What causes bursts of DC Offset?

[24-96 Mastering]

I came across source material with several bursts of DC offset today. Easily removed with a HPF, but I wonder what might have caused them. Does anyone have any ideas?

[dcollins]

Originally Posted by 24-96 Mastering

I came across source material with several bursts of DC offset today. Easily removed with a HPF, but I wonder what might have caused them. Does anyone have any ideas?

Why does it have to be DC? That could just be a burst of LF.

DC

[TranscendingM]

That's true, the better spot to examine for DC offset is where it should be "silent". To my understanding, any electronic device can have perturbations at their outputs because parts are not perfect i.e tolerances
If these voltages are amplified along with the signal they could show up as DC offset.

[haberdasher]

It seems odd to describe that as DC offset?

But to answer your question, I don't know. Does it occur "randomly" or is there some pattern? If the former I'd guess a faulty analogue outboard (connections or similar); if the latter either outboard (old tubes, etc.) or a defective plugin. The closest thing I've seen to that was a few recordings on an album that became offset during the fade-out. Never found out what caused that, either but it a least seemed to be level-dependent.

[24-96 Mastering]

Originally Posted by dcollins

Why does it have to be DC? That could just be a burst of LF.

DC

The term "burst of DC" is probably not well chosen, maybe even somewhat nonsensical, I guess. I wrote DC because the anomaly is very low in frequency. If it's a LF wave, it's at less than 1Hz.

[24-96 Mastering]

Originally Posted by haberdasher

It seems odd to describe that as DC offset?

But to answer your question, I don't know. Does it occur "randomly" or is there some pattern? If the former I'd guess a faulty analogue outboard (connections or similar); if the latter either outboard (old tubes, etc.) or a defective plugin. The closest thing I've seen to that was a few recordings on an album that became offset during the fade-out. Never found out what caused that, either but it a least seemed to be level-dependent.

Thanks for the suggestions. It's on a recording I got in, I don't know anything about the recording setup. There are only 4 instances of the anomaly, and there seems to be a rhythmical pattern:

burst
7 seconds normal
burst
3.5 seconds normal
burst
7 seconds normal
burst

Could be coincidental, of course.

[Noooway]

I have seen those bursts on every single file of a session. I asked those who recorded it if there was like a refrigerator, or kitchen machines, in the room while recording, and YES there was. On the same AC power I guess.
Only good way I found to get rid of them is to zoom very close in on every instance with iZotope RX and use Spectral Repair starting out with a preset like 'Replace unwanted event (slow)'.
One time I recorded a vocal track in that same room without getting those spikes, but I used a power conditioner. And I did not know about that fridge, ...
A refrigerator would even also spike a tape machine ...

[jdg]

someone bumping a mic stand a bit?
or thumping on a tube mic-pre while recording?

[Noooway]

Originally Posted by jdg

someone bumping a mic stand a bit?
or thumping on a tube mic-pre while recording?

I've seen and worked those spikes. It is not peaks. If you zoom close in is very low level spikes/cracks in samples like 4 samples suddenly breaks the line abnormally just a bit. You can for sure notice them by ear. And on a 96kHz recording they easily go up to 40 000 Hz, way beyond e.g. good OH mics.

[Jerry Tubb]

From the looks of it, just a touch Sub LF rumble mainly on the right channel.

I see those occasionally as well, not sure its a big problem unless its audible.

Cheers, JT

[Noooway]

Sorry, ignore my earlier posts. I was interpreting the picture the wrong way. Gotten so used to seeing 'burst' for a while, so what seems only to be the marker I thought was the problem. But of course, that is not a frequency view, so sorry again

[24-96 Mastering]

Here's the event zoomed in, with time line for reference. Looks like sudden offset after all.



PS: And there's a click at the start of the offset:

[jdg]

power supply sh*ting the bed in some recording equip they're using?

[glassmaster]

Bursts of DC are caused when you post about jitter, dither, or Microdynamics on this forum.

[Noooway]

Originally Posted by 24-96 Mastering

PS: And there's a click at the start of the offset:

Attachment 172678

Hmm, actually kind of looks like one of those bursts I was talking about just after the middle between 2m3s and 2m4s, the purple spike. Check it out with RX like I said. Can you zoom all the way in on that burst and post a pic all the way in sample level (not frequency view)? I am almost sure there will be a 'break' in there. I can hear those usually. Smacks the tweeter.

[Noooway]

Originally Posted by glassmaster

Bursts of DC are caused when you post about jitter, dither, or Microdynamics on this forum.

[Bob Olhsson]

I understand that allowing the mathematical rounding of DSP calculations can cause a DC offset.

[Alexey Lukin]

I think I've seen such bursts due to:
* bad phantom power supply to the microphone,
* hitting the mic stand,
* wind or air movement (some mics are pretty sensitive down to 1 Hz).

[macc]

See things like this fairly often... Usually caused by heavy-feedback chorus/doubler-type plugins that haven't been HPF'ed, also in some synths (Massive!) when generating those extremely weird waveforms that dubsteppers like to create.

[24-96 Mastering]

Thanks for the suggestions. I checked with the artist and the recording setup doesn't really provide clues as to where this originated from. But the band likes to use modulation effects, so Macc's suggestion could be it. All other suggestions are possible too though, so I guess there's little point in speculating...

[dcollins]

Originally Posted by Bob Olhsson

I understand that allowing the mathematical rounding of DSP calculations can cause a DC offset.

Sure, but that's not what we see there. it's just a bump, a whump, of LF or something.

Real DC offset causes the entire wave to be shifted from the zero-line. Not to be confused with asymmetry in musical instruments, the HEDD process, etc.

It's pretty much a non-issue, but I think the emphasis on analysis causes people to see a numerical value or unusual waveform and worry.


DC

[Nordenstam]

Originally Posted by macc

See things like this fairly often... Usually caused by heavy-feedback chorus/doubler-type plugins that haven't been HPF'ed, also in some synths (Massive!) when generating those extremely weird waveforms that dubsteppers like to create.

+1

See this in ITB productions. Think it's the digital equivalent to control voltage bleedthrough in modular synths. Certainly looks like it!

[dcollins]

Originally Posted by itsybitsy

exactly!

and i'm sure you'll agree that the use of a proper dithered system will solve any of those issues infinitely quickly anyway, so there's really no problem.

Don't you ever get tired?


DC

[PatrikT]

I only zoom in for the fade-works, hence missing bursts completely.


Regards
Patrik

[24-96 Mastering]

Originally Posted by dcollins

it's just a bump, a whump, of LF or something

Thinking out loud: Visible on the zoomed in pics, it starts with a click, then the whole wave is quickly shifted and slowly returns to zero. If this was a wave, it would be at 1/2 Hz. And we see amplitude one way only (also true for the other 3 instances). And if it was a wave, there'd have to have been an edit to explain the click and amplitude jump, and that doesn't appear likely due to the nature of the recording and the timing of the events.

Originally Posted by dcollins

Real DC offset causes the entire wave to be shifted from the zero-line.

Is there a term for an event that is neither a wave, nor a static offset, but a temporary shift in potential? (hope I can use the word potential this way in english)

Originally Posted by dcollins

It's pretty much a non-issue, but I think the emphasis on analysis causes people to see a numerical value or unusual waveform and worry.

To me, this is not a worry but an interest.

[Bob Olhsson]

Originally Posted by dcollins

...I think the emphasis on analysis causes people to see a numerical value or unusual waveform and worry...

all to true.

[Alécio Costa]

Bursts of "DC" when you mention you work with an RTA
LOL

[bigdoghat]

Just got something with this issue - it IS all above the zero line visually and there actually is something else to consider which I've just discovered

This was a small piece of room tone with a DC offset issue - when I cut a clean section of it to use, there was a small audible click at the beginning and end of the piece - I used the DC offset remover plugin and once this was done the audio no longer clicked at the start and end - without using the DC offset removal plugin, this would be a huge problem, I would have had to crossfade every edit that used this piece of room tone to stop the audio clicking - so sometimes, it does come in to play

[Arksun]

I don't think its the cause here, but just as a side note some plugins can cause DC type offset issues as well (extreme low frequencies of high amplitude, sub-10hz stuff). Either from badly coded or the way they manipulate the waveform through their algorithms. Some soft-synths as well, particaurly if using pulse width modulation.

This could create very similar looking shifts in the waveform display.

[dcollins]

Originally Posted by bigdoghat

Just got something with this issue - it IS all above the zero line visually and there actually is something else to consider which I've just discovered

This was a small piece of room tone with a DC offset issue - when I cut a clean section of it to use, there was a small audible click at the beginning and end of the piece - I used the DC offset remover plugin and once this was done the audio no longer clicked at the start and end - without using the DC offset removal plugin, this would be a huge problem, I would have had to crossfade every edit that used this piece of room tone to stop the audio clicking - so sometimes, it does come in to play

Those are all symptoms of actual DC offset.


DC