Analog signal isolation

[Silenus]

I'm designing a circuit that reads bioelectric signals from the body and then feeds this signal into an MCU for processing. Currently I am working on the isolation of the device. I have a high isolation dc/dc converter to power all the active filter op amps. I have read that it is a good idea to isolate the signal via an optocoupler however I am not sure how I would achieve this. With an optocoupler am I able to feed in an analog signal and have it mimmic the signal on the output? What sort of optocoupler circuit would achieve this?

[coppice]

I'm designing a circuit that reads bioelectric signals from the body and then feeds this signal into an MCU for processing. Currently I am working on the isolation of the device. I have a high isolation dc/dc converter to power all the active filter op amps. I have read that it is a good idea to isolate the signal via an optocoupler however I am not sure how I would achieve this. With an optocoupler am I able to feed in an analog signal and have it mimmic the signal on the output?

Yes.

What sort of optocoupler circuit would achieve this?

Any opto-coupler which is a pure opto-coupler, and doesn't contain extra circuitry to turn it into a logic device. Opto-couplers can operate as analogue devices over a large dynamic range. That said, in most cases you are better off putting the op-amps and MCU together on the isolated side, and feeding the digital results from the MCU through an isolator.

[Silenus]

Any opto-coupler which is a pure opto-coupler, and doesn't contain extra circuitry to turn it into a logic device. Opto-couplers can operate as analogue devices over a large dynamic range. That said, in most cases you are better off putting the op-amps and MCU together on the isolated side, and feeding the digital results from the MCU through an isolator.

So say if I fed this signal into my computer, it would isolate the computer's ground from the ground of the signal acquisition hardware?

[Richard Head]

Use an IL300. It's designed for transfering analog signals across an isolation barrier.

[Silenus]

If I isolate at the digital side, I was thinking of going with a ISO7220xD instead of the optocoupler.

[dannyf]

Must easier to convert the analog signal to digital locally before the isolation.

[Richard Head]

Must easier to convert the analog signal to digital locally before the isolation.
Agreed.

[Gall]

Must easier to convert the analog signal to digital locally before the isolation.

That is exactly what HCPL-7800 does. This is a digital optocoupler with integrated ADC and DAC.

[macboy]

I'm designing a circuit that reads bioelectric signals from the body and then feeds this signal into an MCU for processing. Currently I am working on the isolation of the device. I have a high isolation dc/dc converter to power all the active filter op amps. I have read that it is a good idea to isolate the signal via an optocoupler however I am not sure how I would achieve this. With an optocoupler am I able to feed in an analog signal and have it mimmic the signal on the output?

Yes.

What sort of optocoupler circuit would achieve this?

Any opto-coupler which is a pure opto-coupler, and doesn't contain extra circuitry to turn it into a logic device. Opto-couplers can operate as analogue devices over a large dynamic range. That said, in most cases you are better off putting the op-amps and MCU together on the isolated side, and feeding the digital results from the MCU through an isolator.

well, no, an analog optocoupler is not simply a "non-digital" optocoupler. Many optocouplers are not strictly speaking digital; they are simpy a LED on one side and a photo-transistor on the other. They can be made to operate in the "analog" domain but will not be linear.

This is where true "analog" optocouplers come in. The are highly linear by design. They achieve this by having two very well matched photo-transistors. One is used for feedback in a closed loop op-amp circuit which drives the LED, so that linearity can be achieved. The other is used on the isolated side. The theory is that the second one will track very closely the first one, which is forced to be linear by that negative feedback.

[tomizett]

If you're after a simpe aproach, take a look at this app note:
http://www.avagotech.com/docs/5954-8430E
I first came accross this taking apart a Harvard Instruments medical/laboratory isolation amp - they use the Fairchild HCPL2531 opto.

The idea of doing your A/D conversion on the "remote" side is definately attractive though, if you can take the extra complexity.