DC offset keeps rising throughout every vocal take

[MMensch]

I seem to have a strange issue with the recording set up I'm using for the vocals. My chain is as follows:
Behringer B-1 > A&H Zed420 preamp > FMR RNC > Yamaha A3000 sampler

What I tend to do are three- or four-minute takes, which I then export into my computer for editing. When I then look at the waveform, it's immediately obvious that the DC offset is steadily rising throughout the take - the silences get progressively fuzzier the further right you go (and the "Eliminate DC offset" function in my WaveLab Lite also confirms this). Furthermore, each file/take is roughly the same in this regard - it's not random or anything, just very steady and predictable upwards slope of fuzziness every time.

Any ideas as to what could be the cause of that? I honestly have no idea. It would be easy to point the finger at the B-1 as the only (potentially) low-quality link in the chain, but considering that the situation "resets" with each take, I don't see how this would be possible (ditto for the rest of the gear).

 

[Sonixx]

I'd start by elimination

What happens if no mic is connected?

A Dynamic?

Turn off Phantom power?

Disconnect the RNC?

Connect only the RNC?

Remove RNC from signal chain.

Sounds like a Capacitor is charging with the signal present. Possibly an Open Pulldown Resistor somewhere.

What Happens if you do a take and let the DC build, Stop recording and immediately start recording again? Continuing to Talk/Sing.

Does the DC build to the same level regardless of input level?

 

[SouthSIDE Glen]

each file/take is roughly the same in this regard - it's not random or anything, just very steady and predictable upwards slope of fuzziness every time.
...
the situation "resets" with each take

The question there (for me) is what is it that you're re-setting with each take? If the only thing that you're "resetting" is the record function on the sampler - i.e. if you're not turning off/on the preamp or any other device in the upstream signal chain in between takes - then I'd put your sampler at the top of the suspect list. Otherwise, I'd agree with sonixx that you'd have to troubleshoot the chain.

Also, I just want to clarify what you mean when you say an "upward slope of fuzziness". Does this mean that the amount of "fuzziness" stays the same but the line of it climbs, or does it mean that the actual amount of fuzziness actually increases in amplitude? If it's the former, then it is probably a VLF component being introduced somewhere (which an offset filter or a high-pass EQ filter can easily handle). If it's the later, then that sounds more like it describes an increase in noise than a low frequency offset.

G.

 

[MMensch]

Hi, thanks for the speedy replies. Yes, I'll definitely troubleshoot the next time I record, but I figured I'd ask first in case it was a known problem with a known list of potential remedies (I've only recently noticed it).

The question there (for me) is what is it that you're re-setting with each take? If the only thing that you're "resetting" is the record function on the sampler - i.e. if you're not turning off/on the preamp or any other device in the upstream signal chain in between takes - then I'd put your sampler at the top of the suspect list. Otherwise, I'd agree with sonixx that you'd have to troubleshoot the chain.

I think I'll do a 4-minute recording with nothing even connected to the sampler input first, and take it from there.

Also, I just want to clarify what you mean when you say an "upward slope of fuzziness". Does this mean that the amount of "fuzziness" stays the same but the line of it climbs, or does it mean that the actual amount of fuzziness actually increases in amplitude? If it's the former, then it is probably a VLF component being introduced somewhere (which an offset filter or a high-pass EQ filter can easily handle). If it's the later, then that sounds more like it describes an increase in noise than a low frequency offset.

The fuzziness increases in amplitude, but stays perfectly centred around the 0 line. So you're telling me it may not even be DC offset? But how come the "Eliminate DC offset" function gets rid of it then?

At any rate, I'm attaching a screenshot in case you can make something out from it. You can see the line get visibly fatter (in a totally linear way) even in the space of just a bit over a minute. In case it is an increase in noise as you suspect, do you have any ideas what could be causing it?

 

[SouthSIDE Glen]

It's very hard to say from that pic just what's going on, but that is not the typical profile for DC offet at all. DC offset typically means that the centerline of the wave form shifts off of the centerline of the display; i.e. the rest voltage, or quiet part of the signal is not at 0DC (-inf dBFS).

That's not what you have. The exact nature of what you have is hard to tell because of the low resolution of the zoom level in that pic that makes the "fuzziness" appear as solid red. It looks like either heavy-duty noise or a high frequency oscillation, but it's impossible to be sure at that resolution. So I can't explain what the DC Offset filter thinks it's seeing.

Perhaps you could post a small file of a good chunk of the "noise" section - e.g. the area from ~1:17 to ~1:27 - that we could take a look at and listen to in more detail? I'm curious, if nothing else.

G.

 

[Raw-Tracks]

This may be a long shot, but could it be headphone bleed? Is the track you're singing to building up to the end of the song?

 

[MMensch]

This may be a long shot, but could it be headphone bleed? Is the track you're singing to building up to the end of the song?

No, it's a loop, I'm doing takes of the same verse, plus my Sennheiser HD280s are doing a good job of keeping the spill to a minimum.

And you can take my word for it, because I'VE JUST FIGURED IT OUT - purely by accident. Thing is, I normally transfer the sample into the computer soon after I record it, playing it back from the sampler's play mode. But today, since I didn't need to do any work on it, just make a noise sample for Glen, I simply streamed it from the sampler's hard drive ... and there was *absolutely none* of that fuzziness all of a sudden (pictures on request). So I guess there must be something happening somewhere along the way towards the output once the sample is in the RAM ...

... and, again, have no idea what it could possibly be - the sample is in its utmost default state, there's no filter assigned, no EQ, no master EQ, analogue in is disabled. I'm guessing there must be something there feeding back into itself since the fuzz that you can see in the picture is obviously not part of the file at all. So I guess I'd be interested to hear from other Yamaha A3000 owners if they're having the same issue, and thanks again to everyone who tried to help.

 

[SouthSIDE Glen]

No, it's a loopThing is, I normally transfer the sample into the computer soon after I record it, playing it back from the sampler's play mode. But today, since I didn't need to do any work on it, just make a noise sample for Glen, I simply streamed it from the sampler's hard drive ... and there was *absolutely none* of that fuzziness all of a sudden (pictures on request). So I guess there must be something happening somewhere along the way towards the output once the sample is in the RAM

I want to make sure I understand correctly. Are you saying that you normally "transfer" files to your PC by playing the audio out and then re-recording it on your PC via the audio in?

If so, and if you're going through your built in PC sound card, there's a good bet that the problem is with the PC sound card, especially (but not exclusively) if it's a laptop PC.

G.

 

[MMensch]

I'm sorry, I wasn't being clear - both times, the sample gets played through the sampler's optical output into my soundcard's optical input. However, when it's streamed from the hard drive, there is no fuzz, and when it's loaded into the sampler's RAM and triggered from the play mode (like it normally would be), the fuzz gradually builds up as the sample is played back.

One fact that might be of note and that I'm aware of is is that the A3000 stores the actual waveform data separately from the sample's settings (transposition, filter, enevelopes, what have you). So when you stream the file from the hard drive, it's giving it to you raw so to speak, so perhaps there's a setting that I'm overlooking that becomes active once both the sample's waveform and its settings are loaded into RAM and played back.

I'll try a few more things now, and if I discover anything, I'll report back.