What does DC Offset mean?

  • Thread starter Thread starter realestninja
  • Start date Start date
Imagine you have a pure 80 hz sine wave. It goes from zero to X amount of volts, then swings back thru zero to the same amount of negative volts. That is how it is supposed to work. If you have DC offset, the zero line will be above or below 0, so the swing would go from plus 7 volts to -5 volts.In effect, robbing you of headroom. I'm not sure how many "volts" are put out, or if my explanation is 100% accurate, but the basic principle is there - I hope. :D
 
NL5's explanation is correct. In mathematical terms, if an AC voltage has a DC offset, then there is a constant component.

For example a perfect sine wave will be represented by something like Asin(theta), where A is the amplitude and theta is the phase angle. If there is DC offset, it will be represented by Asin(theta) + B, where B is the DC offset.

So if you're looking at a sine wave on an oscilloscope, adding DC offset will make the whole wave move up if you add a positive DC offset.
 
anyone seen a DC offset that moves the wave DOWN?

just curious....why is it always up?
 
So why would I want to offset it?
Normaly you won't. Any offset should be killed during mastering. Offsets can be side effects of cheap ADC's and are also a comon side effect of many effect processors. If you have a subsonic filter (which also kills the DC offset) involved, it should be placed late in the chain just prior limiting and dithering.
 
anyone seen a DC offset that moves the wave DOWN?

just curious....why is it always up?

A negative DC offset would merely mean a negative constant DC voltage. So a negative offset can exist.

So why would I want to offset it?

You'd want to offset the signal when it has been accidentally done already, i.e. to reverse an offset. So for example your piece of cheap/faulty equipment has introduced a DC offset of +5v, you'd want to eradicate that offset by introducing an offset of -5v.
 
If you have a DC offset in your audio signal, what is the problem? Does it make AD/DA conversions produce errors?

I got nothin :)
 


it consumes headroom for one. and your speaker cones won't be returning to their resting point.
Also, the problem can be compounded as you mix more tracks together with it. It's best to really reduce it as soon as you see it...and not wait for the mastering stage like LogicDeLuxe said. An easy fix can be just to put a hi pass filter on the track...set real low. You'll have some crazy low end going on too if you look at it.
 
there seems to be some confusion here in that your talking about two different kinds of offset... first in the anaolg world... whwere for instance a difference in rail voltages from side to side cause the output to "shift" accordingly... ie.power supply rails of say +15 &-14 would cause an opamps output with 0 signal present to sit at +.5 V.... or a dccomponent on the input can have similar results.... the second kind of offset is an error in an A-D chip where your wave form seems to ride along a line other than 0... not totally sure on this but i think i read somewhere (lavrey's bbs maybe) that this is caused by a problem at the Vref input of the chip...
 
So why would I want to offset it?
Normally you wouldn't on purpose.

However, when building effects or synths in modular environments such as Reaktor for example, you can induce DC offset on purpose to change the behavior of the subsequent DSP operations. For example, inducing DC offset before quantizers, can make the output more sputtery. You can also modulate the amount of DC offset to create quasi-PWM like sounds. Waveshapers also respond rather nicely to DC offset, and again modulating the offset can be another source of sound shaping.

This type of stuff comes in rather handy in sound synthesis, and I am surprised that very few systems allow for it.

However, even in such instances, you want to make sure that your final output does not have DC component to it.
 
What will it look like?
In a waveform editor like Sonar or Cubase, the waveform will look normal except that it's center will be shifted upwards or downwards (usually upwards) off of the center (-inf) line in the waveform display. Or, if you record with no signal on the input, you'll see a flat line go across the display riding above (or below) the centerline.

This is why it decreases the amount of headroom. If you have a DC offset of, say for example, +5dB, the waveform will be shifted 5dB up off of the centerline of the waveform display, meaning there's 5dB less headroom on the positive (top) side of the waveform to fit your peaks.

If your peaks are large enough, then you'll see clipping on the top of the waveform even though you'll still have room on the bottom of it.

G.
 
What will it look like?

here's some obvious ones. First you'll usually see this:
http://www.making-music.com/SampleEditing-DCOffset.jpg

In that one you can see it's biased to one side. However, sometimes this is just naturally what happens with that certain sound.

So then you look for this:
http://www.harmony-central.com/Computer/Articles/Benchmark/dc-offset-wave.gif

see where the wave doesn't start at the zero line? It ramps up instead. You'd probably hear a noticeable pop when you pressed play.
And then in this
http://www.harmony-central.com/articles/tips/eq_dc_offset/fig1.gif
obviously you want it to look like the second wave....not the first. See how much headroom it takes up?



And if you STILL want more examples, here's one I came across a few weeks ago:
https://img64.imageshack.us/img64/5475/untitled1fi6.jpg
The wave form was so thick that it was hard to tell what was going on with it...but I had a feeling that there was a DC offset problem.

This is blown up quite a bit too. Notice I had just started playing the region and even though the track is down around -48, the low end is freaking out. Apply a DC offset or hi pass filter on the region (this one at 20Hz) and you get:
https://img406.imageshack.us/img406/1073/untitled1gd1.jpg

The wave form has been shifted and everything below 20Hz is taken care of.
 
and not wait for the mastering stage like LogicDeLuxe said
If the offset is really excessive, ie. more than a few percent, you would filter right after sampling. Otherwise, the less you have to filter, the better, especially when the goal is a truthful sound.
And in case, the offset comes from effect processors in the chain during mixing, you only could handle it afterwards anyway.
 
Back
Top