Digital vs Analog Volume Control

[Obagleyfreer]

Im currently wanting to buy an amplifier to go with my Stream Magic 6.

I've narrowed it down to two choices an integrated amplifier and using the SM6 as a source only.

or a Power amplifier and using the SM6 as a digital pre amp.

My only worry would be that the sound would be worse using the power amp as I would be using a digital volume control on the SM6.

So, would I notice a real drop in sound quality using digital volume control over anolog volume control?

regards

 

[Overdose]

Im currently wanting to buy an amplifier to go with my Stream Magic 6.

I've narrowed it down to two choices an integrated amplifier and using the SM6 as a source only.

or a Power amplifier and using the SM6 as a digital pre amp.

My only worry would be that the sound would be worse using the power amp as I would be using a digital volume control on the SM6.

So, would I notice a real drop in sound quality using digital volume control over anolog volume control?

regards

A 24 bit digital volume control will give you up to 48dB of attenuation without causing audible artifacts, or reducing the quality of playback below CD resolution (16 bits). I cannot imagine that the SM6 has less than 24bit DVC.

That is more than enough, so don't worry about any loss of quality.

 

[unsleepable]

Recently I did a test on this connecting my irDac directly to my Arcam P38, to see how it handles the volume digitally. It’s not an advertised or supported feature, but through USB it’s possible to control the volume. When the music was very, very low, there was clearly a loss of detail in comparison to using an analogue volume control. The irDac is managed as a 24-bit audio interface, so I assume this is the bit depth at which it attenuates the sound. At normal volume, it sounded great.

I also tried a NuForce DAC-80, which controls the volume digitally. While it’s not a keeper for a number of reasons, it’s very detailed at very low volume—a tad more so than using my A19 to control the volume. The DAC-80 uses 32-bit to attenuate the signal.

So I’d say that the first thing to take into account to determine if 24-bit is enough, is how powerful the power amp that you are going to use is. The more powerful, the more the volume will likely need to be attenuated. And therefore, the more important the bit-depth becomes. The P38 is 105 watts per channel, and when testing the computer was set at around -54 dB. When playing 16-bit audio and modifying the volume at 24-bit, there is 8-bit of headroom, which allows for a maximum attenuation of 48 dB without losing definition. More than that, and data is being lost.

On the other hand, another technique that is used with digital volume is dithering. You can find technical information about this elsewhere, but basically it’s a way of transforming the signal so that it sounds the same, even though there is less data. I’m sure that the irDac does not implement anything like this and simply chops off the extra data as it reduces the volume—after all, volume control is not a feature of the irDac. Since CA advertises the Stream Magic 6 as being able to work as a preamp, it might do volume differently.

Anyways, if you are going to use a 50 or 60 watts per channel power amp, dithering will not even matter, and I think you don’t need to worry about the digital volume. It should be alright. Now, whether the Stream Magic 6 sounds good as a preamp and has synergy with the power amp you want to use with it, is another matter.

 

[Obagleyfreer]

delete

 

[Obagleyfreer]

delete

 

[matt49]

The SM6 upsamples signals to 24 bits, so if you have a CD-quality (16-bit) signal, that leaves 8 bits for the DSP to work with. This is ample for full-range attenuation.

Matt

 

[davedotco]

Recently I did a test on this connecting my irDac directly to my Arcam P38, to see how it handles the volume digitally. It’s not an advertised or supported feature, but through USB it’s possible to control the volume. When the music was very, very low, there was clearly a loss of detail in comparison to using an analogue volume control. The irDac is managed as a 24-bit audio interface, so I assume this is the bit depth at which it attenuates the sound. At normal volume, it sounded great.

I also tried a NuForce DAC-80, which controls the volume digitally. While it’s not a keeper for a number of reasons, it’s very detailed at very low volume—a tad more so than using my A19 to control the volume. The DAC-80 uses 32-bit to attenuate the signal.

So I’d say that the first thing to take into account to determine if 24-bit is enough, is how powerful the power amp that you are going to use is. The more powerful, the more the volume will likely need to be attenuated. And therefore, the more important the bit-depth becomes. The P38 is 105 watts per channel, and when testing the computer was set at around -54 dB. When playing 16-bit audio and modifying the volume at 24-bit, there is 8-bit of headroom, which allows for a maximum attenuation of 48 dB without losing definition. More than that, and data is being lost.

On the other hand, another technique that is used with digital volume is dithering. You can find technical information about this elsewhere, but basically it’s a way of transforming the signal so that it sounds the same, even though there is less data. I’m sure that the irDac does not implement anything like this and simply chops off the extra data as it reduces the volume—after all, volume control is not a feature of the irDac. Since CA advertises the Stream Magic 6 as being able to work as a preamp, it might do volume differently.

Anyways, if you are going to use a 50 or 60 watts per channel power amp, dithering will not even matter, and I think you don’t need to worry about the digital volume. It should be alright. Now, whether the Stream Magic 6 sounds good as a preamp and has synergy with the power amp you want to use with it, is another matter.

This is correct apart from the highlighted section. The rated power of the amplifier is irrelevant in this instance, the important factor is the gain of the amplifier. Essentially this means the input sensitivity of the power amplifier.

For a variety of reasons manufacturers like to produce components with lots of gain, ie the volume gets loud very quickly at even quite modest settings of the volume control, this impresses the punter and sells product.

It is also a nonsense, poor volume control range, easy to overdrive etc, etc. You should strive for a balance gain setup, in this case choose a power amplifier that allows you to use the digital volume close to full volume for normal 'serious' listening, whatever that volume is for you. For this reason amplifiers with switchable or variable gain have a great advantage.

The CA651 is fixed gain, but is not overly sensitive at 1.5v for full output, unless you listen a lot a very low levels, it should be fine. Ask Richer to set one up with your streamer and see what the volume settings are like, if you normal listening levels are below 70-80%, think again.

 

[Obagleyfreer]

Recently I did a test on this connecting my irDac directly to my Arcam P38, to see how it handles the volume digitally. It’s not an advertised or supported feature, but through USB it’s possible to control the volume. When the music was very, very low, there was clearly a loss of detail in comparison to using an analogue volume control. The irDac is managed as a 24-bit audio interface, so I assume this is the bit depth at which it attenuates the sound. At normal volume, it sounded great.

I also tried a NuForce DAC-80, which controls the volume digitally. While it’s not a keeper for a number of reasons, it’s very detailed at very low volume—a tad more so than using my A19 to control the volume. The DAC-80 uses 32-bit to attenuate the signal.

So I’d say that the first thing to take into account to determine if 24-bit is enough, is how powerful the power amp that you are going to use is. The more powerful, the more the volume will likely need to be attenuated. And therefore, the more important the bit-depth becomes. The P38 is 105 watts per channel, and when testing the computer was set at around -54 dB. When playing 16-bit audio and modifying the volume at 24-bit, there is 8-bit of headroom, which allows for a maximum attenuation of 48 dB without losing definition. More than that, and data is being lost.

On the other hand, another technique that is used with digital volume is dithering. You can find technical information about this elsewhere, but basically it’s a way of transforming the signal so that it sounds the same, even though there is less data. I’m sure that the irDac does not implement anything like this and simply chops off the extra data as it reduces the volume—after all, volume control is not a feature of the irDac. Since CA advertises the Stream Magic 6 as being able to work as a preamp, it might do volume differently.

Anyways, if you are going to use a 50 or 60 watts per channel power amp, dithering will not even matter, and I think you don’t need to worry about the digital volume. It should be alright. Now, whether the Stream Magic 6 sounds good as a preamp and has synergy with the power amp you want to use with it, is another matter.

Thank you so much for such an in depth answer, I really appreciate it.

The Poweramp I will be using is the 651W and it was designed to work with the SM6.

It's 100w output.

I was watching this video on YouTube

https://www.youtube.com/watch?v=t03yHNgbZzs

and at 3m20s they mention about it being used as a digital pre amp with the 651W power amp.

They go on to say specifically that the volume is controlled using a DSP system and that this doesn't reduce bit rate or quality at all.

I'd be quite interested to know how they do this and how DSP works, but I can't info on it anywhere.

 

[Obagleyfreer]

Recently I did a test on this connecting my irDac directly to my Arcam P38, to see how it handles the volume digitally. It’s not an advertised or supported feature, but through USB it’s possible to control the volume. When the music was very, very low, there was clearly a loss of detail in comparison to using an analogue volume control. The irDac is managed as a 24-bit audio interface, so I assume this is the bit depth at which it attenuates the sound. At normal volume, it sounded great.

I also tried a NuForce DAC-80, which controls the volume digitally. While it’s not a keeper for a number of reasons, it’s very detailed at very low volume—a tad more so than using my A19 to control the volume. The DAC-80 uses 32-bit to attenuate the signal.

So I’d say that the first thing to take into account to determine if 24-bit is enough, is how powerful the power amp that you are going to use is. The more powerful, the more the volume will likely need to be attenuated. And therefore, the more important the bit-depth becomes. The P38 is 105 watts per channel, and when testing the computer was set at around -54 dB. When playing 16-bit audio and modifying the volume at 24-bit, there is 8-bit of headroom, which allows for a maximum attenuation of 48 dB without losing definition. More than that, and data is being lost.

On the other hand, another technique that is used with digital volume is dithering. You can find technical information about this elsewhere, but basically it’s a way of transforming the signal so that it sounds the same, even though there is less data. I’m sure that the irDac does not implement anything like this and simply chops off the extra data as it reduces the volume—after all, volume control is not a feature of the irDac. Since CA advertises the Stream Magic 6 as being able to work as a preamp, it might do volume differently.

Anyways, if you are going to use a 50 or 60 watts per channel power amp, dithering will not even matter, and I think you don’t need to worry about the digital volume. It should be alright. Now, whether the Stream Magic 6 sounds good as a preamp and has synergy with the power amp you want to use with it, is another matter.

This is correct apart from the highlighted section. The rated power of the amplifier is irrelevant in this instance, the important factor is the gain of the amplifier. Essentially this means the input sensitivity of the power amplifier.

For a variety of reasons manufacturers like to produce components with lots of gain, ie the volume gets loud very quickly at even quite modest settings of the volume control, this impresses the punter and sells product.

It is also a nonsense, poor volume control range, easy to overdrive etc, etc. You should strive for a balance gain setup, in this case choose a power amplifier that allows you to use the digital volume close to full volume for normal 'serious' listening, whatever that volume is for you. For this reason amplifiers with switchable or variable gain have a great advantage.

The CA651 is fixed gain, but is not overly sensitive at 1.5v for full output, unless you listen a lot a very low levels, it should be fine. Ask Richer to set one up with your streamer and see what the volume settings are like, if you normal listening levels are below 70-80%, think again.

So what you're saying is that it does actually reduce the bit rate?

Cambridge audio seem to think there is no loss of bit rate or quality.

My normal listening levels are probably at about 12 oclock or 50%.

So should I perhaps consider the intergrated 651A instead and leave the SM6 in source mode?

It's quite hard for me to decide without having the 2 units at the same time to demo against each other.

 

[Tannoyed]

I would opt for an analogue volume control wherever possible. Perhaps my experiences havenot been with the best euipment but crosstalk seems to me to be the biggest problem with the electronic option. A nice double gang potentiometer is my preference. There will be some coupling because of stray capacitance but at audio frequencies this will not amount to much. Mind you there is so much large scale integration these days that the means of controlling the volume might be the least of our problems! I'm old and used to old things so am probably out of touch. Just a thought.

 

[davedotco]

So what you're saying is that it does actually reduce the bit rate?

Cambridge audio seem to think there is no loss of bit rate or quality.

My normal listening levels are probably at about 12 oclock or 50%.

So should I perhaps consider the intergrated 651A instead and leave the SM6 in source mode?

It's quite hard for me to decide without having the 2 units at the same time to demo against each other.

It would be a unique product if it did not. but I can not say for certain.

Generally processors (DSP) work at 24bit resolution or better so volume reduction has no realistic impact on sound quality, reducing the volume by 8 bits, 48dB, will be pretty quiet, at this level of attenuation other factors are more important than the theoretical loss of bits.

It is worth remembering that many systems, working off regular mains in a normal home environment will struggle to resolve 16 bits anyway, if you are able to set up your system so that for 'normal' listening levels, your volume is close to maximum, you will be fine.

You also need to understand that the position of the volume control does not give any indication of the power being delivered by the amplifier, the way the volume control works is entirely at the whim of the designer.

 

[unsleepable]

So what you're saying is that it does actually reduce the bit rate?

Cambridge audio seem to think there is no loss of bit rate or quality.

My normal listening levels are probably at about 12 oclock or 50%.

So should I perhaps consider the intergrated 651A instead and leave the SM6 in source mode?

It's quite hard for me to decide without having the 2 units at the same time to demo against each other.

If anything, the bit-rate is increasing when transforming the data from 16-bit to 24-bit. It's the resolution of each sample that is reduced when attenuating the signal beyond the threshold of the available headroom. And yes, it will happen at very low volume.

But there are well-known hi-fi vendors that do this, such as Bel Canto. Their DACs perform volume attenuation digitally at 24-bit, so they cope with data loss using dithering. And people seem to be happy with that. Digital attenuation at 32-bit is of course a simpler solution.

The problem here is that Cambridge Audio does not provide detailed information on how their digital volume attenuation works, so they may, or may not, use dithering to preserve the sound at low volume. But if you don't usually play music so very low, it will be more than ok.

 

[unsleepable]

I would opt for an analogue volume control wherever possible. Perhaps my experiences havenot been with the best euipment but crosstalk seems to me to be the biggest problem with the electronic option. A nice double gang potentiometer is my preference. There will be some coupling because of stray capacitance but at audio frequencies this will not amount to much. Mind you there is so much large scale integration these days that the means of controlling the volume might be the least of our problems! I'm old and used to old things so am probably out of touch. Just a thought.

When I tested the irDac direct into the power amp, I was surprised to find that there was absolutely no audible cross-over, vs a bit when using the A19 to control the volume. I tested this unplugging one speaker, digitally balancing all the sound to the unplugged speaker, raising the volume to the maximum and listening to the other speaker. Some audio was listenable when using the A19, vs none when using only the irDac. Also, background noise was much lower only with the irDac.

As for analogue volume attenuator, my favorite choice is the resistor network type. For what I expect of volume controls, I find that they way ahead of potentiometers, and thankfully hi-fi vendors implement them more and more.

 

[Obagleyfreer]

I just received an email from Cambridge Audio and they said this about the Stream Magic 6 volume control:

"We don’t use ‘bit reduction’ like many other manufacturers do in their digital pre amps.

The 851D/SM6/851C/DM Plus keeps the entire signal in the digital domain up until the D/A converters. In order to achieve volume control, we use some really clever Digital Signal Processing. We first convert the signal to 32 bits, then rescale, this means you have -66dB of attenuation before you ever start to lose bit depth.

It gives us the kind of performance only previously possible using an analogue attenuator, but without issues like noise creeping into the signal path."

Sounds pretty good to me!

 

[MajorFubar]

There cannot be crosstalk in digital volume controls. It's just not possible. Digital doesn't work like that. But that doesn't mean the electrical components elsewhere in the circuitry can't suffer from crosstalk. But it will have nothing to do with digital volume controls.

 

[davedotco]

I just received an email from Cambridge Audio and they said this about the Stream Magic 6 volume control:

"We don’t use ‘bit reduction’ like many other manufacturers do in their digital pre amps.

The 851D/SM6/851C/DM Plus keeps the entire signal in the digital domain up until the D/A converters. In order to achieve volume control, we use some really clever Digital Signal Processing. We first convert the signal to 32 bits, then rescale, this means you have -66dB of attenuation before you ever start to lose bit depth.

It gives us the kind of performance only previously possible using an analogue attenuator, but without issues like noise creeping into the signal path."

Sounds pretty good to me!

That simply does not make sense. Many dacs process at 32 bit resolution, some even higher. They talk about bit reduction and the rescaling as if they are different things, they are not.

The reality is that a competent digital volume, even one working at 24 bits will not affect sound quality in the slightest if used as intended, towards the top end of it's scale. There are far more important things to worry about with this set-up.

 

[unsleepable]

With "bit reduction" they probably mean losing significant bits.

So they don't implement dithering, but if the signal is attenuated at 32-bit, it wouldn't be necessary. What piques my curiosity is that once you have a 32-bit signal, you need a 32-bit DAC to convert it to audio—regardless of the bit-depth of the inputs. And the Stream Magic 6 seems to use the Wolfson WM8740SEDS chip, which works at 24-bit. If the DAC does not operate at 32-bit, you are again chopping off 8 bits.

So yes, it doesn't make much sense… It sounds a bit like the typical marketing BS. Probably they didn't expect it would end up in the What Hi-Fi forum! *biggrin*

Anyways, I think we may also be overseeing that analogue attenuators have their own inconveniences and limits, such as background noise and signal-to-noise ratios. Background noise would have to be very low and SNR superb to enjoy the full resolution of even 16-bit audio at low volume.

 

[Obagleyfreer]

I just received an email from Cambridge Audio and they said this about the Stream Magic 6 volume control:

"We don’t use ‘bit reduction’ like many other manufacturers do in their digital pre amps.

The 851D/SM6/851C/DM Plus keeps the entire signal in the digital domain up until the D/A converters. In order to achieve volume control, we use some really clever Digital Signal Processing. We first convert the signal to 32 bits, then rescale, this means you have -66dB of attenuation before you ever start to lose bit depth.

It gives us the kind of performance only previously possible using an analogue attenuator, but without issues like noise creeping into the signal path."

Sounds pretty good to me!

That simply does not make sense. Many dacs process at 32 bit resolution, some even higher. They talk about bit reduction and the rescaling as if they are different things, they are not.

The reality is that a competent digital volume, even one working at 24 bits will not affect sound quality in the slightest if used as intended, towards the top end of it's scale. There are far more important things to worry about with this set-up.

The thing that I'm concerned about is that I don't really listen to music super loud, only ever at around 50% at the most. So would I be best starring clear of using a digital pre amp?
Or am I just being pedantic and in the real world will I not even notice a reduction in quality?

 

[andyjm]

With "bit reduction" they probably mean losing significant bits.

So they don't implement dithering, but if the signal is attenuated at 32-bit, it wouldn't be necessary. What piques my curiosity is that once you have a 32-bit signal, you need a 32-bit DAC to convert it to audio—regardless of the bit-depth of the inputs. And the Stream Magic 6 seems to use the Wolfson WM8740SEDS chip, which works at 24-bit. If the DAC does not operate at 32-bit, you are again chopping off 8 bits.

High bit depth is used in DSP processing to avoid truncation and lack of precision due to integer maths. Taking a 16 bit sample, doing calcs at 32 or 64 bits and then reducing the result back to 16 bits for output to a DAC does not involve 'chopping off bits' in the sense that any of the precision in the original 16 bits is lost.

 

[Obagleyfreer]

With "bit reduction" they probably mean losing significant bits.

So they don't implement dithering, but if the signal is attenuated at 32-bit, it wouldn't be necessary. What piques my curiosity is that once you have a 32-bit signal, you need a 32-bit DAC to convert it to audio—regardless of the bit-depth of the inputs. And the Stream Magic 6 seems to use the Wolfson WM8740SEDS chip, which works at 24-bit. If the DAC does not operate at 32-bit, you are again chopping off 8 bits.

High bit depth is used in DSP processing to avoid truncation and lack of precision due to integer maths.  Taking a 16 bit sample, doing calcs at 32 or 64 bits and then reducing the result back to 16 bits for output to a DAC does not involve 'chopping off bits' in the sense that any of the precision in the original 16 bits is lost.?

 

But most of the music I play is 24bit, so I will notice a loss in quality?

 

[Overdose]

But most of the music I play is 24bit, so I will notice a loss in quality?

No. You are worrying unduly.

 

[davedotco]

With "bit reduction" they probably mean losing significant bits.

So they don't implement dithering, but if the signal is attenuated at 32-bit, it wouldn't be necessary. What piques my curiosity is that once you have a 32-bit signal, you need a 32-bit DAC to convert it to audio—regardless of the bit-depth of the inputs. And the Stream Magic 6 seems to use the Wolfson WM8740SEDS chip, which works at 24-bit. If the DAC does not operate at 32-bit, you are again chopping off 8 bits.

High bit depth is used in DSP processing to avoid truncation and lack of precision due to integer maths. Taking a 16 bit sample, doing calcs at 32 or 64 bits and then reducing the result back to 16 bits for output to a DAC does not involve 'chopping off bits' in the sense that any of the precision in the original 16 bits is lost.

But most of the music I play is 24bit, so I will notice a loss in quality?

The very first thing you need to understand is that there is no hi-fi system on the planet that even gets close to resolving a 24 bit signal. A decent mid-fi system of the kind you are building will, in a domestic situation (with all the electronic 'polution' and poor quality mains), just about resolve a 16 bit signal. This requires a system s/n of 96dB or better, less common than you might think in the real world.

There may be some advantages in using 24 bit downloads but that advantage, if present at all, is nothing to do with bit depth, it is usually a function of better or more careful mastering.

It simply is not an issue, it is more important to optimise gain throughout your system, (to keep noise as low as possible) than to fret about a digital volume control.

 

[Obagleyfreer]

But most of the music I play is 24bit, so I will notice a loss in quality?

No. You are worrying unduly.

That's good to hear, thanks.
I know hi fi isn't a numbers game, but sometimes it's hard to ignore it.
Especially when you buy something to specifically listen to high res music and hear the bit rate will be reduced.
I think I'm just going to bite the bullet and buy the power amp, using SM6 as a pre.
Thank you to everyone for all your help/input.

 

[Obagleyfreer]

With "bit reduction" they probably mean losing significant bits.

So they don't implement dithering, but if the signal is attenuated at 32-bit, it wouldn't be necessary. What piques my curiosity is that once you have a 32-bit signal, you need a 32-bit DAC to convert it to audio—regardless of the bit-depth of the inputs. And the Stream Magic 6 seems to use the Wolfson WM8740SEDS chip, which works at 24-bit. If the DAC does not operate at 32-bit, you are again chopping off 8 bits.

High bit depth is used in DSP processing to avoid truncation and lack of precision due to integer maths.  Taking a 16 bit sample, doing calcs at 32 or 64 bits and then reducing the result back to 16 bits for output to a DAC does not involve 'chopping off bits' in the sense that any of the precision in the original 16 bits is lost.?

?

But most of the music I play is 24bit, so I will notice a loss in quality?

The very first thing you need to understand is that there is no hi-fi system on the planet that even gets close to resolving a 24 bit signal. A decent mid-fi system of the kind you are building will, in a domestic situation (with all the electronic 'polution' and poor quality mains), just about resolve a 16 bit signal. This requires a system s/n of 96dB or better, less common than you might think in the real world.

There may be some advantages in using 24 bit downloads but that advantage, if present at all, is nothing to do with bit depth, it is usually a function of better or more careful mastering.

It simply is not an issue, it is more important to optimise gain throughout your system, (to keep noise as low as possible) than to fret about a digital volume control.

Thank you. This is all new to me, I'm still understanding everything.
I really appreciate your help and info 🙂

 

[davedotco]

Thank you. This is all new to me, I'm still understanding everything. I really appreciate your help and info 🙂

No problem.

The music industry is struggling for sales and are looking at 'new' technology to generate interest.

The recording industry regularly uses 24/96 or better for recording as this allows them to modify and manipulate the signal without affecting the 16 or so 'significant' bits we all listen to at home.

If a 24 bit download sounds better than a 16 bit download it will be because the mastering has been done with more care for a 'premium' product, care that could have been taken with the 16 bit product in the first place. Ie using the same quality master 16 bit is effectively identical to 24 bit, but as you say, 'numbers' impress.