Op amp output protection resistor

  Hi,
To protect an op amp's output from a cable capacitance, we usually put an output protection resistor (around 47ohms) to prevent HF instability.. Am I right?

  My question is, in a piece of gear, if an op amp's output goes to a pot or whatever through a 1ft shielded cable (more or less) should we put an output protection resistor to the op amp or buffer?

Thanks

It's good practice but do you really need shielded cable to protect a low Z source?

JR

I might be totally wrong, but the idea is to prevent crosstalk between several low Z sources..

saint gillis said:

I might be totally wrong, but the idea is to prevent crosstalk between several low Z sources..

Click to expand...

Crosstalk between cables is a function of the impedance, so a low impedance feed, like from an op amp output should not pick up much.

For crosstalk inside a console, a more likely source of problems is from grounds getting corrupted.  Say if left and right faders dump current into a single ground node, the resistance of that ground will allow a build-up of L+R signal...

Use of differentials can manage this kind of crosstalk.

For sending signals over distances much much longer than 1' using balanced terminations and differential receivers helps.

JR

Ok, so for small pieces of gear I'd better use unshielded cable and no output protection resistors at buffers' end?
  And for stereo crosstalk, considering separate grounds for left and right paths or does it seem overweening ?

saint gillis said:

Ok, so for small pieces of gear I'd better use unshielded cable and no output protection resistors at buffers' end?

Click to expand...

I like the idea of still using a small resistor maybe 10-20 ohm.

  And for stereo crosstalk, considering separate grounds for left and right paths or does it seem overweening ?

Click to expand...

There are different ways to skin any cat, and I try to resist broad sweeping generalizations.  If you use separate grounds for left and right you connect them together at some point?

I prefer to think about signals as two parts, namely the signal with respect to some reference.  We ASSume the signal is clean at the op amp output wrt the op amp local 0V reference.  At the pot the 0V it gets connected to won't be the same 0v as the op amp.  The current flowing through the pot and into the 0V trace impedance will cause a voltage.

However if the following stage takes it's signal from that pot differentially  taking a + signal feed from the wiper, and - signal input from the bottom of the pot where it connects to 0V any voltage on that 0V  node will be common mode and cancel out.  Using a differential like that means any number of pots could connect to the same 0V reference and any noise there is automatically ignored.

Warning- this is one of those sweeping generalizations and you can go crazy chasing minute amounts of crosstalk out of a large console.  For a smaller simpler design brute force might work.

good luck

JR

Thank you!

One last question , is it still better to use shielded cable or is it completely useless?

saint gillis said:

One last question , is it still better to use shielded cable or is it completely useless?

Click to expand...

It depends on the impedance of the signal... Something like the wiper of a fader is not low impedance so could benefit from shielded cable, while the low impedance drive to the top of the fader probably doesn't need it, while it's easy enough to just share some 2 conductor shielded.

JR

JohnRoberts said:

saint gillis said:

One last question , is it still better to use shielded cable or is it completely useless?

Click to expand...

It depends on the impedance of the signal... Something like the wiper of a fader is not low impedance so could benefit from shielded cable, while the low impedance drive to the top of the fader probably doesn't need it, while it's easy enough to just share some 2 conductor shielded.

JR

Click to expand...

Sorry, but in that case you are not using your preferred 4 conductor fader connection, to get better kill level... 3 conductors are more hard to find, maybe 4, or dual single conductor as RCA shielded wires so one for input one for output.

Also in some very sensitive cases the better way of doing it is using the shield connected in one side, just to shield. I'm talking of very high impedance sensitive applications, mostly with tubes, things like a mic preamp or high gain guitar amp may show some benefits of doing so.

JS

joaquins said:

JohnRoberts said:

saint gillis said:

One last question , is it still better to use shielded cable or is it completely useless?

Click to expand...

It depends on the impedance of the signal... Something like the wiper of a fader is not low impedance so could benefit from shielded cable, while the low impedance drive to the top of the fader probably doesn't need it, while it's easy enough to just share some 2 conductor shielded.

JR

Click to expand...

Sorry, but in that case you are not using your preferred 4 conductor fader connection, to get better kill level... 3 conductors are more hard to find, maybe 4, or dual single conductor as RCA shielded wires so one for input one for output.

Click to expand...

I was trying to KISS to not confuse the OP.  Typical fader wiring generally keeps everything within a multiple conductor shielded cable, but this does not inform the OP much about what to shield.  A subtle point, shielding  the drive to a fader may prevent it cross-talking into other sensitive paths.

Also in some very sensitive cases the better way of doing it is using the shield connected in one side, just to shield. I'm talking of very high impedance sensitive applications, mostly with tubes, things like a mic preamp or high gain guitar amp may show some benefits of doing so.

JS

Click to expand...

Yes in sensitive circuits the conductor to shield capacitance can be an issue.  OP was asking about op amps, while care must be exercised for any shielded cable used in series with op amp (-) inputs.  For that specific case, connecting the shield to the op amp output would throw the cable capacitance across the op amp negative feedback network, making it more stable not less.

JR

JohnRoberts said:

Yes in sensitive circuits the conductor to shield capacitance can be an issue.  OP was asking about op amps, while care must be exercised for any shielded cable used in series with op amp (-) inputs.  For that specific case, connecting the shield to the op amp output would throw the cable capacitance across the op amp negative feedback network, making it more stable not less.

JR

Click to expand...

Now I'm beginning to get fundamental questionings : "Why is capacitance bad at an op amp output?"

saint gillis said:

JohnRoberts said:

Yes in sensitive circuits the conductor to shield capacitance can be an issue.  OP was asking about op amps, while care must be exercised for any shielded cable used in series with op amp (-) inputs.  For that specific case, connecting the shield to the op amp output would throw the cable capacitance across the op amp negative feedback network, making it more stable not less.

JR

Click to expand...

Now I'm beginning to get fundamental questionings : "Why is capacitance bad at an op amp output?"

Click to expand...

Because real op amps have real output resistance. A cap to ground there creates a one pole low pass filter. Phase shift (lag) associated with this filter will make the op amp less stable.

At very high frequency  if phase shift lag in the NF path exceeds 180' before the open loop gain drops below unity, the negative feedback turns to positive feedback and oscillation occurs. 

JR

JohnRoberts said:

...
Yes in sensitive circuits the conductor to shield capacitance can be an issue.  OP was asking about op amps, while care must be exercised for any shielded cable used in series with op amp (-) inputs.  For that specific case, connecting the shield to the op amp output would throw the cable capacitance across the op amp negative feedback network, making it more stable not less.

JR

Click to expand...

I like that! So I guess for both possibilities at the opamp (-) input, being going to ground or going to the output of the opamp, I guess is good practice to shield in reference to the output of the opamp, in both cases it will be a capacitive load to the output but usually small enough to not being a problem plus adding the extra capacitance in the feedback to improve stability.

JS