why does digital volume control cause loss in info


I have been trying to understand why controlling volume in the digital domain means that sonic information gets lost. Obviously, I'm not super techincal, but I'm not totally lacking in technical understanding. Can someone give me a straightforward explanation, or point me to a reasonably easy to understand reference?

Thanks!

--dan
dgaylin
It doesn't. Digital volume control is commonly implemented using a module that is essentially a digitally controlled stepped attenuator. The signal is analog. The notion of "dropping bits" is pure myth.
Eldartford is referring to an analog volume control that is controlled by digitally operated motor. It's used to provide remote control capability and it doesn't suffer from bit loss, since the volume control is complete in the analog domain. A true digital volume control can suffer from bit loss. IT IS NOT A MYTH! The amount of and its audibility depends on the implementation.

Here's Wadia's explanation of how digital volume controls work.
Hi Onhwy61. Thanks. I've read Wadia's explanation, and didn't get it. In simple terms why do you drop a bit to attenuate the volume digitally?
" In simple terms why do you drop a bit to attenuate the volume digitally"

i suspect because, assuming all bits in each sample are used at maximum volume (amplitude), it takes fewer bits to represent each sample with the same resolution in the same digital audio stream at lower volumes.
Mapman has the right idea. I would slightly reword his statement as follows:

"If all bits in each sample are required to represent a signal at the maximum possible volume, fewer bits will be available to represent each sample at lower volumes."

The higher order (most significant) bits will be fixed in a single state (e.g., 0, as opposed to being able to change between 0 and 1), in order to represent the fact that the signal does not extend above a certain volume.

It should be kept in mind, though, that in concept the loss of resolution with a digital volume control at lower volumes is no different than what occurs with an analog volume control, although there will certainly be differences in degree.

In both cases, the signal level is being reduced relative to the noise "floor" of the parts of the system that are at and "after" the volume control. Simplifying slightly, in the case of a digital volume control the noise floor (referred to as "quantization noise") corresponds to the resolution that is provided by the least significant bit. In the case of an analog volume control the noise floor corresponds to any analog noise that is introduced into the signal path by the circuitry at or after the volume control (noise that is introduced "before" the volume control will be attenuated by the volume control by the same amount as the signal is attenuated, and so the ratio of signal to noise will not change).

In practice, though, noise introduced in the analog signal path following the volume control is usually pretty minimal in a quality system, and the spectral characteristics of analog noise are usually such that the ear can discriminate between the noise and signal levels that are significantly below the noise level. While the 16 bits of the redbook cd standard are significantly less than ideal to begin with, especially considering that two or three of those bits may be thrown away in the recording process to prevent the possibility of clipping the recorder.

Regards,
-- Al
VERY HELPFUL! Thanks Al (and everyone else too).

--dan
The problem with digital volume control is that not all algorithms properly address quantization errors

You need to apply dither in order to randomize quantization errors.

Chances are that the CD that you buy has had dither applied. Now if you modify the bits digitally and re-quantize it (at a new lower digital volume level) then you need to dither it before hand in order to avoid introducing non-random quantization errors.

iTunes applies dither in the recent versions - so you can be confident that the volume control is as good as analog. In older versions it was poor - it just threw away bits.
There is a common digital volume control chip, used in my McCormack MAP1 six channel preamp, which does 1/2 dB steps. It is a stepped attenuator, and no motor is involved. If such a chip follows the D/A conversion no bits are lost. If an analog output is being produced there is no reason to use another implementation.

Now, if you want to have a digital output and attenuate the signal there will be a loss of resolution. No matter how you process the data a 12 bit output obviously has less resolution than a 16 bit output.
Shadorne: Thanks. Not sure, though, that it's fair to say that properly applied dither will necessarily make a digital volume control "as good as analog." It will certainly reduce the subjective objectionability of quantization noise. But I suspect that it could not adequately compensate for the loss of say 4 or 5 bits, corresponding to a volume control setting that is 24 or 30 db below maximum.

Also, it would seem expectable that in a standalone audio component, that most likely does not have the computing horsepower of a general purpose computer, and that has to calculate and apply the digital volume control function and the corresponding dither on-the-fly, in real time, for any arbitrary volume setting, that optimal dithering and noise shaping is not necessarily going to happen.
ElDartford: If an analog output is being produced there is no reason to use another implementation.
Cost, perhaps? In other words, reducing cost by eliminating that chip altogether, and incorporating the digital volume control function within a gate array or other custom digital chip that is required for other purposes.

Regards,
-- Al

Shadorne: Thanks. Not sure, though, that it's fair to say that properly applied dither will necessarily make a digital volume control "as good as analog." It will certainly reduce the subjective objectionability of quantization noise. But I suspect that it could not adequately compensate for the loss of say 4 or 5 bits, corresponding to a volume control setting that is 24 or 30 db below maximum.

You have to consider that most home audio systems have a dynamic range of only about 60 to 70 db to begin with.

Think about a typical noise floor of around 30 db SPL and then a typical max undistorted SPL of 100 db SPL (played at maximum volume)

So, with this perspective, either digital or analog volume of a 96 db dynamic range CD is more than good enough. Most systems can't even exploit a well recorded CD anyway.

When you get into pro equipment with 120 db SPL continuous output at less than 0.3 THD over the entire frequency range then you can begin to benefit from 24 bit technology (in an audible sense when playing back music)

So from my perspective the benefits of analog over well implemented digital volume control become theoretical in most cases (as you cannot actually hear it in a real room although you might measure it with an oscilloscope)

So from my perspective the benefits of analog over well implemented digital volume control become theoretical in most cases (as you cannot actually hear it in a real room although you might measure it with an oscilloscope)
Shadorne

Shadorne, Wow! That statement blew me away.
On a forum where members argue that 3 feet of special power cord can make a huge sonic difference, measurements be damned. That statement surely constitutes blasphemy. I applaud you!
Almarg...I know about this chip because one failed in my MAP1 and was replaced (on waranty). Sorry I can't cite the exact part number. There are three of these 2-channel chips in the six channel preamp, so they can't cost that much.
Eldartford,

You are probably thinking of the Crystal CS3310 chip it is common in many devices.
I took a quick look through the CS3310 datasheet. It looks like a pretty nice device, although I'd expect that designing with it, and achieving good results, would be a non-trivial challenge because of the presence of both digital and analog signals and power on the same device. And I note that its datasheet was initially released in 1991, so there may be newer and better comparable devices available now.

Digikey charges $6.57 for it in quantities of 1000. I believe that a generally accepted rough rule of thumb for equating parts cost with final assembly selling price is to apply a factor of 5. So it would have about a $33 impact on the selling price of an audio component, or perhaps say $50 to $100 factoring in possible requirements for additional surrounding circuitry to interface to it, and additional circuit board real estate that may be required. Not a huge cost, but certainly significant.

A purely digital volume control, on the other hand, could perhaps be implemented within a custom digital chip that is required in the design for other purposes, resulting in negligible impact on recurring cost.

Thanks for pointing out this device, ElDartford and Shadorne.

Regards,
-- Al
It should be kept in mind, though, that in concept the loss of resolution with a digital volume control at lower volumes is no different than what occurs with an analog volume control, although there will certainly be differences in degree.
Al, the main difference is that the analog noise floor is uncorrelated noise (not related to the signal), and the quantization error of a linear PCM system at lower levels is strictly correlated to the signal, so is audible as distortion, not simply noise.

But on the other hand, virtually all DAC chips these days will accept a 24-bit input . . . so in practice if we have 16-bit source material, a digitally-implemented volume control can provide about 48dB (6dB per bit) of attenuation range without adding any additional quantization error, simply because the DAC has more resolution than the source. And 48dB is pretty workable, if the gain structure of the rest of the system is well designed. There will of course still be the DAC's uncorrelated noise floor on top of this, which is the same type as that produced by analog methods.

Now regarding the CS3310, which I have tested extensively . . . it's a very good "off the shelf" digitally-controlled analog attenuator. Its main limitations are that since it's a monolithic circuit, there are some significant limitations to the quality of the resistors that can be fabricated on the same die as the rest of the chip, as opposed to discrete resistors. It has an onboard opamp with gain-ranging, which doesn't provide a constant bandwidth and transient-response regardless of volume-control setting. And while its noise performance is pretty good (though not nearly as good as a low-impedance discrete-resistor attenuator), it's limited to +/- 5v power supplies, so there's not a lot of headroom if the designer wants to optimize the gain structure for lower noise.
Excellent comments, Kirk. Thanks.

Re your first point, to make sure its clear to others who may read this I'll add that the fundamental purpose of dither is to convert quantization distortion into low level analog-like broadband noise. Or even better, into "shaped" noise, that predominantly occupies parts of the spectrum to which the ear is relatively insensitive.

Re your second point, which I hadn't thought of when I wrote my earlier posts, that would say that the bad rap digital volume controls justifiably got a decade or two ago would seem to be much less applicable, and in some cases completely inapplicable, to today's designs.

Best regards,
-- Al
Have you noticed that the lowest settings on digital volume controls actually play the music pretty loud for the lowest setting? For example going from -50 dB (mute) to -49.5 dB is not a tiny little increase in volume. It's something that's a bit too loud for 2 a.m. in the morning drifting off to sleep music. I've noticed this behavior on several digital volume controls.