Transformer Isolation for electronic signals

Is your 2 kV ac or dc ? If it's ac I'm assuming its 50 or 60 Hz. Your reqirement for dc to 10 Hz prevents the use of transformers and capacitors. That limits your selection to optical or RF data transfer systems. Of the two optical links are less prone to interference and (in my experience) are more relaible. Digitizing the signal will provide a means of obtaining the DC component as well as providing nearly noise free information replication.

I agree that optical isolation is more or less free from electrical interference and use of ADC is better as data is retained near zero as well.

My experiance is that optocouplers give the best quality in terms of a noise free dynamic range and minimal coupling to external disturbances and have the least problems in manufacturing. If you digitize the signal you only need one opto but if you use analog you need a second one to implement the "feedback" path across the same voltage barrier.

Since you only need ~40db of dynamic range a single coupler and a low bit depth ADC running at a few hundred kilohertz may be cheaper. If the signal's analog a Dual Opto shouldn't be too expenive, and may prove cheaper than a digital solution.

The biggest problem with isolation comes due to analog zero level noise, temperature drifts and gain change over time. I thought of using VFC and FVC method and this has also same problem and netter than 10-bit is difficult even though 1MHz VFC is common.

Some people feel that programmable logic may give problem so they do not want such things in the signal converters. Now the option is clear analog versions only.