Skip to main content

Hi,

Could someone explain what DC Offset is? I only have a vague idea of what it is. Been stuck trying to explain what it was to someone who had trouble with it the other day and... well, I couldn't...
I did a search on RO, but did not find anything satisfying enough.

Thank you!

Topic Tags

Comments

Thomas W. Bethel Thu, 11/03/2005 - 05:32

From the web

DC offset is an imbalance that sometimes occurs in A/D converters . It is a constant voltage that is present which can eat up headroom and cause clicks and pops during editing.

Imagine a sine wave with the top and bottom wave forms equally situated over the base line. Now imagine that the sine wave is shiffed up or down so it is no longer symetrical to the base line - that is DC offset.

Most times a high pass filter will take care of any DC offset problems.

anonymous Thu, 11/03/2005 - 11:42

I think the most common source of DC offset, is from what is called ADC offset error. Basically, this causes a shift in the input/output transfer function of the ADC.

To state this more plainly, when you apply 0V to a ADC, you should get the binary code 0 out. If you don't, then you have an offset error. Assuming the gain error of the ADC is zero, the same offset will appear for all input voltages. IIRC, offset error happens when the reference volatge in the ADC is off.

RemyRAD Thu, 11/03/2005 - 20:50

Generally DC offset is a result of a cheap soundcard. Most better quality soundcard don't have that problem. Depending on your software, adding a simple filter without manipulating the controls, i.e. flat, will frequently eliminate the DC offset. Adobe Audition, I believe actually has a setting to eliminate DC offset. Computers can record DC values and there is no place for that in audio. If you can see the waveform, when recording silence, and then play that back, you may notice that the baseline may be higher or lower than zero. That's the offset.

audiowkstation Thu, 11/03/2005 - 23:40

Although most engineers are cognizant of DC offset, the way it occurs vary widely. It normally occurs in the analog domain and the main cause is poor power transfer due to a power supply inadequatcy. Often a certain keyboard or synth using a "wall wart" or small power supply will display the artifacts. Certain patches with chamber can cause it. If one channel of your mix has excessive DC offset, this WILL cause issues with sinusoidal offset of the entire mix. When mixing, each channel in the digital domain can be monitored and with an editor, offset can be corrected.

Another common cause is improper gain management. If one component in the chain is overdriven, DC offset will result. Anytime the analog domain is overstressed, wave shape is modified and DC offset displayed.

Is DC offset a bad thing? Sometimes no! If you take a full Marshall stack and overdrive the amplifier, not just the input signal, you will have DC offset. The trick is to know when it is destructive to overall fidelity and destructive to your mix. Listening tests and track by track editing and a/bing your results should be excersized.

Clipping a power amplifier will cause offset. Some natural sounds actually contain offset.

Zilla Fri, 11/04/2005 - 17:19

Any audio equipment which incorporates an analog stage can exhibit DC offset.

Audio signals oscillate in positive and negative polarities. In a word: bipolar. To reproduce these signals analog circuits usually are powered from bipolar power supplies ( +/- 15Volts for example). This allows the signal voltage to swing positive when a compression waveform is reproduced, and negative voltage on a rarefaction waveform. When there is no signal (silence) the output of the circuit should sit at ground potential: 0Volts.

However this is an ideal. In the real world component tolerances, temperature changes, topology, etc., cause circuit drift. So instead of sitting perfectly at 0volts during silence, you may find that it is +40millivolts or -3millivolts, for example. This voltage is not an oscillating quantity. It will be flowing in a single, fixed polarity; a direct current (DC). This is DC offset. A few millivolts offset is nothing to worry about. But if it becomes too great you will suffer with less signal headroom, pops when bypassing, and other maladies.

Another source of DC offset happens during recording. If a very slow draft from an A/C duct is blowing on a mic, this will induce a very low frequency (~1-10Hz) signal in the mic. So slow that it is considered essentially DC in nature and treated as DC offset.

RemyRAD Fri, 11/04/2005 - 21:22

Easier explained, computers can record DC voltage values. This is one of the primary reason why capacitors were invented. A capacitor can block DC voltages from traveling in one direction or the other depending on its polarity, provided that it is an electrolytic capacitor. A lot of audio guys don't like capacitors as they are also simple filters. They've never hampered my recordings. A lot of circuits nowadays utilize servos instead of capacitors to accomplish the same thing a simple capacitor can do.

RemyRAD Sat, 10/28/2006 - 10:19

Sure there are some cards that do not exhibit DC offsets. They are called good sound cards, with good circuit designs and implementation. Sound Blaster and many of the $15 bargain Taiwanese sound cards the exhibit lots of DC offsets problems. So if you have a $15 soundcard and are experiencing DC offsets problems, frequently almost any software has the ability to after-the-fact. Of course they don't tell you about the additional distortion that was created from the DC offsets to begin with but then with a $15 soundcard, you shouldn't worry about it that much anyhow. So if you notice that the centerline of your waveform appears above or below the center baseline of your software, you have a crappy soundcard. So then you can correct for it in software.

The holidays are just around the corner and you might want to talk to the good religious fat Ferry bearing gifts to get you a better soundcard? He'll know if your mixes have been good or bad. So make them good for heaven's sakes. And what about those new Rudolph rednosed microphones?? I understand they record bells very nicely.

I record all of my fine arts concerts with the new Decca Menorah Microphone Tree. You just keep adding a new microphone every day and after the seventh microphone, you will have cut some wax! Then you can kickback and enjoy the fruits of your effort as long as it's not an apple from that nasty guys garden.

Rockem' Sockem' Robots please
Ms. Remy Ann David

dementedchord Sat, 10/28/2006 - 13:23

RemyRAD wrote:
I record all of my fine arts concerts with the new Decca Menorah Microphone Tree. You just keep adding a new microphone every day and after the seventh microphone, you will have cut some wax! Then you can kickback and enjoy the fruits of your effort as long as it's not an apple from that nasty guys garden.

Rockem' Sockem' Robots please
Ms. Remy Ann David

plz....not while i'm drinkin coffee....

RemyRAD Wed, 11/01/2006 - 14:06

Sure it has a relationship to the noise floor. The DC offset will move the waveform above or below the baseline which moves the signal closer to the inherent noise level of the electronics. It also causes considerable distortion.

I live in the land of the Washington DC offset
Ms. Remy Ann David

Groff Thu, 11/02/2006 - 17:01

Here is my 2c about dc offset

I have Apogee Rosetta 200 and I did simple test (I’m not shore if the test is good or scientific, but it's a real life situation) I recorded „silence“, nothing plugged in except AD to PC (apogee x-firewire card via Adaptec FW 4300 /- as recommended by Apogee). The result was the noise around -98 dB on DAW peak meters with some fluctuation up to -96 dB. For 24 bit AD it seems to me… well … higher than I expected. Theoretically that's the noise floor of 16 bit converter. Rosetta’s spec-> THD+N (AD): -105 dB. So 10 dB are missing.

First I thought that was something from PC (even with good parts inside) so I disconnected all my PC hard drives except system disc and all coolers including the one for processor :shock: (but only for few min. 8) ) and repeated the test. The results were more or less the same, around -98/96 dB.

Noise power? I bought Furman’s power conditioner and voltage regulator (which is good move anyway). Nothing. Still at -96 dB.

FFT analysis showed positive dc offset under 200 Hz. I contacted Apogee support and here are three most important points from respond:

This actually does not sound like there is a problem. That db reading that you are getting is actually close to what we get on the AP

(difference of 10 dB - close enough?)

There is a very small amount of DC offset on the Rosetta 200 that is caused by the actual op amp that was used. The engineers found that specific Op Amp to sound the best and that’s why they chose it, as they felt the DC offset was so minimal.

(engineers golden ears!)

Are you hearing this noise in your recordings? Noise down in the -92 to -96 range is not audible.

(the hearing competition ….)

On the end am I too anal about my 2000+ $ ???

RemyRAD Thu, 11/02/2006 - 21:41

So you're only now discovering that most everything is bull crap? You're talking about theoretical noise limits in the equipment. Even silicon based stuff has thermal noise. I don't care what kind of microphone preamps you have. I don't care what kind of converters you have, you will never find an environment that will give you much more than a 65 DB signal to noise ratio in real life so all of the specifications are basically moot. Unless you are a moron who is into the ultimate avant-garde and like recording absolutely nothing at all so you can listen to thermal noise? I don't think anybody will talk about how wonderful your production is because it's so quiet? "Oh listen to that nothing, isn't it wonderful?" Right.

If you are a good engineer and know what you're doing, your recordings will sound marvelous regardless of the mindless specifications which really don't mean anything. It's the sound that means something. So you have purchased yourself a nice piece of equipment that will give you nice results, reliable service for years to come, as opposed to a piece of crap equipment that will give you crap results. Be happy with what you have, it's good.

Your comments regarding the "engineer's ears" are ridiculous and unfounded. Those terrible engineers created API. Those terrible engineers created Rupert Neve. That awful equipment isn't worth owning so please send all of that stuff to me.

Now go out and make some nice recordings and stop worrying about stupid numbers, they mean nothing. Making recordings is an art like being a musician. It is neither clinical nor technical albeit technically oriented but for people who want to justify their existence, you can make it as clinical or technical as you desire. It won't make you a better engineer and nobody will care about your signal to noise ratio if you make good recordings.

Wonderful engineer
Ms. Remy Ann David

Groff Fri, 11/03/2006 - 13:17

RemyRAD wrote: So you're only now discovering that most everything is bull crap? ...

If DC offset comes from Behri-Behri converter you will tell that's bad because of crapy and cheap gear. That's ok. When it comes from Apogee .... suddenly things become more relative, tolerant, like nothing really matters at all and the one who reported it could be spec numbers obsessive moronic thermal noise listener with disputable re skills.

:shock:

My recorded waves are decentered, i did the test (the best i can / if it make any sense), found dc offset and reported.

From my experience i have found Rosetta 200 (i suppose it is/should be at least entry level of high end converter league) is not immune of DC offset. So DC offset is not privilege of cheap and crap gear only.

Except the question „do you get what you paid for?“ the one more appropriate for this topic i guess is:

What's the acceptable value/range of DC offset for high end AD?

Zilla Fri, 11/03/2006 - 15:02

Groff wrote: ...I did simple test (I’m not shore if the test is good or scientific, but it's a real life situation) I recorded „silence“, nothing plugged in... The result was the noise around -98 dB...it seems to me… well … higher than I expected.

I suspect you are getting those unexpected readings because of your test set-up. You should not test for noise and offset with your inputs unterminated. That is, you can't just leave the a/d inputs flapping in the breeze, unconnected. You should plug in a low z dummy source load to the inputs. 50-100 ohms should model most device outputs accurately enough.

Groff Fri, 11/03/2006 - 15:46

Thanks Zilla, I didn't try this but I could.

The input bias source resistor in the AD stage could make large noise in measurements so maybe the better way is to shorting all input pins to ground.

Regardless of my testing, recorded waves from mic and inst are somewhat out of the ∞ center line, and that's more problematic than noise floor by itself.

Zilla Fri, 11/03/2006 - 16:19

Either shorting or low z dummy should be good. This will help you indicate the amount of DC offset as it exists in the AD. You have probably already realized that even if the AD is clean, any device connected to it may introduce DC offset as well. Almost all devices have some amount of offset, even really expensive units. It is the degree/quantity of that offset that impacts performance.