Audio line out driver

[jimon]

As normal audio lines are usually ground+signal, will "just-an-op-amp-buffer" will be sufficient? Or do I need something specialized like http://www.ti.com/product/DRV135 (though it's balanced line driver)? Thanks!

[SeanB]

Yes, just use a series resistor and a terminating resistor on the other end. Series resistor because many opamps are not too happy with a large capacitive load ( the line), and terminating resistor, which does not have to be line impedance, reduces noise pick up if the line is not connected. If using symmetrical power rails coupling capacitors into the line and at the other end are optional on the output, but just be aware on the input you need to have a DC path for input bias currents of the receive side.

[jimon]

And if line is accepted by other audio equipment ... I guess they already do termination and some coupling?

[SeanB]

If you are making a general purpose line output use a 47-100R resistor as the series resistor, and a coupling capacitor, 100-220uF 25V or so. You cannot be sure that the output will be AC coupled, and might be putting a bias current onto your output. As safety just use a 10-22k resistor after the capacitor to ground.

This has been used by literally millions of amplifiers, all the way from having a 741 as output ( still used these days, though more often a dual clone) to the most common being a TLO74 or clone, or a NE5532 as the opamp.

If single supply the coupling capacitor and resistor afterwards is really needed, you can not be sure the other side is not AC coupled as well, and might object to DC being applied.

If you saw the teardown Mark ( Dexterslab13) did on a Sony Betacam player you will have noticed the output stage for audio had some failed capacitors, and I would guess all of them blew for the same reason, they were connected to a mixing deck input, and somebody turned on phantom power to the input, giving 48VDC to the capacitor in reverse. Simple fix is to use 2 470uF 63V capacitors in series back to back as a coupling capacitor, which will survive that for decades with no problem. Sony did not do that, instead using the single 470uF 35V capacitor instead, and connecting it so positive was on the player side, so it was reverse biased with the accidental application of phantom power.

A rough industry standard ( if such a thing exists) for audio is that you need 1V peak of signal, into a load impedance of 1k.  For professional audio it is 0.7VRMS into 600R, so design your output to give either ( they are about the same, just the first is not really controlled in anything other than voltage) but ensure it can drive a 600R load without distortion. That the signal level varies with load impedance is not a worry much, there will be some volume control somewhere in the chain.

Best to design that the clipping level is more than 2VRMS into the 600R load, so the opamp must be able to drive a 600R load with 3VRMS output. If you only have 5V rails, you will not achieve this so try the best you can, using an opamp with rail to rail output capability.

[jimon]

So I guess something like this will do! Thanks!

[SeanB]

Yes, for most cases it will work fine. Put gain ahead of it and leave the buffer to only have to deal with current amplification and save the gain bandwidth for that. If you are tight for space you can use it to give a little gain, but keep it under 2, and beware of some opamps that are not unity gain stable.

[linux-works]

in theory, unity gain is ok but in practice, you would need to find the right op-amp.

I prefer to avoid unity gain and at least have 1k to 10k in the feedback path.  things tend to be more stable that way, in my experience.

the other common thing is to have a buffer after the op-amp.  the OA will give you voltage gain and the buffer will give 'current gain' or 'drive'.  modern designs often will include the buffer IN the feedback loop.  see the 'pimeta' design and ones like it, for examples.

if tangentsoft's website is still up, its a good place to start.