What exactly IS dc offset - and why is it important?

[Category 5]

I am thinking that it relates to the difference between the + and - voltage rails, and that it shifts the waveform above or below the median line.

But this is a guess. Anyone have a concrete answer?

Shane


P.S. Sorry to milk you all for knowledge so much. i am addicted!

 

[SonsOfThunder]

[quote author="Category 5"]But this is a guess.[/quote] Good guess, Shane. Offset will cause asymmetric clipping (among other things)

HTH!
Charlie

 

[Rogy]

If there is a positive DC offset, the positive side of a sine wave will reach the power supply voltage and clip before the negative side of the sine wave. This is called asymmetric clipping. Major problem except of the distortion when clipping is that you loose headroom, because your sine wave clips at a lower level than would have been the case when not having a DC offset on it.

Second problem with DC offset is that some audio gear nowadays is DC coupled; the DC offset at the input of each amplification stage (opamp etc) will be amplified by the same amount as the ac signal.

If a signal with a DC offset passes thru a potentiometer, this will cause cracks and noises when the potmeter value is changed. This is highly undesirable for audio devices.

Greetz,


Rogy

 

[Category 5]

Thanks guys...


So tell me this. I am measuring the DC offset of one of my a-p-i clones at within 2/10 of a millivolt at full gain, and most steps below (at the output of the op-amp btw).

Problem is, the DC offset at gain steps 3, 4, and 5 (12 position Grayhill) is about -6.5, -8.2, and -5.6. Right after I switch gains the offset goes to about -20mv and then gradually settles at the #'s above. I know it's normal for some op-amps to fluctuate with gain settings, so I chalk it up to being older 2520s.

The other module is within 5mv at all gain settings (I think it's the same switch positions that it fluctuates the most too).

Is -8mv anything to worry about? All I am sacrificing is a negligable amount of bandwidth, correct? Seeing as that I am running the op-amps at +/- 17.4 and going straight into my AD converter I assume that any DC offset in millivolts shouldn't affet me that much (i.e. my other gear will clip LONG before the preamp does).

Right?

Thanks,
Shane

 

[mcs]

8mV doesn't sound like a major problem for me.

It only becomes a problem if you amplify that signal in a DC-coupled amplifier. If you have a low-level signal with an 8mV offset, and then add 100x gain you will end-up with an 800mV offset...

Best regards,

Mikkel C. Simonsen

 

[SonsOfThunder]

Is -8mv anything to worry about?

Nah.

All I am sacrificing is a negligable amount of bandwidth, correct?

You aren't sacrificing any bandwidth, only head room.

The reason for the fluctuation in offset would be settling in the ckt. So when you switch gains, there may be some offset initially,then the circuit settles out to the final amount of offset, this would be normal.

HTH!
Charlie

 

[Category 5]

Sorry...I knew it wasn't bandwidth but instead headroom. A brain fart.

So I am correct that since this is at the output of the op-amp, and I'm not amplifying it further, that it should be negligable. I mean the preamp certainly has more headroom than my converters. Probably 30% more.

Shane

Also, what is the amount of voltage increase for an increase of 1db (just for reference).

 

[SonsOfThunder]

dB conversion table:
http://www.uneeda-audio.com/dbtbl.pdf
To the rest of yer question...its all good, dude!

HTH!
Charlie

 

[CJ]

8 mv was good back in the seventies, 8 uv is more like it today, depending on the gain. Some chips have a dc offset compensation scheme , usually pin 5 and somebody else.
If your amp is direct coupled, say like a Valley People pre, then the dc offset will get into the x-former, if there is one, in the following piece of equipment you are using.

 

[Category 5]

There is a DC offset compensation resistor between the - and + terminals, but I have set it to null the DC offset at maximum gain (on a 12 position switch). THe dc offset changes with gain settings, and is usually within 3mV, but at gain position 3 it is 6.5, 4 is 8.2, and 5 is 6.5 again. The rest of the steps are within 3mV.

The op-amp is a 1973 2520 and the preamp design is a SCA A12, but is pretty much the same circuit as a real 312.

The op-amps came as a pair, from the same 1973 console, and the other one stays within 4 or 5 mV at all gain settings. Both preamps sound fantastic.

Regarding your comment on 8uV being more likely today, keep in mind that this pre is based on a vintage 312 design. It is not a new design, nor is it based on ICs or today's super spec op-amps like the Millennia, Forsell, or John Hardy.

I know I could saturate the output transformer prematurely, but isn't 75-100mV more like the actual danger zone than 8mV?

Shane

 

[SonsOfThunder]

There is a DC offset compensation resistor between the - and + terminals

Hmm...somebody straighten me out on this. Data sheet at Dan Alexander says that pin 7 is trim. I have always thought that it was DC trim. Have I misunderstood all along?

From the data sheet: "Output offset may be nulled to zero by means of a 100k trim pot between the +/- V and T terminals. This is normally not necessary in audio applications."

And
"Input Offset Voltage:4mV typ., 10mV max."

But you could always do a little calculation on the DCR of the trafo and the offset and find the amount of power being dissiapted in the trafo... :grin:

Peace!
Charlie

 

[CJ]

Ahh, I see. Maybe some Melcor's like I just finished? Funny thing, I measured about 8mv on mine also.
There can be some confusion between gain trim and offset trim. The gain is controlled by varying the input resistor, so your ratio between the r2/r1 formula is changing the gain. There should be a terminal for this on the card, if you are still using it.
The offset trim is say a 4.7 k resistor if I remeber correctly, at least on my board.
here are some pics:
This part of the circuit sets the gain of the amp. I used a 100k pot, btw, not the 1k, which is shown on a few schematics, which I believe is wrong.

x
This is where you trim the dc offset:

OK, now for some more confusion. I have seen at least three different output arrangements for the API stuff.
Cap output, which means dc offset is not an issue.
Direct output, thatt is no cap, no resistor, just a piece of wire leading to the output tranny.
And a cap/resistor combo:
x

The good news is that I tried all three, and could hear no difference in the sound, so why not use the cap and be done with offset problems, ehh?
cj
z

 

[SonsOfThunder]

[quote author="CJ"]use the cap and be done with offset problems, ehh?[/quote] Hurray, CJ! I was thinking when I was looking at that schemo this afternoon that I remembered seeing or hearing of there being a cap between the OA and the trafo. But there was not one on the particular schemo I was gawking on.

Thanks for removing the veil! :thumb:

Peace!
Charlie