Isolating a Push-Pull Output

I have never designed anything like this but what about 2 opto couplers, back to back, one for the positive and the other connected for negative signals?

Or just boost the DC bias so that you have in the end a positive (or negative) signal that stays always positive (or negative), but varies by 20 volts?

It isn't simple unless you can get away with low precision and some non-linearity.

How accurate does it need to be? How many volts isolation do you need? Do you have a local floating power supply or do you need to create one with an isolated dc-dc converter? What is output current? What is the frequency range?

Analogue optical couplers are simple but non-linear.

For high precision I would be inclined to digitise the signal and use opto-couplers. For lower precision a pwm system through opto couplers might suffice.

If I were doing this I would investigate using a Hall effect transducer with an amplifier but I have never tried it.

Some important informations are missing for a practical solution:

1. Is it an analog or a digital signal?

2. If analog, what`s your requirement in resolution resp. accuracy? What about environmental conditions like ambient temperature changes ?

3. If analog, which bandwith? If digital, how many bits/sec to be transferred?

4. What kind of load (How much resistance, inductive or capacitive load components?

5. Is it ONE signal of differential nature with a total swing of -20V to +20V or do you have TWO signals each against ground?

Independant of analog or digital, with a simple OP-based differential Amplifier you can convert your bipolar signal into a unipolar signal referenced against ground and beeing able to drive an optocoupler only in one direction.

If you have an analog signal this can be easily transferred via an analog optocoupler together with an OPAMP provided that the ambient temperature changes are not too high because these solutions may be quite drifty. The accuracy you get can will be about 0.1% and the analog bandwith may be up to 500kHz or so, mainly determined by the OPAMP.

If it is a digital signal, have you thought about using one of these RS232-converters with internal isolation barrier ? Your +-10V are wide enough to drive an RS232-receiver and you get a potential-free TTL-Signal out (isolation voltages up to 2500V) whith which you can do whatever you want on the load side.

Regards Uwe

GA Uwe.

Here's some more detail and thanks for your comments and ideas.

What i'm driving is a 16ohm coil. When the coil is polarised in direction A it pulls a paddle that runs a hydraulic pilot. Conversly when the coil is polarised in direction B it pulls the paddle the opposite direction.

My PLC output is so filtered that for all practical purposes it's analog. Remember, it's only two wires. The only thing i can think that's going on, and this is only from other controllers of other brands, that whatever of the two pins is OFF, becomes ground for the ON pin.

.625amps is within the operating spec of the output but my cable length is long and there's a likelyhood all kinds of intereference from other systems. I use shielded cable, and that works for what it's intended but cutting or pinching a cable while under load happens too. I hear someone saying that's a design/planning problem. Planning and reality many times are two differnet things. Hence the need for isolation.

The linearity of the ouput is not a concern because the output is part of a proportional closed loop with feedback.

Evidently i need some sort of switching that senses if one of the outputs goes below a specific low positive voltage in addition to optocoupling.

The purchased isolation amplifier, i'm quite sure, digitizes the signal and reamplifies it, and spits it out nice and linear.

Thanks again

You are supplying the power for the coil over the wiring, is it possible to use the wiring for just control and have a local source of power?

What about V/F converters? Convert input to frequency, isolate it with an optocoupler then reconstruct voltage with a F/V.

U can use say LM331 for this purpose.