Analog opto isolator

Hi,

I want an 1:1 opto isolator for the voltage ranges from 0 V to 3.3 V.

Say if 1.2 V is given to the input of opto isolator, i need the same 1.2 V from the output.


Is there any ready made IC available?

or pls suggest a method to do this by using the conventional digital opto isolator..

Hi,

This has been discussed several times here. Do a forum search.

there are "analog optocouplers" with internal monitoring diode for feedback. (like HCNR200)
They usually work with "current". For your voltage_input to voltage_output you need a circuit. There is one in the HCNR200 datasheet.

***
Digitally:
* A sawtooth or triangle signal.
* A comparator to compare input signal with triangle signal. Output is PWM with duty cycle according input voltage.
* Feed to optotcoupler input.
* Output of Optocoupler needs to be low pass filtered to get DC analog voltage.

Klaus

Accurate analog to analog isolation is somewhat surprisingly tricky.

Tell us more about the application and we can make better suggestions. What's the signal and bandwidth requirements. Do you really need analog to analog? Do you already have any digital isolation in the circuit? Do you have isolated power on the secondary of the circuit?

For example if you're going back to a digital system its probably easier to go analog->digital->isolation for example.

If you already have a digital isolator with an extra channel LTC6992 could make a sense by converting analog into a PWM that can be sent through a single digital isolator channel.

For straight analog to analog look at ADUM4190.

For elaborateness it'd be hard to top a module I was
recently asked to bid on reverse engineering - a
20-bit delta-sigma ADC, an opto for the isolation
and a 20-bit DAC with reconstruction filter on the
other side.

Below is a slight modification to a design I did a long time ago using two common opto's to achieve isolation. The original design was for use in an isolated 4-20 mA loop. It performed remarkably well with good linearity. It is not a very fast circuit and bandwidth will be limited to about 10 kHz.

Today there are much easier solutions available, as other members mentioned.

Another option that can be considered is the IL300, for which there are even simulation models available for the LT(Spice).

Here's the LTspice simulation of a linear opto coupled circuit using a standard opto isolator.
It should work for you if your accuracy requirements are not too stringent.
It converts the input voltage to a current at the opto input, what makes the input to output fairly linear.
Its frequency limit is basically determined by the op amps' response times.
The pot in the output op amp's feedback allows gain adjustment to compensate for differences in the opto coupler transfer function gain.

The problem with these types of circuits is that at some low Vin, the opto stops functioning due to insufficient led current. As a result, the output just drops to zero. With my circuit, it happens with an input below 50 mV, which cause the led current to go below 200 uA.
They work best where there is a minimum current flowing with 0 V input, like in a 4-20 mA application.

I'm thinking about this problem this morning. I always have voltages to move across isolation barriers but often its a nice-to-have rather than hard requirement so I'm always searching for smaller/cheaper ways to do it so I can justify it in more places.

Anyway this appears to be a decent compromise of complexity for accuracy compared to say LTC6992. It's a variation on a relaxation oscillator set-up to operate in a psudo-regulated hysteretic mode where Vin equates to duty cycle out. It uses just a single comparator or op amp.

So the output of the opamp/comparator is a square wave that's suitable for transferring through any number of isolation options (not shown) including opto, digital isolator or pulse transformer. The variable PWM is nice because it can be easily sent to an FPGA/micro directly or just as easily filtered to recover an analog signal (the two pole RC shown simulates that).

The linearity test compares the input to the recovered output. It could be linearized further with another opamp stage although there are diminishing returns because the isolation method with add at least a couple percentage points of distortion.