As I only use digital sources, I'm looking at connecting my Buf32 directly to a power amp and use the Volumite for volume control. Has anybody tried this?
From what I understand, the Volumite controls volume by getting the DAC to shift bits around. As the whole thing is 32 bits, it is possible to 'shift' 8 bits without losing any resolution in the signal if the source is 24 bits, and 'shift' 16 bits if the source is 16 bits. At least, that's what I think... Is this more or less correct?

What I don't know is how volume can be cut without affecting the sound. Is there an way to simulate something like this? My main concern is that when I turn down the sound to reasonable levels with a heavy power amp I start to lose resolution.

That is correct, according to my understanding of it.



Another option...... once you have it all there in front of you, try optimising the gain of the output stage. Reduce the gain such that you stay in the top 'half' of the volume control. Pretty easy with the Buffalo32 - it is controlled by just a couple of optional resistors, which I think you could probably socket for initial testing.

Sort of, but this is really a bad description of what is happening when you control the volume digitally. If the volume was varied by shifting bits to the left and right, you would only have 16-steps for a CD signal, and 32-steps for a 32-bit signal. This is not at all how it works.

Each binary number represents the amplitude of a time slice of an analog waveform. So, the 16-bit binary number 0011010010011110 binary represents 13470 decimal, on a scale of 0-65535 (1111111111111111 binary). When adjusting the volume, you want a smaller scale image of the original waveform. So to reduce the volume by 10%, you take half of our original value (13470) and get 12123, which in binary is 0010111101011011. For the Buffalo, the volume is adjusted using 32-bit words and has 256 steps (0 to 4294967295 in steps of 16843009).

All of this does not take into account varying the volume logarithmically (I did it linearly for clarity).

If you "shift bits" to the left, each step would reduce the volume by half.

Essentially when you adjust the volume digitally, you get a scaled image of the original waveform, just as you would (hopefully) if you adjusted it in the analog realm. The main differences are that in the analog realm, it's more of an approximation of the original wave form (due to material variance, stray capacitance and inductance, etc in the pats involved), more noise from the resistors that must be used, and tracking error between channels.

In both schemes, you lose "resolution," but this does not mean binary resolution. It means you lose signal level. Since generally speaking, the noise level of an amplifier is constant, you have a lower signal to noise ratio with lower volume. Of course, if you volume control adds noise (analog), you get not only reduced signal level but increased noise (!), to the SN ratio is much worse.

Hope this helps clear things up a bit.

[EDIT] I totally forgot about mentioning the varying output impedance of a pot and it's effects on the next stage. Just more badness.

Edited by user Tuesday, October 20, 2009 7:26:24 AM(UTC)  | Reason: Not specified

Power amp noise is constant in both approaches. You volume control will not affect it. Noise from the DAC is infinitesimally small, especially if you are using balanced outputs. There is no noise from the interconnects, or none worth mentioning (0.00000001% or something).



JT will not really reduce output noise from the DAC, as the resistors in it will introduce more noise then in present in the DAC outputs. This is just a fact of life with resistors.

Brian,

Just to understand it a bit better, if you for example use foobar and a 16 bit wave file, output to the DAC at 32 bits and use the volume control in foobar, does that have the same result as using the Volumite? I think foobar internally processes at 32 bit so the "scaling of the original waveform" is all calculated at 32 bit and then sent to DAC.

Or is the scaling in the DAC itself much more sophisticated than a standard digital volume control operating at 32 bit?

Thanks

Wasn't it so that the internal volume control had an equivalent resolution of 52 bits?