Best way to get digital signal INTO computer

[Abstraction]

I guess this is the appropriate forum. There is much discussion here about getting a signal out of a computer to a DAC. I need to go the other way.

 

I have an NAD M2, which I am very pleased with. One of its tricks is that it takes audio from a phono line stage, digitalizes it and reprocesses it through its DAC. It also has optical and SDIF digital outputs.

 

I have been trying out Pure Vinyl the last few days, using a cheap toslink cable straight into a Mac G5, and because I happened to have it, a cheap M-Audio Fast Track Pro, with SDIF in and USB out to the G5. I have been surprised how both of these completely minimal alternatives sounded.

 

I would like to hear from our experts here. What are the issues? What are better and best ways of making this connection? Would there be the same advantage to using an expensive card (Lynx perhaps) or what?

 

[Abstraction]

I realize that, and for the present I am only toying with doing it. I have about 2000 LPs--about eight hours a day for 6 months, if I calculate in my head correctly. And, of course, I may want to go for an interface that will record hi-rez. The M2 will play hi-rez inputs but not output audio inputs in hi-rez, it seems.

 

I would also be interested in hearing about the experience of others with Pure Vinyl or in other ways.

 

[Audio_ELF]

Your question is very open ended... Everything from £20 Behringer interface to £4,500 Weiss ADC2 will do the job...

 

You may need to define it a bit more...

 

Eloise

 

[Abstraction]

or at least not primarily so. For starters are the technical problems--except for direction-- the same when inputting to computer as outputting? What is jitter as input?

 

 

 

[Audio_ELF]

Well if you're using digital from M2 to Computer you're going to have issues of jitter to contend with.

 

You may get superior results using a ADC interface to computer rather than trying to take the digital out of the M2.

 

Elosie

 

[Miska]

Well if you're using digital from M2 to Computer you're going to have issues of jitter to contend with.

 

Let's remember that jitter manifests itself at AD or DA conversion time. When doing ADC over S/PDIF the ADC is the one who drives the clock and receiver (computer) is slaved to this. When doing DAC over S/PDIF the source (computer) is the one drives the clock, this is much more difficult situation. So if the ADC clocking in M2 is adequate and low jitter, any interface that can correctly read the bits from the incoming S/PDIF flow is good. Since the data is just stored, there are no timing issues on the computer side.

 

To this direction, there are no issues with computer interface, unlike opposite direction. Since the clocking master/slave situation is opposite. IOW, S/PDIF for ADC for recording purposes is just fine. The situation would be different if feeding DAC straight from ADC over S/PDIF, because then the indefinite buffer of harddisk wouldn't be in the middle and DA conversion would be clocked by the clock recovered from ADC.

 

 

[Audio_ELF]

I bow to your superior knowledge Miska; but a question...

 

What happens where the ADC is working with one clock and the interface a different clock: isn't there an issue if the two clocks are not identical (I.e. In extreme case the ADC has 44101Hz clock and the interface a 44099 clock)?

 

Or am I talking complete rubbish?

 

Eloise

 

[Miska]

What happens where the ADC is working with one clock and the interface a different clock: isn't there an issue if the two clocks are not identical (I.e. In extreme case the ADC has 44101Hz clock and the interface a 44099 clock)?

 

S/PDIF input always runs at clock recovered from the incoming datastream. Interface just recovers the clock using PLL and doesn't have clock of it's own. Same thing as with DACs, unless you have ASRC.

 

So on S/PDIF sender has the active master clock and receiver has slaved PLL clock recovery. No matter which side has what kind of role in the picture. (S/PDIF transmitter chip has data and clock INPUTS, S/PDIF receiver chip has data and clock OUTPUTS)

 

So think the process as complete reverse of playback process, with the difference that the jitter-sensitive part is still at the same side, between analog and digital worlds. In digital world, you can alter time; it's flexible variable abstract thing, while in analog world you practically cannot change time flow (OK, I don't want to go to hardcore light-speed physics)...

 

 

[Abstraction]

 

Miska wrote:

 

"So think the process as complete reverse of playback process, with the difference that the jitter-sensitive part is still at the same side, between analog and digital worlds."

 

So this is the reason that when I input analog to the M-Audio Fast Track directly, it sounded awful, but when I used it as a link between the SPDIF out on the M2 and the computer USB input, it sounded quite good. Right?

 

Always nice to have a little theory to back up what your ears are telling you.

 

However, given this new camera that doesn't have to be focused until after the picture is taken (today's NY Times), perhaps it is time to start thinking light-speed physics in audio.

 

Don

 

 

 

[Miska]

However, given this new camera that doesn't have to be focused until after the picture is taken (today's NY Times), perhaps it is time to start thinking light-speed physics in audio.

 

Well, that would be still there on analog side, in physical world.

 

Digital signal processing tricks are a nice way to help that analog physical world, be it in camera or audio equipment.

 

If we return to definition of jitter... Sampling at frequency fs (44.1 kHz) each sample is assumed to be dT apart (22.676 µs). Jitter j is non-constant variation of this resulting real timing being dT+j when compared to time of analog signal representation which is the time of physical analog world (relative, not absolute). When the samples are transferred/copied in digital domain, they are not put into relation with the time in terms of signal, internal timing can be still varied by modifying the data, for example with ASRC (which can be thought as digital time-stretch-device that can compress and expand time in terms of the signal)

 

Having clocks of two different frequencies vs data is called clock drift, not jitter.

 

 

Now, going back to "light-speed physics", in electrical and optical signals if you place your audio equipment in a craft moving fast enough, the time in that craft moves slower than time outside it, and would cause pitch shift in the music when listened outside.

 

In acoustics, probably everybody is familiar with Doppler-effect, when source of sound is moving relative to listener, taking into account speed of sound in air, the sound is "compressed" when it's approaches and "stretched" when it's going away.

 

These two could also cause jitter, if the speed would be varying.

 

(ok, we could also play with thoughts like compressed air and such)

 

 

This is the reason you can transfer your digital music files over internet, S/PDIF, Firewire or USB, fast or slow in relation to playback time (song length), without affecting the signal...

 

 

[Miska]

I'll write a little bit more about the topic...

 

So this is the reason that when I input analog to the M-Audio Fast Track directly, it sounded awful, but when I used it as a link between the SPDIF out on the M2 and the computer USB input, it sounded quite good. Right?

 

Probably bigger difference comes from quality of the analog stages before ADC chip and from the ADC chip itself. But part of it could be coming from the clock quality too, yes.

 

Rewording previous explanations again... In S/PDIF connected ADC it has the master clock and that runs the AD conversion and data transfer - so jitter depends solely on quality of this clock - how stable it is.

 

In S/PDIF connected DAC it has the slave clock ("best effort guess"-clock) and that runs the DA conversion. Except if it has ASRC that compares the local master clock and guessed slave clock and modifies the data so that input and output data flow speeds match the relative difference of clock frequencies.