Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
NVE's Isolation Products Business Group is building a continually expanding family of galvanically-isolated signal interface products based on our patented IsoLoop® technology. This technology allows on-chip isolation, providing unprecedented performance characteristics. The IsoLoop's characteristics include high speed, small size, minimal pulse width distortion and high common mode rejection coupled with a high stand off voltage. When this device is integrated with the appropriate on-chip electronics, a wide variety of integrated devices having isolated interfaces can be created.
You can do a trick to make the digitally oriented couplers more linear for analog signals. On the signal source side, have an op amp drive two coupler inputs in series (couplers are current mode devices and this will make the two coupler inputs have identical currents). Take one coupler output to wherever it needs to go and take the other coupler output back around to the op amp input as a feedback circuit.
This will improve things assuming that both couplers have similartransfer curves. Also, if your signals do not go to DC, any form of amplifier will work.
What I mean is a coupler that is not optimized for a linear transfer curve. They have a LED input and a PIN diode or BJT output. They are only guaranteed for output levels with two end points of the transfer curve range having greater than or less than values.
I forgot to say that you have to add a DC bias to the input signal to make it unipolar and then subtract the same DC bias from the output. A precision voltage reference can be used at each end of the signal path.
There are two other methods. Use a vco at one end and a frequency to voltage converter at the other. There are commercial products that use transformers.
I have used the vco method. The carrier was 5-10 times the highest input frequency and the deviation was +- the highest input frequecy. In FM terms, the beta was one. The sending end had a f-v converter just like on the receiving end and was used as feedback around the transmitting circuit. That way any systematic nonlinearities in the f-v circuit was "feedbacked" out.
The opto with feedback is worth a try since it is the cheapest.
If you are only using a few of these the commercial method would be cheaper and faster than the NRE. Burr Brown and others sell "isolation amplifiers" that can isolate a few kV.
A very expensive method would use an ADC followed by digital opto isolators on the data output and data ready lines. On the other side you would have a latch followed by a DAC. You could get 16 bit quality signals if necessary.