I'm wondering if someone can give a definitive answer:

With the M-Audio Audiophile 24/96 (and I think other Deltas), two options are offered in the Control Panel applet, Hardware Settings tab under Variable Signal Levels/Outputs: "Consumer" and "-10dBV."

(1) Is this a digital change in volume, or is it truly a change in analog op-amp output?

(1a) If digital -- which is the "true" level? Is there a boost at the "Consumer" level, or a truncation/other cut in volume at the "-10dBV" causing some less significant bits to be lost?

(1b) If analog -- how is this done? Is there a relay or other circuit on the card that changes the output voltage of the op-amp? How would the drivers be able to trigger such a thing digitally?

(2) Is there any imaginable reason to suspect one output level might sound different or 'better' than another (regardless of volume)? Note that I would do an RMAA test to check dynamic range and such, except the "-10dBV" doesn't provide a high enough output volume to satisfy RMAA. I think I'm hearing some kind of improvement at -10dBV (when I increase the output level of my headphone amp in conjunction), but could easily be imagining it.

TIA...

Edit -- this is what the user manual says (not sure how it reflects on my question though, or what they mean by "least sensitive" and "most sensitive"):

My Terratec shows similar options for input and output ... see the image



From what I've understood, the most sensitive setting is the standard setting whereas the "lower" setting is (pure) analog attenuation of the input/output signal to obtain the optimum "working point" with different input/output hardware ... the digital signal is not being influenced by this settings. It therefore is not a boost in terms of amplification ... I think the OP amps will just get lower base voltage which ends up in lower overall amplificaton and thus lower output signal.

Overall sound quality should not be affected by this IMO

It's just a change in the analog output level, no digital gain/attenuation involved.



This very common in pro audio equipment, and I don't think difficult to trigger digitally. See, in most consumer cards (Creative, TurtleBeach, others), output levels can be controlled digitally even when the output gain is done in analog domain. It's done with analog output amplifiers whose gain is controlled digitally. For a single Consumer/-10 dBV output level switch, things should be easier. I don't know the details, but I believe there's no need to use relays for that.

Thanks (to both of you). I seem to be hearing a subtle difference in SQ, perhaps less distortion at -10dBV (very difficult to verify, however because of the different levels). It seems conceivable that a lower op-amp output might sound different -- either because of the op-amp itself behaving differently at a lower voltage, or the way my headphone amp is reacting to the quieter signal.

I don't know which setting to use... it's a matter of being very picky either way. Maybe I'll do some more comparative listening. I suppose some ABX testing would be possible, but level matching would be difficult and I'm inexperienced in getting stuff like that right.

It would be great if there were someone else who also had an Audiophile 24/96 and experienced at level matching in digital recordings willing to do a few ABX tests from recorded loopback of different music... but I won't ask anyone to do it, as it's my 'issue'. Unless they wanted to...

I'm thinking perhaps... if the analog hardware in the headphone amp is better than it is on the soundcard, I may be getting a benefit from doing more of the amplifying with the headphone amp, using the lower output setting from the soundcard.

AFAIK, my headphone amp has a "fixed" output level that's attenuated with the volume pot so perhaps it's a better idea to use closer to its full potential with a lower-level input signal. Am I completely off-base with this assumption, or does it make sense to anyone?

I'm not sure what you're asking, so I'll provide several different answers.

1) It's likely that your headphone amp is better than the op-amp that sits on the card. If nothing else, the voltage supply signal should be cleaner.

2) I don't think the amp will necessarily sound better JUST because you're not attenuating the sound as much (unless you feel that the pot affects the quality of the amplified signal). It will probably sound better because you're only running it through one amp and not two.

Hopefully I'M making myself clear.

You may bypass the whole volume potentiometer in the headphone amp and use the soundcard to control volume (just make sure you use 24bit out in this case) - it may sound better...

I know nothing about the M-audio card, but as far as the option in the TerraTec config panel, it controls the output (or input) level (for output the +4dB option outputs sound with a higher amplitude, for input it expects sound with a higher amplitude). It says so in the manual and I have verified it by simply switching between the two, which (at least for output) gives a very audible difference (in terms of volume, I couldn't say anything about quality).

The percieved increase in sound quality, apart from placebo (who wants to be a "consumer"?), is perhaps due to the output stage and/or the headphone amp exhibiting a non linearity, which leads to a different frequency response depending on how much amplifying each individual amplifier has to perform.

Maybe you could record a sine sweep playing around with the amount of amplication of each stage in a similar setup as you did with your recent tests. Level match them and compare.

This could help you (German!): http://www.heise.de/ct/ftp/projekte/fgang/. It's a little tool that outputs a frequency sweep through your soundcard and records it allowing you to compare different devices you put into the loop. Never tried it and don't know what language the program uses.

It must be placebo:

http://shs82web.home.comcast.net/Comparison.htm

Unless the headphones are responding differently in some way that the soundcard-amp-soundcard loopback isn't showing (but it seems highly unlikely). Ahh well, so it goes...