is Naim Pre-Amp necessary with i-tunes?

[jcbrum]

Your comments regarding alac are agreed Nik.

It seems Jack has his W7 sound system controlled by a Realtek device driver which was set to output 24/48 by default, so bit word-length and sample rate translation will be operating on all songs which are not in that format, and 'bit-perfect' transmission will be lost.

This is nothing to do with iTunes, nor digital volume control, but is simply due to the design and configuration of his computer, and the operating system.

Probably he could alter it if he wished to do so, but he seems happy with the sound it makes, and why not ? - I doubt he would hear much difference if he did.

On a W7 system, digital volume control is doing nothing compared to the translations his Realtek driver is doing.

JC.

 

[Nik]

I remember that thread Joe. I did have a little play with it but didn't study it closely.

I think it's generally accepted now that high bit rate compressed files are audibly undetectable from uncompressed ones for ordinary use. There are a few 'die-hards' and 'golden-ears' who can't come to terms with that though. They usually fail a proper blind test.

What I am certain of though, is that if you up-sample it doesn't improve the sound quality beyond what was available from the original format. So if you start from a CD you can play it at 16/44 and get the best available results.

These days many new recordings are made typically at 24/96, and I can't see any real reason to alter that for release, except to significantly reduce the file size, where that is a desirable feature. In that scenario, compression is very valuable.

I did quite a bit of research a few years ago, into Reference Recordings HRx DVD's which are a non-standard format using 176.4 sampling rate, obviously because they were old recordings developed from modified RB CD format, which used to be their specialist area. This was to provide some advice to Keith at Purite Audio after he had been to an American Audio show (CES iirc).

I came to the conclusion that a standard Mac running suitable software was the cheapest way to play them and all the special 'trick-shit' PC's and expensive sound cards weren't really necessary. All that was required was a suitable DAC to handle the bit and sample rate.

Overwhelmingly what makes a 'killer sound' is the quality of the recording in the first place, almost regardless of format, and of course, as you know, good loudspeakers.



JC.

As a Mac user myself, I'd like to read that in full if you have a link somewhere.

Speakers aren't everything, though. Amplifiers can sound very different too, despite what the measurements say.

 

[prowla]

I would NEVER run a computer directly into a power amp; it is all too easy for an update/reinstall/something else to set the volume to max.

 

[Nik]

Your comments regarding alac are agreed Nik.

It seems Jack has his W7 sound system controlled by a Realtek device driver which was set to output 24/48 by default, so bit word-length and sample rate translation will be operating on all songs which are not in that format, and 'bit-perfect' transmission will be lost.

This is nothing to do with iTunes, nor digital volume control, but is simply due to the design and configuration of his computer, and the operating system.

Probably he could alter it if he wished to do so, but he seems happy with the sound it makes, and why not ? - I doubt he would hear much difference if he did.

On a W7 system, digital volume control is doing nothing compared to the translations his Realtek driver is doing.

JC.

Again, I agree. 48 and 44.1 don't have simple common factors so the conversion must be "messy". OTOH it could be that the DAC is able to produce a better analogue result because there's more "room" above the upper hearing cut-off (24 kHz vs 22 kHz) to implement the brick-wall filter (or is the difference too small to matter?).

 

[Nik]

I would NEVER run a computer directly into a power amp; it is all too easy for an update/reinstall/something else to set the volume to max.

That would be scary! Good point.

 

[jcbrum]

I have been doing it that way for about five or six years now, .... with two provisoes ...

1. ... on AVI ADM9 systems (sorry chaps ), they have a built-in, digitally remote controlled, analogue, level control, as part of the dac -> pre-amp -> active crossover, with mute and input switching also available on the remote controller. So the computer can always be run 'flat out'.

2. ... on my other systems which use amps running flat out permanently connected to loudspeakers, I am careful to choose amps which don't clip unless overdriven. The maximum sound output is then governed by the loudspeaker ('cause that's what decides the current drawn) and providing you realise straightaway what is occurring, they should be able to stand full volume briefly.

The only time I damaged a loudspeaker was with a Cyrus amp, which when delivering full volume went straight into severe clipping, and more or less put DC across the speakers.

I just refuse nowadays to connect amps of that category at all to that type of system.

Shitty amps which clip all the time on a standard line-level signal are only fit for the bin.

JC

 

[Nik]

You're right. But if you use some software with a proper dithered volume control such as PureMusic then this doesn't affect the resolution. In iTunes you should make sure that all the EQ and volume normalising functions are disabled.

To return to this ...

Dithering means that quantisation errors average out as innocuous random noise instead of less pleasant distortion. It can do nothing to retain resolution.

If you reduce the "volume" by 20 dB (a realistic value IMO), you are reducing the numerical values of all the samples by a factor of 10, IOW from a max of 65535 (2^16 - 1) to 6553.5. This is less than 2^13 (8192). If you reduce by a further 20 dB, the max value becomes 655.35 or less than 2^10 (1024). What you are sending to the DAC are therefore only 13 and 10 bit files (or more likely 12 and 9) for these 2 cases. Makes you think, when everyone's going on about 16 bit being no good, doesn't it?

Anyone (jcbrum?) like to check my figures?

 

[Nik]

I have been doing it that way for about five or six years now, .... with two provisoes ...

1. ... on AVI ADM9 systems (sorry chaps ), they have a built-in, digitally remote controlled, analogue, level control, as part of the dac -> pre-amp -> active crossover, with mute and input switching also available on the remote controller. So the computer can always be run 'flat out'.

2. ... on my other systems which use amps running flat out permanently connected to loudspeakers, I am careful to choose amps which don't clip unless overdriven. The maximum sound output is then governed by the loudspeaker ('cause that's what decides the current drawn) and providing you realise straightaway what is occurring, they should be able to stand full volume briefly.

The only time I damaged a loudspeaker was with a Cyrus amp, which when delivering full volume went straight into severe clipping, and more or less put DC across the speakers.

I just refuse nowadays to connect amps of that category at all to that type of system.

Shitty amps which clip all the time on a standard line-level signal are only fit for the bin.

JC

You say "full volume" but isn't this just a matter of allowing too much gain? This is a common trick to make an amp seem more "powerful" but doesn't mean that at all.

 

[Uzinusa]

To return to this ...

Dithering means that quantisation errors average out as innocuous random noise instead of less pleasant distortion. It can do nothing to retain resolution.

If you reduce the "volume" by 20 dB (a realistic value IMO), you are reducing the numerical values of all the samples by a factor of 10, IOW from a max of 65535 (2^16 - 1) to 6553.5. This is less than 2^13 (8192). If you reduce by a further 20 dB, the max value becomes 655.35 or less than 2^10 (1024). What you are sending to the DAC are therefore only 13 and 10 bit files (or more likely 12 and 9) for these 2 cases. Makes you think, when everyone's going on about 16 bit being no good, doesn't it?

Anyone (jcbrum?) like to check my figures?

Nik, i sort of see. Thanks for a more informed response FWIW I keep the volume on 0db PureMusic, un-attenauted on the DAC, and only use the pot on my Starfish to turn it up.

 

[Nik]

Nik, i sort of see. Thanks for a more informed response FWIW I keep the volume on 0db PureMusic, un-attenauted on the DAC, and only use the pot on my Starfish to turn it up.

Glad it makes sense.

It seems to me that the best answer is to do what you do, i.e., do the volume control (attenuation) as late as possible. This preserves the best S/N (analogue) and bit depth (digital).

 

[jcbrum]

You say "full volume" but isn't this just a matter of allowing too much gain? This is a common trick to make an amp seem more "powerful" but doesn't mean that at all.

It's just a badly designed amplifier that clips severely when set to it's maximum gain setting.

I suppose you could fit a 10dB attenuator to the input and derate it to say 7 watts, but what use is that, if you want 100dB spl from your speakers.

Just buy a proper 100w+ amp in the first place that doesn't clip on full signal.

Or better still a 250w one.

JC.

 

[jcbrum]

To Uzinusa, Amarra is of little use on a modern Mac running current versions of the OS X.

The OS already has a fully dithered 24 bit level control, and simply doesn't require Amarra for this function. It makes no difference at all in that respect. W7 is similar. The only useful function which Amarra performs is on-the-fly rate switching instead of auto re-sampling. I think it's pretty useless and possibly introduces extra distortions.

Dithering means that quantisation errors average out as innocuous random noise instead of less pleasant distortion.

It can do nothing to retain resolution.

I don't think that is valid Nik. It is the dithering which avoids the quantisation errors, and therefore preserves the resolution.

.. If you reduce the "volume" by 20 dB (a realistic value IMO), you are reducing the numerical values of all the samples by a factor of 10, IOW from a max of 65535 (2^16 - 1) to 6553.5. This is less than 2^13 (8192). If you reduce by a further 20 dB, the max value becomes 655.35 or less than 2^10 (1024). What you are sending to the DAC are therefore only 13 and 10 bit files (or more likely 12 and 9) for these 2 cases. Makes you think, when everyone's going on about 16 bit being no good, doesn't it?

Anyone (jcbrum?) like to check my figures?

Your mathematics are correct, but they only apply to systems running software which is about six years out of date, e.g. Windows XP and iTunes 6, and software of that era.

Current systems just don't operate with 16bit volume controls, they all use 24 or 32 bits which don't have the truncation problems, and also use dithering to avoid quantisation errors.

I repeat the advice given by Weiss, quoted earlier ........

. . . with today’s 24 bit converters – no question that a level control with a 24 bit wordlength easily rivals the best analog level controls. By the way, 24 bits means 16,777,216 quantization steps . . .

That's the trouble with the Benchmark advice sheet, - it's just so out of date to be useless.

JC.

 

[Nik]

I don't think that is valid Nik. It is the dithering which avoids the quantisation errors, and therefore preserves the resolution.

Quantisation errors are inevitable when digitising a signal and cannot be avoided. Is is the results (character) of these errors that dithering modifies. You cannot reduce the bit depth and preserve the resolution as if by magic.

Your mathematics are correct, but they only apply to systems running software which is about six years out of date, e.g. Windows XP and iTunes 6, and software of that era.

Current systems just don't operate with 16bit volume controls, they all use 24 or 32 bits which don't have the truncation problems, and also use dithering to avoid quantisation errors.

I repeat the advice given by Weiss, quoted earlier ........

Increasing the bit depth will reduce the quantisation errors when resizing the samples to reduce the volume, but at some point you convert back to 16 bit (unless, of course, you preserve the greater bit depth through to the DAC). You must still end up with 12-13 bit and 9-10 bit to represent the reductions by 20 and 40 dB, respectively.

That's the trouble with the Benchmark advice sheet, - it's just so out of date to be useless.

JC.

Keeping up to date with hardware and software in the computer world is a nightmare.

 

[prowla]

Whether or not the amp blows the speakers (either because of distortion or because it is just too loud for them) is only one of the risks of a computer app coming on at full volume when the computer is straight into a power amp.

The other is that it could give you a heck of a fright, wake somebody up if at the wrong time, or some other variation.

So, I would suggest a pre-amp is required.

 

[jcbrum]

Nik, it works as if it were 'magic'. Dithering restores the bit depth by interpolation, but if you use a 24 bit level control you are never short of bit depth or resolution anyway.

You don't end up with12 bits or 9 bits if you operate on 24 bits.

JC

 

[jcbrum]

I would suggest a pre-amp is required.

A pre-amp is just another amplifier stage contributing to overall gain.

I think what you are trying to say is that you want a volume control available to reduce the signal into the amplifier chain.

Whilst in legacy systems this has been placed in the early stages of the amplifier chain, all it does is reduce the signal to a low level by means of attenuation.

There is no reason why, in a digital system, signal attenuation cannot be accomplished in the digital domain. It amounts to the same thing.

I think that you simply feel uncomfortable with a digital attenuator, rather than an analogue one.

I think that is a misplaced assessment of the technology.

JC

 

[Purité Audio]

A properly dithered digital attenuator, Weiss for example is completely transparent, you may of course like what a preamp adds.
Keith.

 

[prowla]

A pre-amp is just another amplifier stage contributing to overall gain.

I think what you are trying to say is that you want a volume control available to reduce the signal into the amplifier chain.

Whilst in legacy systems this has been placed in the early stages of the amplifier chain, all it does is reduce the signal to a low level by means of attenuation.

There is no reason why, in a digital system, signal attenuation cannot be accomplished in the digital domain. It amounts to the same thing.

I think that you simply feel uncomfortable with a digital attenuator, rather than an analogue one.

I think that is a misplaced assessment of the technology.

JC

No, I was answering the OP...

The question was do you need a pre-amp between a computer and a power amp, or could you just plug the computer straight into the power amp.

And I responded that I would not do that, and gave a pretty obvious practical reason.

I didn't even mention the technology, whether the pre-amp should be analogue active, analogue passive, or digital.

 

[Nik]

Nik, it works as if it were 'magic'. Dithering restores the bit depth by interpolation, but if you use a 24 bit level control you are never short of bit depth or resolution anyway.

You don't end up with12 bits or 9 bits if you operate on 24 bits.

JC

1. I think you are mistaken about dither. This is not what it does, or ever can do. If this were true, why would we ever need even as many as 16 bits, let alone 24 or 32?

2. Explain to me what happens when you have a 16 bit signal, reduce the level by (say) 20 dB in the digital domain, and feed it to a DAC set for 16 bits.

 

[Nik]

A pre-amp is just another amplifier stage contributing to overall gain.

I think what you are trying to say is that you want a volume control available to reduce the signal into the amplifier chain.

Whilst in legacy systems this has been placed in the early stages of the amplifier chain, all it does is reduce the signal to a low level by means of attenuation.

There is no reason why, in a digital system, signal attenuation cannot be accomplished in the digital domain. It amounts to the same thing.

I think that you simply feel uncomfortable with a digital attenuator, rather than an analogue one.

I think that is a misplaced assessment of the technology.

JC

Please explain what happens to the effective word length (bit depth) when you reduce the level in the digital domain. Take -20 dB as a simple example.