Digi 001 discontinued?!

[Lowfreq]

qoute-
How did I take it out of context? I was under the impression that you were saying that the 002 sounds better because it uses Firewire instead of PCI. Is that not right?

Your impression (and previus statement) was that Firewire has nothing to do with sound quality, they just transmit data. Any noise in the chain affects the sound quality. Firewire has a standardized noise quality to meet to be certified a IEEE1394 device. The 001 PCI does not need to meet a recognized noise spec. What you had taken out of context was stating that if Firewire was 'better' than 001 PCI card, then with the thinking above, then it had to be better than the HD PCI systems.

[SKI]

Isn't bit rate the resolution? In other words the number of times that signal is referenced in the digital domain and sample rate how often it looks at the signal?
Since both analog or digital circuits can be represented mathematically it is easy to plot the effects and results with a minimal amount of points of reference but it requires a better design in the converter, higher sample rates allow for improvements in converter design to be more effective since it is looking at the signal more often almost like error correction since it is seeing more actual change in the signal and predicting less.
What the higher sample rate really offers is the ability to design better converters.
Improved number crunching in the software for the 96k and better converters would change the sound.
This brings up another thought about the 001, which is what part of the circuit is in the breakout and what is on the PCI card.
Is it split analog (breakout) and Digital (PCI)?
If this is the case Digi could easily upgrade the 001 with a new PCI card just like MOTU did when the went from the 324 to the 424 card.
That would be an interesting upgrade path.

[Carl Z]

Any noise in the chain affects the sound quality.

<font size="2" face="Verdana, Arial">I don't get it. It's a digital signal from the converter box on out - isn't it? How do you induce noise into a digital signal?

[Kickin.da.speaker]

Duardo is 100% right.

However, to go back into the real world, sampling at 96KHz DOES make a difference in frequencies under 20KHz, but only due to the filtering process of the A/D and D/A conversions. The higher the sampling rate, the lower the slope of the filtering, hence the less distortion around the cutoff frequency.

So if you sample at 40KHz, you need a brickwall filter at 20KHz. Doesn't exist.

If you sample at 80KHz, you need a brickwall filter at 40KHz, or a steep filter at 35KHz (and distortion around 35KHz is ok since we don't hear it).

And so on: the higher the sampling frequency, the farther you push the distortion introduced by the converters in higher frequencies that we don't hear. That's the benefit of high sampling rates. The "better resolution" is not a benefit of the sampling rate, since we do NOT hear it.

If you want a better resolution that you will hear, you'll need to increase the bit rate.

[Kickin.da.speaker]

1. 2 samples of a waveform is all that is needed to accurately capture it digitally and then have it recreated "exactly" by a good DA.
Only true of a sinusoidal waveform. Actually it's 2 samples for every cycle of the highest frequency component of your signal. If you have a 20 Hz pure tone, then 40Hz sampling is all you need (in theory. In practice, you'll want a little more to minimize the distortion introduced by the converter's filters). If, however, you have a 20Hz waveform that includes harmonics that are 500Hz, then you'll need 1KHz sampling rate, and so on.

2. Higher sample rates don't technically increase the resolution of any audio information less than 1/2 of the acutal sample rate. It is just a waste of samples in that area.
100% correct. It's redundant information.

3. Higher sample rates only increase the resolution of higher frequencies. (related to above).
Right.

4. Higher sample rate converters CAN and/or DO sound better becuase:
a - They can capture harmonic content above 20k
Wrong: we can't hear it anyway.

b - The low pass filters are around 96k and not as
harsh in the more audible areas around 20k.
Right for a 192KHz converter. The cutoff would actually be a little lower than 96KHz.

c - They can be a newer higher quility converter than
current 44.1/48k converters.
That's usually right but that's another topic.

d - They may have a beter clock making it seem that
they higher SR is the reason for the beter sound.
That's the same thing as {c}

6. I won't need 96/192k until those converters a way better than any available 44.1/48k converters.

They already are!

[fenris]

Wow!!! This thread is...humbling! [img]images/icons/blush.gif[/img] It is a shame that even in university (I studied in music ten years ago) we are taught a simplification that seems to be a bit misleading.

[Duardo]

Your impression (and previus statement) was that Firewire has nothing to do with sound quality, they just transmit data. Any noise in the chain affects the sound quality. Firewire has a standardized noise quality to meet to be certified a IEEE1394 device. The 001 PCI does not need to meet a recognized noise spec.

<font size="2" face="Verdana, Arial">Firewire doesn't have anything to do with sound quality. It's true that any noise in the chain affects the sound quality, but Firewire doesn't add any noise, and in the case of the 001 neither does the PCI interface. In both the 001 and the 002, the converters are in the boxes, and once the sound has been converted to digital it's transmitted digitally to the system via the PCI cable and the Firewire cable, respectively. Neither affects the sound or adds noise at all.

There are PCI systems where that's not the case, like the M-Audio and Echo interfaces, where the box with the jacks is just a box with jacks, and the converters are actually on the PCI card. In that case, sure, you could have noise introduced over the PCI cable, since it's transmitting an analog signal to the PCI card. But that's not the way it works with the 001.

What you had taken out of context was stating that if Firewire was 'better' than 001 PCI card, then with the thinking above, then it had to be better than the HD PCI systems.

<font size="2" face="Verdana, Arial">I'm sorry about that. I realize now that you weren't saying that Firewire isn't inherently better than PCI...but it still seems like you're saying that the improved quality of the 002 has something to do with the fact that it connects to the computer via Firewire, which isn't the case...it's all in the box.

Isn't bit rate the resolution? In other words the number of times that signal is referenced in the digital domain and sample rate how often it looks at the signal?

<font size="2" face="Verdana, Arial">Both bit depth and sampling rate can affect "resolution"...the term gets thrown about oftentimes without being specific as to what type of resolution you're talking about.

Since both analog or digital circuits can be represented mathematically it is easy to plot the effects and results with a minimal amount of points of reference but it requires a better design in the converter, higher sample rates allow for improvements in converter design to be more effective since it is looking at the signal more often almost like error correction since it is seeing more actual change in the signal and predicting less.

<font size="2" face="Verdana, Arial">Nope, that's still not quite right. The Nyquist theorem proves...PROVES...that you don't need more than double the highest frequency you're trying to capture. Any problems with converters have more to do with other components of the design of the converter...filters, etc...than the sampling rate itself.

Is it split analog (breakout) and Digital (PCI)?
If this is the case Digi could easily upgrade the 001 with a new PCI card just like MOTU did when the went from the 324 to the 424 card.
That would be an interesting upgrade path.

<font size="2" face="Verdana, Arial">The conversion takes place in the breakout box...so to upgrade the converters in the 001 you'd need to replace the breakout box, not the card (just like the 1296 was able to do 96 kHz with the older PCI 324 card).

So if you sample at 40KHz, you need a brickwall filter at 20KHz. Doesn't exist.

<font size="2" face="Verdana, Arial">No, it doesn't, at least not in the analog domain. But if we use a gentle filter and oversample, then we can use an extremely steep digital filter without the phase problems we'd have trying to do it in the analog domain.

1. 2 samples of a waveform is all that is needed to accurately capture it digitally and then have it recreated "exactly" by a good DA.
Only true of a sinusoidal waveform. Actually it's 2 samples for every cycle of the highest frequency component of your signal.

<font size="2" face="Verdana, Arial">Right...and the highest frequency component we can hear is (rougly) a 20 kHz sinusoidal waveform. If it's anything other than sinusoidal, then by definition it includes information that's higher than 20 kHz, and that information will have already been filtered out by the anti-aliasing filters.

b - The low pass filters are around 96k and not as
harsh in the more audible areas around 20k.
Right for a 192KHz converter. The cutoff would actually be a little lower than 96KHz.

<font size="2" face="Verdana, Arial">Check into a good high-end converter. The filters aren't harsh at all, even at 44.1 kHz.

d - They may have a beter clock making it seem that
they higher SR is the reason for the beter sound.

<font size="2" face="Verdana, Arial">You can have just as good of a clock at 44.1 kHz as you can at 192 kHz.

6. I won't need 96/192k until those converters a way better than any available 44.1/48k converters.
They already are!

<font size="2" face="Verdana, Arial">True, the converters already are better...but many of them are just as good at 44.1 kHz as they are at 192 kHz...

-Duardo

[rwhitney]

Just a quick correction to my previous, earlier post: I was referring to the tc electronic 2290 digital delay that employs a 1MHz sampling rate. (not a model 1176, as erroneously stated, which is an analog compressor by a different manufacturer).

[B-Grade]

Man, just when I thought these types of arguments were over!

One thing I can say about some conjecture of Converts sounding the same at any sample rate. We use a Radar system daily. Some consider it to be one of the better sounding digital recorders out there (I'm agree). I can say this without a doubt-

The Radar sounds way better at 96K. And 192? Fahgetabouddit. Same clock. Same converters.

Recording drums and acoustic guitar is night and day better at the higher rates. Night and day. And this baby sound great at 48. This is all I should ever need to say 192K is better, my ears.

Now for the more analytical side. AD/DA converters all have errors. They all have some form of smoothing filters to reassemble the data. This all adds up to guesswork. Simply put, take a clearer picture in the first place, spend less time cleaning it up later using math for guessing what it should have been.

[Duardo]

Man, just when I thought these types of arguments were over!

<font size="2" face="Verdana, Arial">They'll never end...

The Radar sounds way better at 96K. And 192? Fahgetabouddit. Same clock. Same converters.

<font size="2" face="Verdana, Arial">Can't argue with that.

AD/DA converters all have errors. They all have some form of smoothing filters to reassemble the data. This all adds up to guesswork. Simply put, take a clearer picture in the first place, spend less time cleaning it up later using math for guessing what it should have been.

<font size="2" face="Verdana, Arial">"Guessing" isn't the best word. We know exactly what we had after the analog signal was filtered...so when the D/A converter puts it back together, it's not guessing. The problems are the filters themselves...they're getting better all the time, but they're still not perfect. Get the filters "perfect" and the recreation of the information we can hear will be "perfect" as well at 44.1 khz. No guesswork involved.

I don't think it'll ever be perfect, but I think it's entirely plausible that we may reach the point where the differences are inaudible. I think some converters are there already...but I'm not sure all converters ever will be. I'm not sure they all want to be.

-Duardo