Eliminating series resistor at line output?

[Omega_Void]

On most schematics I've seen, there's normally a resistor 100R in series with the line output. I believe this is to swamp the effects of capacitive cables.

Is it possible to achieve the same effects without having any resistance in series? The intended output may need to drive fairly long cable runs at times, load capacitance will vary.

I have no oscilloscope to work with, I'm afraid. I can observe up to 48kHz using my PC soundcard, and I have built Uncle Jed's oscillation detector described here.

Is it possible to dispense with the resistor without fear, or must I live with it?

Many thanks.

[Bonsai]

Yes. Assuming a non-inverting design using an op-amp:-

1. take your feedback point after the output series resistor
2. place a small value cap (usuallyfew 10's of pf) between the inverting input and the output before the series resistor.
3. Select the cap by driving a squarewave into the worst case expected capacitive load and then adjustinmg the cap to ensure no oscillation with minimal rise time degradation.

This technique can also be applied to discrete designs. You may not use this technique on CFA's or amplifers that are not unity gain stable.

For most applications though, I would just stick with the series resistor - 50 Ohms is usally quite enough.

[nigelwright7557]

I would leave it in.
It also protects against short circuits on the output.

[DF96]

The resistor protects from shorts, and also isolates the output from the cable capacitance so that stability is not compromised. You could argue that it is most needed for long cables, as these will have the highest capacitance. If your cable is too long then you need to add a buffer stage, or redesign the existing output stage.

A 100R resistor can drive 53nF up to 30kHz, which is a lot of cable! For a 1V signal at 10kHz this would need 3.33mA of drive current (4.7mA peak) so in many cases the output stage would run out of current drive before the resistor became significant.

[Omega_Void]

Thanks for the responses.

@ Bonsai: I think I understand what's happening. LTspice says it works anyway.

Output current is limited in most opamps anyway (OPA2134, LM833, TL071 all indicate this on the datasheet), so a short shouldn't be fatal should it?

[AndrewT]

Originally Posted by Omega_Void

Output current is limited in most opamps anyway (OPA2134, LM833, TL071 all indicate this on the datasheet), so a short shouldn't be fatal should it?

I don't know how opamps are protected, but I suspect they will run hot if the short output is maintained while signal at the input is still live.

This may not be a concern with an opamp operating from supply rails @ 25% of maximum, but could be very different if operating @ 90% of maximum.

[DF96]

As the resistor does so much good, and so little harm, why remove it? For those people who, totally unnecessarily, worry about transmission line impedances at AF the resistor will even act as a partial back-termination so it is good for both science and voodoo.

[Omega_Void]

Originally Posted by AndrewT

I don't know how opamps are protected, but I suspect they will run hot if the short output is maintained while signal at the input is still live.

I might try that and see over the weekend.

I have a ±15V 10A regulator. A couple of 6A fastblow fuses for the rails should protect that and still give the opamp enough to fry itself, no?

I've got enough spare TL072 to risk killing one in the name of research.

[Omega_Void]

Originally Posted by DF96

As the resistor does so much good, and so little harm, why remove it?

I'm experimenting with grunging up the sound using some little mains toriods I picked up cheap. Yup, this ain't for hi-fi.

One possible configuration is a step-up with 50:1 impedance ratio. Means my 10k line inputs look like 200R.

Tried it, I like the sound, TL072 seems happy driving it (signal voltage is very low), but the 100R on the output means I don't get maximum benefit from the step-up.

So if I can ditch the resistor without sacrificing reliability it keeps my hookup options open.

I could just buffer the output after the transformer, but since so many balanced inputs are not transformer-coupled (like my soundcard for example) this would lose the loop-breaking advantage of a transformer-coupled output.

[DF96]

A transformer will also, to some extent, transform the capacitive reactance although leakage inductance and transformer resonances will confuse the issue.

I have never tried to deliberately degrade sound so I can't really give you any advice about that!