DC coupling advantages/disadvantages?

[earthtonesaudio]

Assuming first that you're using a bipolar supply and the signal is referenced to 0V (ground), it seems like there are a couple advantages to direct-coupling your signal.  Low frequency response goes down to zero Hz (DC) and of course no capacitor means one less part to buy, drill holes for, solder in place, etc.  Any other benefits?

As for disadvantages, most obvious to me is that a DC coupled input is vulnerable to sources that have a non-zero DC level on their output.  As for DC coupled outputs, I can't think of too many disadvantages.  Possibly short circuits to some non-zero power supply rail?  Are there any other reasons not to do this?

Having "grown up" in the stompbox world, my experience is based around the single 9V battery.

[puretube]

Rumble...

[Cliff Schecht]

There are some big advantages with DC coupling. The obvious one, bandwidth, is only so relevant to music applications. There is some stuff that NOBODY, no matter what they claim, can hear (the human ear only goes up to about 22 kHz, and that's a good ear!) Also, many people claim that passing a signal through a capacitor will change how something sounds no matter how high quality of a cap they use. I personally think this is bullsh**, but DC amplifiers do serve their purpose. There is a definite challenge in designing one properly because you have to take into account not only the amount of output voltage one stage is feeding another, but also how much current and how the small signal "wiggle" will affect said bias. Very fun stuff IMO! I haven't played around with tube DC amps (although I've studied them), but transistor based variants are just as challenging to work with.

[Sir H C]

Most op-amps you see are DC coupled high gain amplifiers.  They have to be as the caps would be huge in there.

[R.G.]

While there do exist a very few types of amplifier that are inherently not DC amplifiers, they are esoteric enough that I suspect that you can't name one. Other than those few, *all* electronic amplifiers are DC amplifiers. The ones that seem not to be just have caps blocking their inputs and/or outputs.

So if the question is about advantages and disadvantages of not using a cap to block DC at the input and/or output (that is, leaving them DC coupled), then:

Advantages:
- response to 0Hz
- lack of phase shift at low frequencies
- fewer parts, etc. as you've noted.

Disadvantages:
- response to 0Hz; this is what Ton was trying to be cryptic about, letting in low frequency noise. In turn tables and all other mechanical sound playback stuff, this comes through as rumble. In general, you can also have problems with flicker or 1/F noise, and with having offsets that can drift your output devices out of range. Speakers are DC response devices, but they have a limited range. If you have  slowly increasing offset, it can push the speaker's "dc" position to one edge of its range where you get distortion, excess dissipation, and inexplicable death. Having a wider bandwidth than you need is a Bad Thing in some cases, especially if you amplify stuff you can't detect.
- drift; always a concern, drift gets critical if you want high gain. One amplifier section is problem enough, but using two or more DC coupled can get you in trouble. For instance - if you look at the datasheet of any opamp, it gives you a number for input offset voltage. The opamp will act as though this amount of voltage offset is present at its input. The 10-50mV number doesn't sound like much until you want to do a gain of 100; then it becomes an OUTPUT offset of 1 to 5 volts. Worse yet, if you have a circuit with several sections doing different things, like envelope detection, modulation, etc, and your input opamp has a 15mV offset, then the offset is carried through all the sections, being amplified as an error term in each one of them and gumming up the works. The alert reader will have noticed that I said "drift" and talked about offsets. Offsets are not static, they wander around with temperature, humidity (yes, that is correct!), mechanical pressures, and age, so you can't even null them out and have them always stay there.

I my first new-bought house, we moved into the place and engaged in the normal pursuits of new homeowers, including getting a lawn to grow. In Texas, this often means removing the rocks from the ground so the grass CAN grow, so it's common to use a pick, shovel, and steel breaker bar to remove said rocks, which can vary from fist size to you-can't-lift-it size. My neighbor was doing the same thing, and over time he and I did a pretty good job. But you could always tell when there was a rock you'd missed, because the grass would simply not do very well in a spot, and you'd know to go dig the rock out and put soil back in the hole. My neighbor had such a spot, which vexed him for a couple of seasons until he decide to end this once and for all. He took his shovel, pick, and breaker bar and set to work. About 4" down he found a fist sized lump, but it would not move. So he enlarged the hole, and found it wasn't fist sized, it was about two feet across - bad for grass! More picking, shoveling, breaking, sweating and the occasional swearing ensued. At some point he had uncovered the rock to a diameter of about two meters. At that point, he had the realization that what he had found was not a rock in the dirt, but rather a thin spot in the dirt letting the greater bulk of the planet show through. That's a very different case than he had in mind.

And the reason I berated you with the story is that you have uncovered such a rock. DC or not DC response sounds pretty simple until you realize that what you're really dealing with is the whole topic of noise bandwidth and DC accuracy as well as side issues of thermal and environmental responses. There's a lot hiding under there.

[Cliff Schecht]

Good point on the 1/F noise and op amp offset voltage multiplication. For the offset voltage, some sort of compensation is required (usually just a trimpot), but this becomes a huge problem when you don't cancel everything out to damn near zero over a wide temperature range. The heating (and cooling) of your components in the amplifier will cause these voltages to drift over time, which is a LOT of work to counter against. I know the method used in high-dollar DC coupled audio amps is the "servo method", but I honestly will admit that I'm not too familiar with this technique (haven't studied it yet ).

[R.G.]

One of the things we had to do in one of my circuits design courses was to design and demonstrate an amplifier with an AC gain of 100. One guy had a JFET input setup whose output DC voltage swung wildly when people walked by the bench, disturbing the air near it. Blowing on it from several feet away did the same. Holding an index finger near the input transistor did the same.

Amp servos are integrators. They take the output  DC voltage as the input to an integrator. The integrator output slews in response to non-zero average output. The output of the integrator is supplied to a DC input of the amp in question in a direction that corrects the output drift and causes the integrator to slew the opposite direction. Done properly, it will correct any offset and/or drift. Of course, they have to have DC response and a high frequency cutoff well below audio or they try to correct the audio, too. 

[Cliff Schecht]

Makes sense. It's just a form of negative feedback with a very limited bandwidth.

[earthtonesaudio]

Thanks for the input everyone!  And thank-you physics for capacitors.   

[aron]

Did you guys see this plugin called Volta made by MOTU? It's awesome! It uses the DC coupled outputs of an audio interface to send CV voltages out!!!!! I really want that plugin. I will be able to control my analog synths again!

[Cliff Schecht]

Did you guys see this plugin called Volta made by MOTU? It's awesome! It uses the DC coupled outputs of an audio interface to send CV voltages out!!!!! I really want that plugin. I will be able to control my analog synths again!

That's pretty frickin' cool! I'm planning on expanding the crap out of my analog synth collection when I get money, everything I have is digital hardware or all-in-one units like the PAiA Fatman (I just finished my second one last night, lots of fun!).