T&L Publications

Hi,
Does anyone know if an optocoupler can be used for audio line level? Or of a circuit that would allow this?
Thanks,
RT

Well...I'm no genius, but here are my ideas:<p>1) The thought of using a standard opto-isolator (or coupler) with an audio signal seems ideal. Flaw? The fact that an opto-isolator creates a digital 1 or 0. So, you can get your frequency correct, but not the amplitude of your audio signal.<p>Of course, an insanely complex solution (but it's all i can bear to think of during work) is to have a PWM driver/decoder circuit with the optocoupler in the middle. The PWM driver is a fast-acting sub-circuit that generates different pulse-widths according to fluctuations in voltage. On the other end, after transferring through your high-speed opto-isolator, the pulses are converted back to analog signals.<p>Is there an easier way? Sure. A/D conversion on the pre-isolated side and D/A on the post-isolated side of the circuit.

You're talking about optocoupling an audio signal (line level), right? A couple of questions:<p>1) How much do you want to spend?<p>2) How much distortion are you willing to live with? Is DC voltage important, or only the AC signal?<p>3) What's your upper and lower cutoff frequencies?<p>There are a lot of good solutions out there, but you have to provide more information.

By the way, Chessman's idea is not insanely complicated -- it's been done and it's on the shelf, except that instead of an optocoupler, it uses modulated/demodulated transformer isolation (which also, incidentally, provides an isolated +/-15VDC supply along with the signal). It's called the Analog Devices AD210, and it's available from Newark for about $70.00 in unit quantities.<p>AD210 Datasheet (.pdf format)<p>That module can be a lifesaver. It's got bandwidth (-3dB) to 20KHz with low drift and 0.012% distortion, and it's a genuine ace up the sleeve if you need it.<p>[ July 11, 2003: Message edited by: Chris Foley ]</p>

Opto couplers which have transistor output are fairly linear. The current gain is 1 or 2 (check the data sheet).

........isn't it crazy that I have an AD210 chip I can gratis you? I got a couple in a trade..and I always wondered what they were for... <p>of course, it's just the IC...

The Analog Devices AD210 is a potted epoxy package 2.1" X 1.0" X 0.35". If you've got one, save it -- you never know when you'll need it!<p>Russ is right. The least expensive method of analog optocoupling is simply to mix a bias current for the LED of a 4N28 or other optotransistor, say 12 to 16mA, with the line signal changed from 1VAC to 2mA AC input to the LED. Typical turn-on and turn-off times for the 4N28 are 2 microsec., which should be good enough for 20KHz. This modulated signal would be seen at the output optotransistor, which you would bias into the most linear part of its curve. One problem is the gain is kind of dependent on the 4N28 (time/temp drift), the DC level would be changing into the more non-linear areas of the curve, and there would be a certain amount of distortion (even if you set the Q point in the middle of the curve, you would still probably get a bit). Gain control would be a bigger problem than linearity, if your peak-to-peak signal at the output of the 4N28 were less than 10% of Vcc.<p>A better, but still only moderately expensive way to do this with optocouplers, is to have a second matched optocoupler which is used as a servo to keep the output co-ordinated with the input. You also need at least one good op-amp on each side of the opto to make this work. Clare makes a line of linear optocouplers (LOC-110, -111, and -112) and opto-isolation amplifiers (LIA-100, -101, with the op-amp built in) which are commercially available, go well past audio frequencies, and may be good enough for all but those with a "golden ear". The higher priced spread (LIA-100) sells in the Newark catalog for $8.99 ea. in unit quantities.<p>Clare Linear Optocoupler Page<p>Agilent has a line of linear optocouplers which use a delta-sigma a-d converter optically coupled to a D-A converter, which they claim has a bandwidth of 100KHz, with 5% gain tolerance and 0.1% distortion. Their chips are primarily made for motor control, but I understand they can be used for audio signals, too. I've never used one -- I wonder if anyone here has experience with them?<p>HPCL-7850 Datasheet in .pdf format<p>I hope this has been of some help.<p>[ July 11, 2003: Message edited by: Chris Foley ]</p>

Thanks for the replies. OK,here's what I'm trying to accomplish:
An XM sat.radio is mounted in my vehicle and the line level outputs are fed into a FM modulator which feeds the main sound system.<p>I want to use the XM sat line outs to also feed my low power FM transmitter so as to be able to recieve it on my boombox.<p>I've done this by simply using a Y adapter and it worked OK for a while,then I had to replace the sat receiver because the line out ground of the XM is not "vehicle ground".<p>So this is why I want to completely isolate the transmitter inputs from the XM receiver.As to how much do I want to spend,as little as possible.
I would much prefer to build a solution.<p>Thanks very much,
RT

I believe you generally use a ground loop isolator audio transformer for this problem. I think Radio Shack sells one for $14.99 (270-054). You just plug the line out RCA plug in on one side, and plug the line in RCA plug on the other. Self-powered, simple, and you'll probably get real good results. Here's a link from Electronic Circuits Schematics<p>Isolation Transformer For Audio Lines<p>As it says in the link, if you're desperate, get a 600 ohm : 600 ohm phone transformer, and live with an impedance mismatch and slightly lower audio quality. Happy hunting.<p>[ July 11, 2003: Message edited by: Chris Foley ]</p>

Yeah Chris,I did that already and used the Y on the output side of the GLI and it decreased the volume of the vehicle radio a noticeable amount.<p>I also tried coupling the transmitter inputs to this with various values of capacitors,but that affects the sound quality of the vehicle radio and the transmitted audio as well.<p>It looks like the Agilent or LIA-100, -101 may be the way to go.<p>chessman,I checked out the AD210 you and Chris refer to and I'd like to explore other solutions before going to that.<p>Thanks, please advise more if you can.

Wow. All this discussion and all Riptide really needs is an audio isolation transformer.<p>Goes to show that it's good to find out what people are trying to accomplish whan they ask a technical question.<p>Bob

To paraphrase the Oracle,<p>"What's really gonna cook your noodle is when you ask yourself why people answer them..."<p>By the way, when you use a "Y" from the line out (nominal 1K ohm) to two line loads, you have in effect a 500 ohm load. The impedance mismatch will mean that you will lose 33% of your signal. It won't kill the electronics, but you might adjust by just turning up the volume, or using a preamp. Happy hunting.<p>[ July 11, 2003: Message edited by: Chris Foley ]</p>

BTW,an audio isolation tfmr was the very first thing I tried,both on input and output of the GLI and without the RS GLI.
As far as turning up the volume,the thing with that is the high freqs are attenuated noticably.<p>I didn't think about a preamp though.<p>As to the "all this discussion",I think most of the questions and discussions thereof are very interesting and often trigger new ideas on how a problem(even if not the same problem) can be solved.<p>One reason I subscribe to N&V is the fact of working out solutions with custom built circuits or modifying equipment rather than simply buying something off the shelf like any numb nut can do.I still would like to have the second output totally isolated and will continue with the ideas most of you have presented.<p>Thanks alot,
RT<p>[ July 12, 2003: Message edited by: Riptide ]</p>

Less expensive audio isolation transformers (such as the one from RS) do have the tendency to drop off at the high and low ends. You know, one of the things you might want to try is to build a preamp/buffer (as simple as using a dual audio quality op amp to create two voltage followers for impedance matching. I like the idea of trying to optoisolate the analog line signal, even though it's not the shortest distance from point A to point B. After all, the journey is part of the trip as well as arriving at the destination.<p>Good posting. Let us know what you come up with, RT, and if you need any more advice.