What is DC Offset?

This is one of the strange questions one comes across when investigating audio. Can someone tell what DC offset is and how important it is in the mastering process?

Thank you for bearing my lack of audio knowledge.

 

The change in input voltage required to produce a zero output voltage when no signal is applied to an amplifier.

http://www.twysted-pair.com/dictd.htm

 

More than you ever wanted to know.

In a nutshell, DC is unwanted and distortion to our ears when traveling with the AC signal towards your speakers. I know nothing about mastering but garbage in, garbage out seems to apply.

DC Offset, distortion and You

 

Yes

If you look at a sine wave it has what we call zero crossings. The signal goes up to a peak and then comes down to zero and then continues to a negative peak and back up. If the center point is at 0 volts, there is no dc offset. If there is 1 volt DC offset, the center point will be above the zero point. In audio circuitry, a series capacitor is often used between stages to eliminate the dc offset. Of course there are now other ways of eliminating offset than series capacitors. DC will not pass through a capacitor, but signal will. The value of the capacitor determines how low a frequency will pass. If there is dc offset, it will mean the swing of the voltage can clip on the top of the waveform. The peak to peak signal of a sinewave can not exceed the maximim power supply voltage. If you raise the center above zero it can clip if the signal is too strong. This will be a non symmetrical clipping, as the bottom of the wave will not clip. Of course if you have a negative DC offset, the bottom can clip instead. Also when there is DC on an audio signal, when you go through a switch, it can make a clicking noise.

 

Thanks, but what importance does it have in a recording/mastering environment?

 

It will lower the maximum possible amplitude of the signal before clipping. It will not cause distortion in all cases, but should be avoided. You want the signal centered between the postive and negative voltage of the power supply. If not, you will have less headroom to work with.

If I could draw you a picture it would make more sense. Clipping is bad, but asymetrical clipping can be even worse. Both should be avoided unless you're trying to make a fuzz box.

 

Quite interesting, Doug!

Am I wrong, or can DC offset also be measured as a percentage?

 

Sure. If you had a scope you can easily see it. The sine wave will look the same with DC offset as long as you don't reach the maximum voltage of the power supply, but the sine wave will shift up or down on the scope depending on whether the offset is positive or negative.

There is a ratio between the amount of offset and the power supply voltage. You can filter the offset after an amplifier stage, but once again, the amount of voltage swing of the signal before that filter will be limited. As long as you don't clip in that stage with the offset, and go through the filter, there will be no problem, but it should still be avoided to insure maximum headroom. That is why there were tons of electrolytic capacitors in vintage consoles between sections. They filter the offset. Of course there should not be any offset if things are right, but the capacitors insure that if there are they are not continued through the circuit.

And once again, modern circuits often use servo circuitry instead of capacitors to lessen interstage offset, because capacitors don't sound that good. They also can cut the low frequency response of the circuitry unless larger caps are used. I used to modify consoles by putting in larger value interstage capacitors to allow lower frequencies to pass, and then bypass them with high frequency capacitors to get the best possible high frequency response. There is no such thing as a pure capacitor. They also have resistance and inductance and can also slow down high frequencies. I used networks of capacitors to insure the best possible signal flow, sort of like a crossover on a speaker. A big capacitor is great for bass, but causes high frequency issues. Using a special high frequency capacitor in parallel can open up the high end.

Btw, these caps when they dry out can pop, and then there is no signal getting from one stage to the next. I would say that that was the most common problem I ran into when servicing consoles. I generally didn't need test equipment for that problem. I'd just look for the cap with the end popped off. Of course things weren't always that easy to fix.

Sorry if this is too much info.

 

No problem Steve. I love talking about these things. I guess this stuff is like riding a bike. I've not thought about it for years, but fortunately it comes back when I need it.

 

Although just barely within my grasp your info is quite interesting.

See, I came across this issue while trying to backup an old radio program (31 years old in mono) I had on cassette to my hard disk in stereo to make a CD out of it. Don't know why I was recording to stereo (I later continued the rest of the work in mono). Anyway, I did a DC offset check on it and the least dramatic numbers I got were a -5% DC offset on both channels (there was one case where I had gotten 12% on one channel and 5% on the other IIRC). I later did a DC offset correction on them and merged the stereo to mono and, finally, normalized the resulting WAV file. I did notice more detail in the file after the whole process. I hope I didn't overdo the processing on it.

Any suggestions anyone as to how I should have focused this case so as to have gotten the best results? (I still have some 3 files/cassettes to go).

 

Sorry, if I was difficult to understand. I try to simplify things so they are easily understood but often miss the mark. In person, I'll often try many approaches until the person asking totally understands. That is harder to do in this type of situation. I'm convinced I can explain anything I know to anybody if I take enough time to find something that they can relate to and explain the concept in those terms. Of course some times the person asking gets frustrated and calls the discussion off.

Regarding your project, I'd say if it sounds better to you than you probably did just fine. After all how it sounds is all that matters. How you got there is irrelevant IMO. If it sounds good, it likely is good. I'd say that our ears are capable of hearing more than we know how to measure, but even that has been a subject of much debate around here. . Of course judging sound quality can also vary from person to person as is also plainly obvious from reading posts in this forum.

The only suggestion I'd make is to stay in mono if you start in mono. If there is any difference at all between the stereo tracks that came from mono, as a result of processing or whatever, than there will probably be some issues when combining them.

 

Something like this.

The first image is a sine wave centered on the 0dB axis. Second sine wave has a positive DC offset...with an example of how the same amplitude of sine wave can be clipped if the DC offset is not removed.

 

Yes, I should have started working in mono as I am now. Maybe it's just a matter of not having stayed up all night with it as I did yesterday.

As to your explanation: I too imagine that a couple of drawings would have illustrated your point easily. It's OK, these are the limitations of a forum, but I've also gotten to understand this better through your words so I guess it's got its good and bad points.

On the other hand, it seems your words have also been intersting to Steve, so they are working.

And, no I feel no inclination to call the discussion off.

Again, thanks.

 

Hey, Rudy, great! Undoubtedly a picture is worth a thousand words. Now I can see what Doug has explained. Thanks!

So my DC offset negative figures were showing an increased possibility of the waveforms clipping in the bottom phase on my recording. Right?

 

Rudy to the rescue, again. Thank buddy.

This should help Metoo.

 

Correct. However, if your levels are low enough, you won't have to worry about clipping. If the sine waves in my examples were at a lower amplitude (lower volume), there would not be a clipping problem...but regardless, it's best to keep it centered. If you did any further processing (increasing volume, mainly), you do stand the risk of clipping the lower part of your waveform.

 

No problem--created a sine wave in an audio program, then trimmed the image in Photoshop. Figured it was much easier to see it than try to describe it...

 

So, if I understand it correctly the DC offset could be produced by my old cassette deck. Or is it the recording, or some type of incompatibility between the recording and the tape deck?

 

My guess would be the cassette deck.

 

If I get something on a CD-R to work on, the first thing I will do is check for DC offset. If it's there, I will remove it. If I hear clicks between tracks, I know there are problems that need correcting without checking.

A reading of about .001%, for example, is not enough to even be concerned about. .5%, as you've stated is excessive!

When working in a DAW, the better the soundcard, the less DC is a problem. When recording into one, you can set things for automatic removal on the fly if the card produces DC.

 

An idea (and I might be talking out of my a** here): have you tried running a ground wire from the case of the PC to the case of the cassette deck? I wonder if that'd help?

 

It means that this is the trim pot inside your amplifier that you NEVER-EVER touch or try to tweak on your own.

 

That would help with equalizing any ground voltage differences between the components. These are usually AC voltages, and I guess they could be considered offsets of sorts, but these are much more difficult to get rid of after the fact. They manifest themselves as 60 cycle hum. Since they are AC, they are basically audio that can be heard. DC offsets can't be heard directly, though they can lead to problems which can be heard.

The problem with ground loops occurs when there is a duplicate ground path. For example if 2 components have 3 prong electrical cords, there could be a problem. These two components can have their case grounds tied together if the third wire grounds are tied together where the power is plugged in. Now if you hook an rca cable up between these components there is also a connection between the two cases. Now consider that there is rarely a true ground at 0 volts. Sure they are close to 0 volts, but there is usually a very small voltage present. If the length or resistance of the ground connections are different, different components can have different voltages on their cases. When you hook the interconnect up, the difference in the voltages between the two components can now be amplified. The only way to remove this is to isolate the grounds or insure that they are at virtually the same potential. But once this hum is there you can't easily filter it out.

With a DC offset, it is pretty easy to remove after the fact.

Ground loops are often difficult to find. In a studio, when you remove the worst grould loop offender, you can lower the noise floor, but then the next worst offender stands out. This can go on for quite a while, but with a fair amount of work, you can lower the noise floor considerably. There are many techniques that can be used to do this, and they are probably way off topic to discuss here.

I will add one more thing. For some reason, some studios or audio installations are wired haphazardly without obvious ground loop problems. In other cases, great care can be taken in wiring, and one small problem can wreck everything. Finding that one problem can sometimes take a long time to isolate and correct.

 

I will add that DC offset behaves much the same way with speakers. When there is no signal going to a speaker it is centered in it's throw range. If there is a DC voltage present the speaker will move one way or the other depending on the polarity of the voltage and this will now be the center position. This will present the same type of problem, in that there is a limited amount of throw from the offset position to the extreme of the throw range.

I often hook up a battery, which is of course DC voltage, to an array of speakers to verify polarity. If they all go the same direction from center they are in phase.

Also, biasing of transistors in a circuit is also a form of DC offset. When you first apply voltage to a transistor it does not turn on until you reach a certain voltage. This is not a large amount of voltage and is usually around .6 volt. As a result, bias (or offset) is usually applied to make sure the transistor never operates below this level. If it does, there will be distortion and nonlinearity. As you might imagine from the other examples in this thread, this does lower the efficiency, or headroom, a bit, but the tradeoff is well worth it if low distortion is desired. Of course this offset must be filtered out before the output of the circuit.