How low output impedance is too low?

[JimRayden]

...and what starts happening when it's below that limit?


----------
Jimbo

[Paul Perry (Frostwave)]

You can never be too rich, too thin, or have too low an output impedance. If you are talking about stompboxes, anyway.THhe only disadvantage to an ultra-low output impedance is this.. if the output is shorted to ground, how much current flows? Will it damage anything? When you look at op amp specs, you see some are guaranted for short cicuited output, but some aren't. In which case, you might care to bump up the output imprdance with a 1K resistor.
Am ideal op amp has zero impedance output.
Power amps are another story entirely.

[R.G.]

As Paul notes, for signal purposes, the correct number is 0.00000000000000000000000000000000000 ohms, but for safety, a lower practical limit makes sense. I have grown to liking inserting a 1K or so in series with the output of an opamp that goes off the circuit board. A shorted wire then does not take out the opamp even if the opamp's protection doesn't work well.

Something between 100 ohms and 1K is good, unless it's a power amp. For power amps, I like to couch these numbers in current ability. I like power amps with output stages with current abilities in the 30A to 100A region. Driven properly, you should be able to arc weld with the output of an amplifier without damaging it. Haven't seen one of those, but I think it's a good goal. 

[JimRayden]

Thanks for the replies.

Anyone know a good output impedance for a tube preamp? I'm making a tube preamp/OD effect and am designing the switching circuit between the two modes.

----------
Jimbo

[R.G.]

Unless you have done something special, the output impedance of a tube circuit may be taken to be the value of the last resistor actually connected to a tube that the signal comes out of. For instance, a typical triode gain circuit with a 100K plate resistor may be considered to have an output impedance of 100K. A cathode follower with a cathode resistor of 47K may be taken to have an output impedance of 47K.

That's not strictly true, of course. The plate resistance of the tube gets into it, and so does the overall feedback, if any, and the gain of the tube, that kind of thing. But those will tend to lower the value, so this approach will yield conservative results.

Another way to look at it is that tube circuits, excepting power amplifiers, should never be expected to drive an input impedance of less than 100K.

[puretube]

A cathode follower with a cathode resistor of 47K may be taken to have an output impedance of 47K.

?

I`ve been taught: Z = 1/S (S=transconductance)...

[R.G.]

I`ve been taught: Z = 1/S (S=transconductance)...

That's true - pulling up. A cathode follower can't pull down, the resistor pulls down, so the downward impedance is different from the upward impedance for any except quite small signals. This can be fixed to a degree by using an active load instead of a resistor for the follower load, so the active load can pull down as hard as the followeris pulling up.

As I said, "That's not strictly true, of course." I was quoting rules of thumb, with conservative results.

[Elektrojänis]

I like power amps with output stages with current abilities in the 30A to 100A region. Driven properly, you should be able to arc weld with the output of an amplifier without damaging it. Haven't seen one of those, but I think it's a good goal. 

Haha...

Q: "Thats a smooth looking joint. What did you use to weld it?"
A: "AC/DC, Led Zepplelin and some jazz standards for the finishing touch."

[seanm]

As Paul notes, for signal purposes, the correct number is 0.00000000000000000000000000000000000 ohms, but for safety, a lower practical limit makes sense. I have grown to liking inserting a 1K or so in series with the output of an opamp that goes off the circuit board. A shorted wire then does not take out the opamp even if the opamp's protection doesn't work well.

Interesting.  So you mention sticking a 1k resistor in series if it is an opamp. How about a FET based output, like, oh I don't know, a muamp?

And would the answer change if you where going straight into the power amp? I am asking because I want to use a mini-booster as a preamp and currently have a 100k resistor to ground on the output. But I notice that the power amp has a 10k input impedance and the 100k is probably way too large. But I would also like the mini-booster to work as a booster.

[davebungo]

I would have thought you would be OK with a discrete FETmu-amp or for that matter most similar looking configurations because there is already an intrinsic output impedance which at AC is a combination of the load R and the output resistance of the device itself  (I can't remember the equations but I'm sure someone will volunteer).  Most pedals though usually have a low impedance driver because you can't guarantee exactly what you will be feeding i.e. short or long cable, high impedance guitar input or low impedance line input etc. and it is generally this which needs some protection whether it be a series R or an in-built output impedance.

[R.G.]

Dave is correct. You have to take into account the intrinsic output impedance of the device that drives the output. in the case of a mu amp, the output impedance is modestly high already. A short circuit or worse, a low impedance, won't hurt it because the impedance is so high already. An opamp has a fairly low output impedance plus the feedback lowers it more. Although most opamps these days have short circuit protection, the 1K helps a lot for strange loads and for incoming transients.

And would the answer change if you where going straight into the power amp? I am asking because I want to use a mini-booster as a preamp and currently have a 100k resistor to ground on the output. But I notice that the power amp has a 10k input impedance and the 100k is probably way too large. But I would also like the mini-booster to work as a booster.

It depends on what you mean by "straight into the power amp". If they're in the same box, not unpluggable, you don't need the protection against transients and the mu amp will never be shorted. However, you're onto a good point with the input impedance of the power amp. If it's 10K, that's a fairly tough load for a mu-amp. It would be better to put a buffer after the mu amp to prevent it from being loaded down and having its gain lowered by the loading. The exterior pull down resistor has almost nothing to do with it. It's the INSIDE impedance that is important, not what we hang on the outside. The outside 100K pulldown is just another load to the mu-amp.

[zachary vex]

...and what starts happening when it's below that limit?


----------
Jimbo

it depends on what you intend to feed with your effect.  for example, many octave devices sound great direct into fuzzes... in this case, the octave device should have an output impedance that's relatively high so the fuzz will enjoy the signal... perhaps 10k-25k or more.  this is easy to do... use a 50k output pot and when it's parked halfway up you'll get the impedance you need.

it's madness to use very low output impedances to drive guitar gear.  it's a disease perpetrated by people who think with the technical half of their brain... low-impedance sources tend to interact with guitar gear in negative ways.  sorry Paul!  8^)

[R.G.]

it's madness to use very low output impedances to drive guitar gear.  it's a disease perpetrated by people who think with the technical half of their brain... low-impedance sources tend to interact with guitar gear in negative ways.

Kewl... other than the high impedance loving fuzz face circuits and some wahs, do you have any concrete examples of the madness? And how some of those negative ways happen?

The techical half of my brain makes me list a few counter examples:
- The lower the impedance that drives a cable, the less treble tone sucking the cable can do. It's the mirror image of needing quite high input impedances to eliminate tone sucking. You can eliminate it on the driven end of the cable instead of the driving end.
- Triodes grid-clip more softly the lower the impedance they're driven from. 
- Pentodes in output stages produce more power and softer overload when driven from followers directly on their grids.
- The lower the impedance that drives a transformer, the less core distortion you get and the wider the bandwidth it will produce.

To my way of thinking, it's madness not to know and understand all of what input and output impedances both can and may do and be able to do either or both as needed.

[Phorhas]

So driving guitar gear with a 600Ohm DI/line driver is no problem, yes?

I thought so anyways

and while on the subject, where can I read some info on o/p impedance of OAs, how do I determine exact numbers and such (yeah... I neglacted that subject a bit - learning by myself is a PITA)...

Regards,
Dan

[R.G.]

So driving guitar gear with a 600Ohm DI/line driver is no problem, yes?

Yes, it's no problem, given that you pad the voltage level down to where the guitar gear isn't continually blasted with massive signal levels. A 600 ohm source will NOT harm the input of your tube amplifier.

where can I read some info on o/p impedance of OAs, how do I determine exact numbers and such

That's hard to find. The problem is that the o/p impedance of a high feedback amp is usually so low that it can be assumed to be zero at all currents right up to where the opamp current limits (modern opamps) or dies (older, fragile opamps). The output of most opamps is a class B emitter follower, complementary, or quasicomplementary follower, so the no-feedback output impedance is already pretty low, maybe a few ohms, well under 50 ohms. Adding feedback lowers the output impedance by the feedback factor. If the amp has an open loop gain of 100K (and that's low for IC opamps), then a gain-of-ten stage uses up only a factor of ten of that 100K. The other 10K of the gain goes to making the output match the input, so the output impedance is lowered by that factor, or 1/10K of what the open loop impedance is.

It's simpler to talk about their current limits. The datasheets give you the output current limits, and whether the amps are short circuit protected or not. The opamp will deliver its max output current into any load impedance that generates a voltage from that current less than the high level output swing of the opamp.

For instance, if the opamp data sheet says that the amp will deliver 10ma max, and the output will swing to within 2V of the power supply, then for a +/-9V supply, then a 7V/0.01 = 700 ohm resistor will just make the amp current limit. For loads bigger than that, the amp will voltage limit before it current limits. For resistances less than that, the amp will current limit before it voltage limits.

Fortunately, the exact number of output impedance of an opamp almost never matters. If it does, there is generally another, better way to do whatever you're trying to do.

[Phorhas]

Thanks R.G

So if I take a moderd audio OA like an NE5523 or LM833 with an open loop I can just stick a 620Ohm resistor in series with it's output and I have DI line driver. then puting a switchable pad of 6dB isn't an issue - just a resistor to ground. kewl.

It's better to use a rail-to-rail output AO for such use, right ? (I can't use mor than 9-12v)

[R.G.]

So if I take a moderd audio OA like an NE5523 or LM833 with an open loop I can just stick a 620Ohm resistor in series with it's output and I have DI line driver. then puting a switchable pad of 6dB isn't an issue - just a resistor to ground. kewl.

Shouldn't be. Both the 5523 and 833 are designed to drive 600 ohm loads. It depends on how much voltage you want them to drive. The outputs still have that voltage limit to the power supply, so they won't go all the way to ground or power.

It's better to use a rail-to-rail output AO for such use, right ? (I can't use mor than 9-12v)

Depends on what you're driving, and whether it really needs driven all the way from ground to power supply.

There are some second order considerations. If you're using this for a comparator, you typically damage the inputs of the opamp by exceeding either the input common mode range or differential mode range on the inputs, so the opamp input degrades with time. You can prevent that with resistors to the inputs and diode clamps in most cases. But using a comparator gets you immune inputs and rail-to-rail performance.  Generally using an opamp open loop means you're using it as a comparator.