What makes a high end audio amp "better" then a low end unit?

[Fungus]

Oh, come on. The reconstruction filter has been built into DACs for decades.

(that's why they've started calling them "codecs").

[paulca]

15 years ago, he has a 192Khz capable sound card, back then, that's not consumer grade.  That's a studio grade sound card.  (You can see he is able to select higher sample rate settings....)

You have to remember that in audio production you don't "sample" to just record.

If you wanted to sample, say, a piano, you could take a multi-part sample, one sample per octave (or half or quarter octave depending on your equipment/sampler and it's memory).  But a lot of basic samples are single samples which when used in music might be played across multiple octaves. 

So it is common practice to over sample at much higher rates to retain quality when played on lower notes.

A 44.xKhz sample of a piano at middle C might start to sound rubbish two octaves down when you replay it at 11Khz.

[David Hess]

The 44.1/48 kHz 16 bit samples are resampled at a multiple of the sample rate so the requirements of the analog antialiasing filter are considerably relaxed.  Digital integration makes this more economical than an analog solution.

[paulca]

The 44.1/48 kHz 16 bit samples are resampled at a multiple of the sample rate so the requirements of the analog antialiasing filter are considerably relaxed.  Digital integration makes this more economical than an analog solution.

I believe that is one way of doing it.  The other is to use high sample rates.  I confess it's been a decade or two since I've used a professional hardware sampler though.

[coppice]

The 44.1/48 kHz 16 bit samples are resampled at a multiple of the sample rate so the requirements of the analog antialiasing filter are considerably relaxed.  Digital integration makes this more economical than an analog solution.

Sample rate conversion like that is normally done by using a polyphase filter that bandlimits to the new Shannon rate and produces output samples at the required rate all in one go.

[Buriedcode]

It seems most DACs for audio, and - at least as far as I have seen in consumer grade audio gear - ADCs are Sigma Delta, which greatly lessens the burden on the antialiasing filter.  It seems many have forgotten this when talking about 44.1kHz as a sampling frequency.

https://en.wikipedia.org/wiki/Delta-sigma_modulation
http://www.ni.com/white-paper/11342/en/

You can buy perfectly adequate ADCs, DACs /Codecs for <$2 that require only a clock source and a simple RC filter.  Often the less-than-stellar results one gets from PC and laptop soundcards is nothing to do with the codec itself, but rather the power supply used (for example, they rarely have a decent mic bias, and headphone amps tend to be right next to noisy parts).  This is why external cards - even ones with *worse* specced codecs often perform much better, they don't have to cope with all the EMI generated inside a computer.

Also, the fact a converter can do 192kHz sample rate, does not mean it is 'studio grade' or 'consumer grade' or say anything about its ENOB.

[David Hess]

I was just keeping things simple.  Oversampling is implemented in the ADC/DAC which considerably relaxes the requirements of the analog antialias filtering.  Delta-sigma converters also do this to an extreme to relax the requirements of the modulator and implement noise shaping to improve dynamic range.  But the result is that 44.1/48 KHz and 16 bits are sufficient for transparent audio because of this processing being done by the ADC/DAC.

There are all kinds of other limitations including noisy power, ground loops, crummy analog signal paths including the power amplifier, and poor clocking.  The later is especially a problem with interfaces which do not provide the sampling clock like USB although I wonder if any common interface on a PC does.  Like power amplifiers, clocking is a solved problem but there are lots of examples out there which are broken.

[Fungus]

The weird thing about all this theory is that most of it was figured out in the 19th century. Long before "electronics".

[coppice]

Oh, come on. The reconstruction filter has been built into DACs for decades.

(that's why they've started calling them "codecs").

The term codec is a contraction of code and decode. It was a term that started with telephony ADC + DAC + A-law/u-law conversion chips, I believe. By any measure a DAC only or an ADC only device is not a codec, whatever bits of its part of the signal chain it includes or misses out.

[Zero999]

All this talk of sampling is interesting and very relevant if one is using a sound card to record something, but for CD playback, all of the sampling and anti-aliasing will have already done before the CD was recorded. The DAC in the playback device doesn't have to worry about filtering or anti-aliasing. All it has to do is convert the 1s and 0s on the disc to the correct analogue value at the right speed.

[coppice]

All this talk of sampling is interesting and very relevant if one is using a sound card to record something, but for CD playback, all of the sampling and anti-aliasing will have already done before the CD was recorded. The DAC in the playback device doesn't have to worry about filtering or anti-aliasing. All it has to do is convert the 1s and 0s on the disc to the correct analogue value at the right speed.

Rubbish. What comes out of the DAC needs band limiting to recover the original signal. That filter might be integrated with the DAC on a single chip, but its still a vital filter beyond the actual DAC.

[Zero999]

All this talk of sampling is interesting and very relevant if one is using a sound card to record something, but for CD playback, all of the sampling and anti-aliasing will have already done before the CD was recorded. The DAC in the playback device doesn't have to worry about filtering or anti-aliasing. All it has to do is convert the 1s and 0s on the disc to the correct analogue value at the right speed.

Rubbish. What comes out of the DAC needs band limiting to recover the original signal. That filter might be integrated with the DAC on a single chip, but its still a vital filter beyond the actual DAC.

I thought that the only stuff which needed filtering, on the analogue side of the DAC was above the sample rate of 44.1kHz and therefore ultrasonic in this case, so the filter isn't critical, or have I missed something?

[Dubbie]

Yes you have hero999. Read a primer on DACs and you will see that filtering before digitising is essential as is filtering after converting back to analog. With both of these taken care of you can get practically a perfect reconstruction of the original waveform within the bandwidth constraints you are working in.

If you want to learn more, I can recommend this book: http://www.dspguide.com/
I learnt (and still am learning) a lot from this book.

You can read the whole book for free on the site. Chapter 3 will fill in the blanks for you.

[coppice]

All this talk of sampling is interesting and very relevant if one is using a sound card to record something, but for CD playback, all of the sampling and anti-aliasing will have already done before the CD was recorded. The DAC in the playback device doesn't have to worry about filtering or anti-aliasing. All it has to do is convert the 1s and 0s on the disc to the correct analogue value at the right speed.

Rubbish. What comes out of the DAC needs band limiting to recover the original signal. That filter might be integrated with the DAC on a single chip, but its still a vital filter beyond the actual DAC.

I thought that the only stuff which needed filtering, on the analogue side of the DAC was above the sample rate of 44.1kHz and therefore ultrasonic in this case, so the filter isn't critical, or have I missed something?

You've missed that if all that ultrasonic stuff reaches a typical audio amplifier the results won't be pretty. Also, if anything downstream re-digitizes the signal (perhaps as part of a digital power amp) its going to digitize something pretty funky if you haven't band limited the signal.

[Zero999]

All this talk of sampling is interesting and very relevant if one is using a sound card to record something, but for CD playback, all of the sampling and anti-aliasing will have already done before the CD was recorded. The DAC in the playback device doesn't have to worry about filtering or anti-aliasing. All it has to do is convert the 1s and 0s on the disc to the correct analogue value at the right speed.

Rubbish. What comes out of the DAC needs band limiting to recover the original signal. That filter might be integrated with the DAC on a single chip, but its still a vital filter beyond the actual DAC.

I thought that the only stuff which needed filtering, on the analogue side of the DAC was above the sample rate of 44.1kHz and therefore ultrasonic in this case, so the filter isn't critical, or have I missed something?

You've missed that if all that ultrasonic stuff reaches a typical audio amplifier the results won't be pretty. Also, if anything downstream re-digitizes the signal (perhaps as part of a digital power amp) its going to digitize something pretty funky if you haven't band limited the signal.

I see your point, but I think you've overlooked the fact that the output filter, on a DAC, is far less critical, than the input filter, on an ADC.

Any decent audio amplifier should already have a low pass filter, guard against demodulation of high frequency signals and protect a piezo tweeter, connected to the output. Anything which digitises the signal again (which isn't best design practise anyway: ideally a digital amplifier will convert the digital input from the DAC directly to analogue PWM to drive the speaker) should already have a good low pass filter.

[coppice]

I see your point, but I think you've overlooked the fact that the output filter, on a DAC, is far less critical, than the input filter, on an ADC.

In practice its a little less critical, but it still needs to be taken seriously.

Any decent audio amplifier should already have a low pass filter, guard against demodulation of high frequency signals and protect a piezo tweeter, connected to the output. Anything which digitises the signal again (which isn't best design practise anyway: ideally a digital amplifier will convert the digital input from the DAC directly to analogue PWM to drive the speaker) should already have a good low pass filter.

Show me an audio amp with a brick wall filter that will clean off everything beyond the Shannon rate of an incoming 44.1ksps signal. As for re-digitizing a signal, what happens when you re-digitize an unfiltered 44.1ksps signal at 96ksps or 192ksps? Those digitizers won't filter away the mess below about 48kHz or 96kHz.

I expect the reason you don't see this filtering as important is you haven't experienced what happens when it isn't there. Most audio DACs have had the filtering integrated since the mid 80s. in early CD players they used to up sample by 4x or 8x the 44.1ksps rate during the ZOH pre-compensation, just to make the analogue post filter simpler. Now most DACs are sigma-delta, the filtering requirements are must easier to meet, but still important.

[Fungus]

Any decent audio amplifier should already have a low pass filter, guard against demodulation of high frequency signals and protect a piezo tweeter, connected to the output. Anything which digitises the signal again (which isn't best design practise anyway: ideally a digital amplifier will convert the digital input from the DAC directly to analogue PWM to drive the speaker) should already have a good low pass filter.

Show me an audio amp with a brick wall filter that will clean off everything beyond the Shannon rate of an incoming 44.1ksps signal.

I don't think an audio amp has any business doing anything at all to a signal other than making it louder.

Any filtering should be in the source. The source knows exactly what sort of filtering is needed (if any), the amplifier doesn't.

[alexanderbrevig]

I don't think an audio amp has any business doing anything at all to a signal other than making it louder.

Any filtering should be in the source. [...]

https://en.wikipedia.org/wiki/Class-D_amplifier

You would be very disappointed with a Class-D that has no filtering.

[Zero999]

Any decent audio amplifier should already have a low pass filter, guard against demodulation of high frequency signals and protect a piezo tweeter, connected to the output. Anything which digitises the signal again (which isn't best design practise anyway: ideally a digital amplifier will convert the digital input from the DAC directly to analogue PWM to drive the speaker) should already have a good low pass filter.

Show me an audio amp with a brick wall filter that will clean off everything beyond the Shannon rate of an incoming 44.1ksps signal.

I don't think an audio amp has any business doing anything at all to a signal other than making it louder.

Any filtering should be in the source. The source knows exactly what sort of filtering is needed (if any), the amplifier doesn't.

The audio amplifier should filter out any ultrasonic frequencies on the input, which the designer knows can cause unintended operation. It is important for EMC purposes. If poor/inadequate filtering is used, the speakers will buzz loudly when: placed near a switched mode power supply or a nearby mobile phone receives a call, or it might pick up and demodulate the signal from a nearby AM transmitter. To stop this, a filter with a cut-off above 20kHz should be added to the inputs, along with ferrite beads. This is more important with amplifiers, connected to long leads, carrying small signals, such as microphones.

I don't think an audio amp has any business doing anything at all to a signal other than making it louder.

Any filtering should be in the source. [...]

https://en.wikipedia.org/wiki/Class-D_amplifier

You would be very disappointed with a Class-D that has no filtering.

What filtering? On the input or output?

Filtering is often omitted from the output, especially when the speaker is close by. The inductance of the voice coil, acts as a low pass filter preventing large currents from flowing at the PWM frequency.

There should be some filtering in the input, for EMC reasons, mentioned above.

[Fungus]

https://en.wikipedia.org/wiki/Class-D_amplifier

You would be very disappointed with a Class-D that has no filtering.

A class D filters its output, obviously.

[Zero999]

I see your point, but I think you've overlooked the fact that the output filter, on a DAC, is far less critical, than the input filter, on an ADC.

In practice its a little less critical, but it still needs to be taken seriously.

Any decent audio amplifier should already have a low pass filter, guard against demodulation of high frequency signals and protect a piezo tweeter, connected to the output. Anything which digitises the signal again (which isn't best design practise anyway: ideally a digital amplifier will convert the digital input from the DAC directly to analogue PWM to drive the speaker) should already have a good low pass filter.

Show me an audio amp with a brick wall filter that will clean off everything beyond the Shannon rate of an incoming 44.1ksps signal.

There are no audio amplifiers with a brick wall filter at that frequency, because it's not necessary. If the amplifier is properly designed, then a small amount of 44.1kHz and harmonics present in the input, will not cause any problems. Generally most amplifier designs don't misbehave, until the frequency is well above that anyway.

As for re-digitizing a signal, what happens when you re-digitize an unfiltered 44.1ksps signal at 96ksps or 192ksps? Those digitizers won't filter away the mess below about 48kHz or 96kHz.

I expect the reason you don't see this filtering as important is you haven't experienced what happens when it isn't there. Most audio DACs have had the filtering integrated since the mid 80s. in early CD players they used to up sample by 4x or 8x the 44.1ksps rate during the ZOH pre-compensation, just to make the analogue post filter simpler. Now most DACs are sigma-delta, the filtering requirements are must easier to meet, but still important.

I see your point about re-digitising the signal and the problems it may cause.

[oldway]

All this has nothing to do with the subject of this topic ....

After 14 pages of audiofoolery, I'm still waiting for someone to tell me what a "high end" amplifier is, to know where we can find the list of these "high end" amplifiers, what sets them apart from low end amplifiers, (NB: apart from perlinpin powder that can not be measured or confirmed in an objective and impartial double-blind test)

If these tests can not be done because the hering memory is limited to 10 seconds, how can we say that one amplifier sounds better than another ....?

How is it possible that there is a hearing difference between 2 amplifiers, while the measurements made with high precision instruments are equivalent?

Unlike a microphone, or a loudspeaker where there are complex phenomena like resonances, an amplifier is an electronic component whose characteristics are measurable with high precision ... where would the magic of the different sound come from? perlinpinpin powder?

[Fungus]

After 14 pages of audiofoolery, I'm still waiting for someone to tell me what a "high end" amplifier is

A mythical beast.

How is it possible that there is a hearing difference between 2 amplifiers, while the measurements made with high precision instruments are equivalent?

It isn't.

(not unless people are measuring the wrong things)

[trys]

Cotton buds ('Q Tips' in the US) make a big difference.

[Buriedcode]

Show me an audio amp with a brick wall filter that will clean off everything beyond the Shannon rate of an incoming 44.1ksps signal. As for re-digitizing a signal, what happens when you re-digitize an unfiltered 44.1ksps signal at 96ksps or 192ksps? Those digitizers won't filter away the mess below about 48kHz or 96kHz.

I expect the reason you don't see this filtering as important is you haven't experienced what happens when it isn't there. Most audio DACs have had the filtering integrated since the mid 80s. in early CD players they used to up sample by 4x or 8x the 44.1ksps rate during the ZOH pre-compensation, just to make the analogue post filter simpler. Now most DACs are sigma-delta, the filtering requirements are must easier to meet, but still important.

As you stated, most, if not all modern audio DACs are sigma delta, - their output is updated at 64/128x the sampling rate, pushing the sample rate waaay up the spectrum, meaning the a simple output filter, combined with the frequency response of an amp greatly attenuates it.  I won't go into interpolation/digital filters etc.. as ultimately it just means that at the output, the sample frequency is now in the MHz range, and is already attenuated.  Unless one is plugging the output of their DAC into an RF amp, I don't see why output filtering is particularly important. Yes, it is needed, but I can't see it being something that requires a great deal of design?

The conventional (say, R2R ladder DACs) that really do require strong attenuating filters are, these days, the ones designed and built by 'audiophiles' because they believe that sigma delta DACs sound 'sterile' and because they've seen 1-bit SD modulation waveform and thought it doesn't 'look' like nice clean audio.  The irony being that in an effort to bash modern audio converters, they've resorted to older technology that performs measurably worse, but 'looks nicer' on a scope (and also costs a lot more because of very tight component tolerances, which justifies the high cost for those silly enough to pay for them).

Some sources of that misunderstand the process about it:
http://www.mother-of-tone.com/conversion.htm
http://www.msbtechnology.com/faq/why-ladder-dacs/
https://hifiduino.wordpress.com/2014/10/12/r2r-for-the-rest-of-us/

Note the last one isn't snake oil or anything, I'm sure its a very well designed R2R DAC, but seems like a LOT of effort and money for something that performs very similarly to a $2 IC.