How much DC Offset is too much?

[Pedro Teixeira]

Basically how much DC Offset is too much to have it removed? I was mixing a band and sent it to WaveLab and analyzing one track it detected 0.0337% on L channel and 0.0249% , I used the Remove DC Offset built-in Wavelab and it went down to 0% but I hear nothing... I know this may seem really stupid as it is a stupid question but at what percentage is it "worriable"?

 

[TheWinterSnow]

http://en.wikipedia.org/wiki/DC_offset it can summarize better than i can

basically if the left and Right Channel's are set off too far, you can get phase and artifact issues. If both numbers are extremely high, you loose headroom. The figures you pulled from the DC offset are very normal, they might kill barely a db of signal to you headroom, but will correct equal pushing and pulling for each channel, improving speaker response.

EDIT: DC offset is really for old analog gear that uses op amps since they were known of having drastic DC offsets, when you use all digital, you see figures in the fraction of the percent. I master with WaveLab and use DC offset to compensate for some weird uneven frequencies i sometime get in the bass, has something to do with my bass guitar's active pups. still my results are low, usually what you are getting. in reality there is not need to do it. unless there is a huge DC offset, your only killing headroom, which means you loose maximum possible volume before turning your tracks and guitar tone into the jaymz in death magnetic.

 

[Pedro Teixeira]

So basically removing DC offset on my values does nothing right? Only if I had high values it would be killing dynamic headroom although not hearing it right?

 

[ahjteam]

Lets use a smallscale example. Lets say you have a 8 unit waveform and you want to maximize the incoming signal. Here you have a DC normalled audio, that can be normalled 4 units before reaching the maximum volume; two up, two down.

--------------------------
1
2
3   x     x     x     x
4  x x   x x   x x   x x
--------------------------
5 x   x x   x x   x x   x
6      x     x     x
7     
8
--------------------------

But here you have the same a piece of audio with 1 unit offset, which means that you can only maximize the volume 2 units before reaching the maximum volume; one up and one down, even tho you have 2 extra units left at the bottom of the wave

--------------------------
1
2   x     x     x     x
3  x x   x x   x x   x x
4 x   x x   x x   x x   x
--------------------------
5      x     x     x
6     
7
8
--------------------------

Get the difference? Removing DC offset enables louder audiotracks without distortion

 

[Pedro Teixeira]

ahjteam thanks for the great explanation, I do get that, what I don't get is the real value of the "values" that WaveLab displays, like my mix got 0.0337% on L channel and 0.0249% on WaveLab, I did remove the DC offset to 0% and hear no difference at all, so what I'm asking is at which percentage is "troublesome" like by 5% up or something? That's what I don't understand

 

[ahjteam]

ahjteam thanks for the great explanation, I do get that, what I don't get is the real value of the "values" that WaveLab displays, like my mix got 0.0337% on L channel and 0.0249% on WaveLab, I did remove the DC offset to 0% and hear no difference at all, so what I'm asking is at which percentage is "troublesome" like by 5% up or something? That's what I don't understand

You should do it in the individual tracks, not mixdown. For example a CD is 16-bit audio and has 96dB dynamic range and 24-bit audio often used in the recording business has 144dB dynamic range and 0.0337% from that is about half a dB (144db * 0.0337% = 0,48528dB), but considering that 3dB increase doubles the volume, it means that you have lost about 15% (edit: 0.15%) of the maximum volume possible.

100% = +0.0dB
116% = +0.5dB
133% = +1.0dB
150% = +1.5dB
166% = +2.0dB
183% = +2.5dB
200% = +3.0dB

"The decibel's logarithmic scale, in which a doubling of power or intensity always causes an increase of approximately 3 dB, corresponds to this perception." -wikipedia

edit: oops, no it isn't that much, its 0.0048528dB! Damn zeroes.
100% = 144dB
10% = 1.44dB
1% = 0.144dB
0.1% = 0.0144dB
0.01% = 0.00144dB

0.00144 * 3.37 = 0.0048528dB

edit2: but for your original question: I would say 1% and up is too much, for the above reasons, but if you are going to analyze them anyways, go ahead and do it even if there is a little bit of offset, it only takes a few seconds more to do it

 

[TheWinterSnow]

^^explained it better than i could. you have to remember the DC offset program in WavesLab is finding the DC offset that will yield highest headroom and volume without distortion by compensating with charge could sacrifice some volume to give to the opposite polarity which has some that isn't being used, this allows you to normalize to 0.0db with an average rating closer to 0. In reality the numbers that you pulled in your reading are so low your not going to hear a volume difference. Though it is still worth it, because if your not compressing like the commercial stuff, but want to compete with volume, you take all that you can get.

 

[TheWinterSnow]

edit2: but for your original question: I would say 1% and up is too much, for the above reasons, but if you are going to analyze them anyways, go ahead and do it even if there is a little bit of offset, it only takes a few seconds more to do it

the program is a DC offset remover, so the "percentage that you are given as DC offset is what was detected on your track, it then fixes it by adding opposing polarity DC offset for each channel to compensate.

that means if you had a 10% DC offset detected, if you corrected it, you have just gotten 1.44 db extra headroom.

 

[Pedro Teixeira]

Hm... I think I get it, couldn't be better explained!! But why not remove it on the mixdown? What's wrong with it?

 

[ahjteam]

Hm... I think I get it, couldn't be better explained!! But why not remove it on the mixdown? What's wrong with it?

Its the "fix problems in the earliest stage possible" reason

 

[Pedro Teixeira]

So if I just select each track at a time and remove the DC Offset it is impossible to have any negative effect right?

 

[ahjteam]

So if I just select each track at a time and remove the DC Offset it is impossible to have any negative effect right?

Exactly. Btw, most DAWs have DC offset fixing features, so you don't need to bounce the tracks first and then re-import

 

[Pedro Teixeira]

Thanks a lot for the great explanation!! Yes I have Cubase I know it has built in

 

[TheWinterSnow]

DC offset is created from cheap or old op-amps, and AD/DA converters, which means you won't see huge numbers when you scan for DC offset. most of it will come from an imbalance of pressure since no two sides of the wave will be the same...when you average the total ac voltage both positive and negative, you will see that one might be more, but it will be a very small number.

so in reality if you are recording with all digital gear with instrumentation that is analog and being mic'd, you won't any disturbing numbers.

i run the DC offset fix in my mastering process, because like i said, the totally volumes will vary, and correcting those averages is my way of squeezing out ever db of headroom i can possibly get. Other than that i never worry about it, up close magnification of the tracks show that i have virtually no DC offset, that you can see.

 

[TheWinterSnow]

unless you are using older analog gear before your interface, you won't see noticeable offset.

and by noticeable offset...read this article, you will see what i mean by the pictures, that's when you need to worry about it.

http://www.harmony-central.com/articles/tips/eq_dc_offset/

 

[VH100R]

So if I just select each track at a time and remove the DC Offset it is impossible to have any negative effect right?

select all the tracks->process->remove DC offset (cubase does the rest for you)

I do this to every track before starting to mix the project.