|
How do you avoid DC offset? (pg. 2)
|
View this Thread in Original format
| shockwavedj |
| quote: | | I have DC offset when I record stuff with my 2$ microphone |
Is it a capacitor microphone? |
|
|
| BelgianGuru |
| I have no clue what that is. It's one of those things people game with, only worse ... :D It's this tiny thread it hangs on, and the mike itself is about 2 inches long lol, and maybe half an inch wide :D It's ridiculous how bad it is, but it gets the job done for what I need it :) |
|
|
| shockwavedj |
Sorry, man, I understood 2K $ not 2$... xDDDD That's why I asked you this. For 2$ it's "a bit" difficult to buy a capacitor micro.
But I'm not for hi-q microphones for dance. It's stupid to pay a lot of € or $ to record a voice that's gonna be distorted, vocoded, echoed... |
|
|
| alanzo |
| The DC offset in my tracks was caused by too much bass which would of distorted had I not lowered the levels on my tracks to avoid distortion.. but doing this caused the DC offset |
|
|
| alanzo |
i'm having this problem again with my a-station.. the original recording has 0 DC offset..
but when I try to EQ out the bass below 600hz.. it causes one (but not when I EQ out higher frequencies)
any suggestions on avoiding this? |
|
|
| shockwavedj |
| if you are using a low-shelving attenuation, instead of it try to use a high-pass filter. It can be due to that, but... mmmm, you have many problems with dc offset... are you sure you aren't doing anything else? |
|
|
| Dj Thy |
Yeah, I suggest you try to find out what causes the DC offset first.
Try to do stuff step per step, from scratch (starting with the recording or the synthesis). Some EQ's can cause a little DC offset, but not enough to throw things of balance.
If you consistently find yourself with DC problems, it's not the best thing to keep fixing it afterwards, but it's better to find what causes it.
So like I said, do it step by step. If you find out that some application or plugin causes the DC offset, try either with another similar plugin if the problem persists (for example you mention EQ, try another one for the same process). Especially in digital, if the algorhythm is flawed, scrap it, and use something else. |
|
|
| alanzo |
I did do it step by step and I found out that it was the EQing.. the original recording from my synthesizer had 0 DC offset.. but once I put on a high pass filter starting at 600hz, it created a substantial DC offset.. it did the same with several EQs that I tried..
there no no other FX on the recording (which had 0 DC offset before EQing) |
|
|
| shockwavedj |
| please, put here one snapshot of the spectrum before and after eqing and, if possible, the value of the eq parameters. that's so strange... ??? |
|
|
| halo |
DC is not always bad.
Ok, analog equipment will always be DC decoupled but that's a technical problem.
Let's have a look at a simple square wave. Shockwavedj already statet that DC offset basically is the mean value of the waveform calculated from negative to positive infinite. As our signal is periodic it's the same for only one period.
Now our square waves amplitude ranges from let's say -1 to 1. As times in positive and negative are balanced the average is 0, no DC offset.
What if we apply a PWM of 1:3? Times in positve and negative become unbalanced as negative time is three times the positive. Calulating the average we get a DC offset of -0.5 which is NOT bad at this time as or amplitudes still range from -1 to 1. If we removed DC at this point our diqital equipment could clip as it is proberly defined only in ranges from -1 to 1 and our DC free wave now ranges from -0.5 to 1.5.
What is the point?
Well as long as you keep it all digital, don't bother DC offsets in most cases they are caused by the waveform itself and removing it properly without clipping or losing signal energy is impossible.
As you mixdown (or transfer to analog) remove the very low end of frequency range. Reasons for that were mentioned above.
As for alonzos problem. Phase distorion and removal of low frequenies will always generate DC offset, as positive/negative balance mostly comes from high amplitude low frequency components. In most cases higher partials are insignificant to the fundamental so Offsets might be very small. It's all depending on your original waveform. |
|
|
| halo |
| btw... please notice that calculating DC offset in realtime can never be 100% acurate, as you don't know the exact time of one period. If you dont't calculate over the whole track, it's just plain lowpass filtering at a very low frequency... maybe at a frequency above the lowest in your signal (which is to be EQed out). |
|
|
| shockwavedj |
| quote: | Originally posted by halo
DC is not always bad.
Ok, analog equipment will always be DC decoupled but that's a technical problem.
Let's have a look at a simple square wave. Shockwavedj already statet that DC offset basically is the mean value of the waveform calculated from negative to positive infinite. As our signal is periodic it's the same for only one period.
Now our square waves amplitude ranges from let's say -1 to 1. As times in positive and negative are balanced the average is 0, no DC offset.
What if we apply a PWM of 1:3? Times in positve and negative become unbalanced as negative time is three times the positive. Calulating the average we get a DC offset of -0.5 which is NOT bad at this time as or amplitudes still range from -1 to 1. If we removed DC at this point our diqital equipment could clip as it is proberly defined only in ranges from -1 to 1 and our DC free wave now ranges from -0.5 to 1.5. |
Mmmm... but following your explanation, we shouldn't bother (or do) also with DC over analog equipment, since analog has limited dynamic range as well. I mean analog signals go from one level to another, as digital ones.
PWM is one of the causes of DC. If the synth doesn't remove the very low freqs, it'll be generating a signal ranging from -1 to 1 even the square waveform has a 0% duty cycle. But, we can see that this is ridiculous so what we are generating is a continuous signal with amplitude 1 (or -1), a static DC offset!
Of course, I'm with you: a dc remover should try to not overload, but the signal you have argued if possible should range from -1'5 to 0'5. Why? Because if you amplify that signal a bit more (or add another signal), the original signal will clip (-1'1 to 1'1) at 100% of the whole time but the dc removed only clip (-1'65 to 0'55) at 25%.
IMO DC offset is always bad, beacuse you are wasting bits and dynamic range for something (a ULF signal) that not only you can't hear but, for example, it'll be removed by the soundcard output stage. |
|
|
|
|
