic field generated by the current in the bar. For high performance systems like these large mu-metal torroids are use which are very expensive. The have to be mu-metal because very low inductive currents are required to achieve the required accuracy.

The existing design used three torroids in a system which was able to compensate for the DC component. The problem with it (aside from cost) was that the low bandwidth associated with high transformer inductance and DC compensation scheme.

My supervisor suggested I try a sampling approach, and this is what I did.

Essentially, there is a circuit on the secondary of the transformer which forces the transformer out of DC saturation for a few milliseconds. During this time the secondary current is measured with a precision analog to digital converter.

The first 1000A prototype is shown right.

And here is the production unit with the bus bar shown going through it.

I also developed a 100 Amp version on a plug in card shown here.

 

I was at CERN for one year. I am not sure happened to the 1000A version but I know the 100A version was used in the positron storage ring built the next year.