Negative effects of digital gain & need Squeezebox info!

[IEMCrazy]

I'm hoping the science and measurements crowd can help with a bit of a dilemma I'm in with regard to how to configure my gear, and at what point I'm reducing quality.
 
I've been running my Squeezebox Touch at "Fixed 100% volume" to prevent gain manipulation in the digital domain.  However compared to my prior digital source, the SBT seems to be very "hot".  On my Schiit/Lyr/HE-400/K702/HD650 combos, I barely get the volume knob over 7:00 to 8:00 for normal listening, compared with 9:00 to 10:00 for my prior source.  As far as I know, amps generally don't perform their best when attenuating all the way down to the bottom of the dial, though that's a different question.
 
My bigger issue starts now that I have an Objective 2 amp and Denon D5000's.  I have it set up with stock 2.5x gain configuration, and if I use it as a portable, that's an appropriate low gain setting so I don't want to clip the resistor to bring it to 1.5 or 1.0x.  (I have a set of Y-Cables from Bifrost DAC feeding O2 and Lyr.)
 
The very hot SBT at 100% and the standard 2Vrms out from Bifrost resulting from it gives me MAYBE one degree of rotation on the O2 volume pot.  While this is barely sufficient to have any volume control, worse, there's horrible channel imbalance down at the bottom of the O2's pot, forcing me to drive the music much too loud for prolonged listening to get the channels balanced.  Additionally I noticed a track that was so hot it was either driving the DAC into clipping or overloading the input of the O2 (which shouldn't even happen at low gain.)  Something tells me the SBT 100% volume is doing things it shouldn't be doing anyway.
 
This has forced me to drop my SBT volume down to at least 75% to get appropriate volume control on the amp.  That means I'm getting into the realm of digital gain control in the digital domain, and the associated resolution loss.
 
This is where the need for SBT info comes in.  Does anyone know (either via knowing the tech specs or by having plugged it into an AVR/pre-pro that displays the current bitrate) does the SBT output in the native bit depth/sample-rate of the source file, or does it universally convert bitrate to 24 internally no matter what?   I know the Bifrost doesn't click the relay when changing digital gain, meaning it's either doing 16->16bit conversion (really bad), or it's doing 16->24bit upconversion ALWAYS and applying gain on top of that, meaning with the extra headroom, I should have quite a bit of digital gain reduction headroom before quantization hits the noise floor. 
 
So for the measurement geeks out here, or those who know a thing or two about Squeezebox Touch coax outputs, am I suffering resolution loss or quantization artifacts by doing digital domain gain adjustment?  Are there any alternate solutions I'm not thinking of other than clipping resistors in the O2?

 

[stv014]

If you are already using the O2 on low gain, you can get even lower gain by some simple modding (i.e. 1x and 2.5x instead of 6.5x and 2.5x), clipping two resistors.

 

[IEMCrazy]

Quote:

If you are already using the O2 on low gain, you can get even lower gain by some simple modding (i.e. 1x and 2.5x instead of 6.5x and 2.5x), clipping two resistors.

 
Yes, I realize that.  Ultimately it could come down to that, but I'd like to avoid that if possible since the default setup is still good for other sources.  I was mostly interested in more details from those in the know of the damage, or lack thereof, of digital domain gain modification as it relates to the output depth of the SBT.
 
I of course "hear a reduction in detail", but I'd be interested in some more real info on whether that's pure audiophile voodoo, real damage to the resolution, or possible a combination of reality not mixed with damage, but with a difference between the low band and the rest of the volume dial (given volume imbalance, it's very possible it has frequency alteration issues at the very bottom as well.)
 
Unfortunately there's not a huge crowd that gathers here at the Sound Science part of the forums

 

[anetode]

If the SBT upsamples all digital input then there shouldn't be any appreciable loss of detail, only of the extreme ends of dynamic range. If not, then unfortunately you may be looking at the lesser of a handful of evils, ranging from changing the O2's gain to selling/buying new equipment. Ultimately it doesn't much matter if what you're hearing is a real problem or voodoo, you decide on how much it bothers you and whether a drastic change is necessary.

One thing I do know from experience is that the balance differences introduced at the lowest range of the pot are going to be more annoying than truncating a couple of bits.

 

[wakibaki]

Just install a permanent attenuator in the form of a resistor (in each channel) in the signal path of the SBT. The O2 has a 10k input impedance, try a 10k resistor, which will halve the starting volume. If it's not enough, try a bigger resistor. A 10k resistor will increase the idling hiss (Johnson noise) by 1.4 (root 2) times, but this will be inaudible with music playing, and may be inaudible anyway depending on the sensitivity of your phones.
 
w

 

[IEMCrazy]

Quote:

If the SBT upsamples all digital input then there shouldn't be any appreciable loss of detail, only of the extreme ends of dynamic range. If not, then unfortunately you may be looking at the lesser of a handful of evils, ranging from changing the O2's gain to selling/buying new equipment. Ultimately it doesn't much matter if what you're hearing is a real problem or voodoo, you decide on how much it bothers you and whether a drastic change is necessary.
One thing I do know from experience is that the balance differences introduced at the lowest range of the pot are going to be more annoying than truncating a couple of bits.

 
So the only detail remaining is if the SBT upsamples all output or not.  if it does it's not "bit-perfect" as some claim, but then, Logitech never said anything about "bit-perfect."  If it's really working in 24-bit space, one could drop the gain quite far I'd think.  I supposed 24-bit audio would suffer a bit more (presently I have none in my collection), but even then it wouldn't be bad.  In 24-bit space the pre-occupation against digital modification is a bit crazy, considering the 16-bit master was the result of digital processing in 24-bit space in most cases to begin with....
 
Quote:

Just install a permanent attenuator in the form of a resistor (in each channel) in the signal path of the SBT. The O2 has a 10k input impedance, try a 10k resistor, which will halve the starting volume. If it's not enough, try a bigger resistor. A 10k resistor will increase the idling hiss (Johnson noise) by 1.4 (root 2) times, but this will be inaudible with music playing, and may be inaudible anyway depending on the sensitivity of your phones.
 
w

 
Well an attenuator resistor wouldn't help the DIGITAL output of the SBT, so you're talking about an attenuator from the Bifrost or the Lyr.   That gets kind of messy.  Also the primary headphones bothered the most by the level of output are the Denons (the planars can barely scrape by without issue), and the denons are amazingly sensitive, so we'd be talking about adding a hiss.  Which is much worse than whatever slight resolution loss may or may not be occurring.
 
Personally, compared to other digital sources I've used, I still believe the SBT to be very strangely hot.  I wonder if it is padding to 24-bit space and that is having an effect on the gain by definition due to the greater available dynamic range.