Digi 001 discontinued?!

[Dale-c]

By the way, those of you who wish to compare a 001 with other goodies (converters, pres, etc) with an out of the box 002/002R are hardly being fair. Obviously anyone who is happy with the system they have, 'stock' or upgarded should stick with what they have. Lastly, let me address the money issue. The 001 wasn't exactly 'cheap' when it was introduced. In fact the 002R is fairly close to the same price when introduced.

<font size="2" face="Verdana, Arial">My thoughts too.
If you add all of that to the 002 then there would be no advantage.
As for the price, I think the 002R is the same price as the 001 was originally at $1200.
I still would like to see the 001 go and a lower priced Firewire replacment take it's place.
We will see.

[muspro]

Sorry Duardo but you are completly missing the point of "high resolution" sampling. It increases the "resolution" meaning it can approximate the original analog signal much more accuratly.

It is simple as grid paper. If your gird paper was made of 1" square blocks and my paper had 1/32" square blocks and we were both asked to draw a 1" circle using the grid lines.....guess what.........you would have a SQUARE! On the other hand I would have a very close approximation of a circle that would be 32X more accurate than yours.

I'm not sure what you are thinking here.

Also to clarify the Nyquist Theory: "According to the Nyquist theorem, to capture a sound wave accurately, you need to sample at at least two times the frequency of the signal you're trying to capture"

It isn't to "capture accurately", but to capture it all! You have to have a minimum of 2 samples to capture a wave form! This yields the absolut worst representation of that wave form. At 48k your 6k info has 8 samples representing the anolog wave. (4up and 4down)

This is not a very accurate picture (like your square circle) and why high sample rates have a much nicer and smooth sounding top end.

Hope this helps clear up some issues.

[clorox]

Originally posted by LSRdigital:
I just wanted to point something out, as I have seen the same inaccurate comments in other posts on the DUC.

We all know that the general range of human hearing is from 20 Hz to 20 KHz. However, I am seeing a lot of people get this confused with the SAMPLE RATE of digital recording equipment, which is also measured in KHz. All Hz means is cycles per second. These two measurements have absolutely nothing to do with each other.

<font size="2" face="Verdana, Arial">Thank god you posted this before I had to. I was getting depressed.

[img]images/icons/grin.gif[/img]

[clorox]

I'm with everyone here that's saying the higher latency is unacceptable.

I use outboard converters (dbx 386 mic pre, triple C compressor), so the new converters don't mean anything to me.

But having your drummer's hits always 256 samples behind the click, and to be able to hear the echo of himself in his headphones. . . . that sucks.

[img]images/icons/frown.gif[/img]

[Burton]

Monitor through your mixing board = 0 latency.

[Duardo]

Duardo,
so if you owned an 001 and had $1200.00 to spend on an 8 channel A/D D/A converter (for home drum tracking), which would you choose:
1. the RME ADI-8? (to go with my 6 pre's plus the two 001 pre's) 8 total
OR
2. the 002 rack? (heard the conversion is that close to the RME?) plus 4 useable pre's for a change?

<font size="2" face="Verdana, Arial">I haven't been able to A/B them, but I'd probably go for the RME...if I already had the 001, I figure that would give me the most flexibility for future expansion.

There isn't a defacto standard noise spec for the the 001 to meet. There IS a defacto standard noise spec. for Firewire I/O's to meet before getting that logo on there. Does this mean the 001 is noisier than an airport? Hardly. But a Firwire certified I/O should simply provide a cleaner signal than the 001 PCI I/O.

<font size="2" face="Verdana, Arial">The Firewire spec has nothing to do with signal quality. That all rests on your converters. The Firewire cable just trasmits data.

By that logic, then, the higher-end HD systems should not be as quiet as the 002, since they use PCI cards? That's not the case.

I think LSRdigital did a fine job of explaining sample rates. A 96k sample rate offers a (twice the resolution as stated) better sample of the analog wave form than 48k. A better sampled waveform offers better clarity, better accuracy of nuances of the sound that was recorded.

<font size="2" face="Verdana, Arial">That is only true for audio above about 20 kHz. It takes twice the samples, yes, but that only manifests itself in frequencies above the Nyquist frequency. For audio we can hear, a "better" sampled wafeform does not offer any more clarity or accuracy. This is not an opinion of mine, this is the way it is. Go out and find any scientific book on digital sampling theory and you'll see that that's the case.

Maybe you can or can't hear the differance, but there IS a differance.

<font size="2" face="Verdana, Arial">There can be a difference, depending on the design and implementation of the filters and other components in the system...but there's nothing inherently better as far as 96kHz is concerned that we can hear.

As for the price, I think the 002R is the same price as the 001 was originally at $1200.

<font size="2" face="Verdana, Arial">The Digi001 has always retailed for $995...it sold for pretty close to that when it first came out, and the price dropped over time. The 002R retails for $1295 and sells for about $1200...who knows, its price may drop over time as well.

Sorry Duardo but you are completly missing the point of "high resolution" sampling. It increases the "resolution" meaning it can approximate the original analog signal much more accuratly.

<font size="2" face="Verdana, Arial">No, the whole point is that it captures more frequencies. It can also allow for gentler filter designs and that sort of things, but with today's oversampling converters that shouldn't be an issue.

It is simple as grid paper. If your gird paper was made of 1" square blocks and my paper had 1/32" square blocks and we were both asked to draw a 1" circle using the grid lines.....guess what.........you would have a SQUARE! On the other hand I would have a very close approximation of a circle that would be 32X more accurate than yours.
I'm not sure what you are thinking here.

<font size="2" face="Verdana, Arial">It's not even close to that simple. The main flaw with your example is converters don't draw along the grid lines. Here's how it works:

Take your grid with 1" square blocks. If each block represents a half of a second, then the resolution of your system is two blocks per second. (We can't do a circle since audio isn't represented that way...we'll have to "unfold" it into a sine wave).

Now, let's put one point at zero on the X- and Y- axes. We'll talk in coordinates from now on...so that point is at (0,0). Then put one at (1,1). Then one at (2,0), then (3,-1), then (4,0). If you just connect those dots any way you want, you could have a triangle wave, or a sine wave, or just about anything. But that's not the way it works.

With those points in that grid, the only way to connect them without "breaking" the rules would be to connect them as a sine wave. If you connected them point to point, that would represent audio that's higher than the Nyquist frequency in terms of content. If you had your sine wave go up to "2" on the Y-axis between 0 and 1 on the X axis it wouldn't be able to make it back down in time...it would have to "turn around" at a rate that's higher than the Nyquist frequency.

If you were to sample that same sine wave at a higher rate...if you took an extra x point between each existing one...the sine wave would look exactly the same. If you took ten between each one, it would still look the same. In fact, if you had your first sample at (0,1), and every sample after that was at "0" on the X-axis after that initial sample you'd have a sine wave at one block per second, which would be the highest-frequency wave you could recreate that is, not coincidentally, half of your sampling rate (which was two blocks per second).

Also to clarify the Nyquist Theory: "According to the Nyquist theorem, to capture a sound wave accurately, you need to sample at at least two times the frequency of the signal you're trying to capture"
It isn't to "capture accurately", but to capture it all! You have to have a minimum of 2 samples to capture a wave form! This yields the absolut worst representation of that wave form. At 48k your 6k info has 8 samples representing the anolog wave. (4up and 4down)
This is not a very accurate picture (like your square circle) and why high sample rates have a much nicer and smooth sounding top end.
Hope this helps clear up some issues.

<font size="2" face="Verdana, Arial">That's not right. According to the Nyquist theorem, you either capture it or you don't. There's no "absolute worst" representation of the waveform...either you can represent it or you can't. If you follow the rules of the theorem, there's one and only one way to connect the dots, and having those extra dots in there doesn't give you anything except for the ability to capture higher frequencies. No "squares" allowed because they represent higher-frequency information.

Go back to our paper example. If you did have one more point in between each one on the X-axis...in other words, if we could have a point at .5...then you could capture signals at a higher frequency...up to two blocks per second, since now your sampling rate is four blocks per second.

Hopefully this helps clear up some issues.

-Duardo

[where02190]

Yeehawwwww....we ain't had a good ole fashioned p*ssin' contest on here in a hoot long time!!!!!!!!!!

Praise the Lord and pass the ammunition!!!!!!!

FWIW< IMHO, if you're buying, the 002 rack looks pretty sweet for the price, but also IMHO it's unfortunate it only runs on PT6......which...IMHO is still being beta tested......

Hope this is helpful.

[muspro]

Duardo,

Sorry man, we are on two different planets. If your example were true, we could sample bass frequencies (60 Hz) at 120 samples/sec, mids (500 Hz) at 1000 samples per sec. etc. etc. This sure would save a TON of disk space and processing time. We should invent a frequency divided sample rate system. All frequencies just get 2 samples per wave form! That would sound horrible!

There are 2 major things that I can think of that higher sample rates effect:

1. Higher frequency. We both agree on this (nyq. theory).

2. Higher resolution. More slices of audio represent the analog wave more accurately.

Hold on........I just re-read your post. Here is the flaw in your statments:

"If you were to sample that same sine wave at a higher rate...if you took an extra x point between each existing one...the sine wave would look exactly the same. If you took ten between each one, it would still look the same."

Your talking about taking a 48k digital wave form and up-sampling to 96k. You are just making a copy of a digital signal. This would of course yeild the same digital signal. This is not what goes on during a typical recording process.

When recording an analog signal (human voice through a mic), the 96k recording will have twice the resolution. A much more accurate representation of the analog signal.

One other note, analog signals are NOT square, digital signals are. The more resolution we have the less square are digital signals "appear".

Imagine a 20k sine wave. It goes up. It goes down. It's a very smooth and rounded wave form. In a 44.1 recording there are essentially TWO SAMPLES to record this wave. That digital sample would be represented as a straight up signal and instantly straight down. NOT AT ALL what really happend. Now take into account the timing of the sample VS. the original analog signal and the digital signal may not even come close the the actual voltage (volume/bit level) as the original signal.

Example: Draw three vertical lines. The two areas between the lines represent a 40k sampling space (for simplicity I am using 40k instead of 44.1). Now, in those 2 spaces draw a 20k sine wave. It has to go up, round off at the top, go down through 0, continue down, round off at the bottom and then go back up to 0. That would represent one full 20k sine wave. That was four smooth lines that has to be represented by 2 digital square blocks. The voltage or level of those blocks would be an average of the content in that sample.

Again, there is no way to represent that analog signal accuratly. However higher resolution would.

Are we on the same page now?

I will see if i can find a resource on the web that shows diagrams of analog wave forms trying to "fit in" to digital blocks and it's shortcomings.

[Duardo]

Sorry man, we are on two different planets. If your example were true, we could sample bass frequencies (60 Hz) at 120 samples/sec, mids (500 Hz) at 1000 samples per sec. etc. etc. This sure would save a TON of disk space and processing time. We should invent a frequency divided sample rate system. All frequencies just get 2 samples per wave form! That would sound horrible!

<font size="2" face="Verdana, Arial">Actually, that's right...you could sample 60 Hz frequencies at 120 Hz and so on. But how often do you need to do something like that? 44.1kHz was chosen because it is able to sample all the frequencies we can hear. Everything from 20 Hz up to 20 kHz. But sure, if you were recording a signal that went no higher than 10 kHz, there'd be no need to sample at a rate higher than 20 kHz.

2. Higher resolution. More slices of audio represent the analog wave more accurately.

<font size="2" face="Verdana, Arial">I don't disagree with you on this entirely...yes, more "slices" represent the analog wave more accurately...but we don't hear that accuracy. The do not represent what we can hear more accurately when we go above 44.1kHz.

Hold on........I just re-read your post. Here is the flaw in your statments:
"If you were to sample that same sine wave at a higher rate...if you took an extra x point between each existing one...the sine wave would look exactly the same. If you took ten between each one, it would still look the same."
Your talking about taking a 48k digital wave form and up-sampling to 96k.

<font size="2" face="Verdana, Arial">No, I'm not. I'm talking about sampling that same analog sine wave at different sampling rates. As long as the sine wave is below the Nyquist frequency, it doesn't matter how many samples you take of it, it will be reproduced exactly. Perfectly.

When recording an analog signal (human voice through a mic), the 96k recording will have twice the resolution. A much more accurate representation of the analog signal.

<font size="2" face="Verdana, Arial">Only if it's picking up information above 20 kHz. If it is, then yes, it's more accurate, but it doesn't matter...you won't hear a difference.

More is better to a point...but once you get past the point where you can hear the difference, why bother?

One other note, analog signals are NOT square, digital signals are. The more resolution we have the less square are digital signals "appear".

<font size="2" face="Verdana, Arial">You are exactly right. Except it doesn't matter how digital signals "appear". We don't hear digital signals, we hear analog signals. Sure, what you're storing may look like a bunch of sqare waves. But that's not what you hear. What you hear is an analog signal that is reconstructed from the digital samples that are taken. Antialiasing filters get rid of everything above the Nyquist frequency (as well as their audible artifacts) on the way in, and reconstruction filters get rid of everything above it on the way out. So while what you record may be stored as a square wave, what you hear will be a sine wave that sounds (and looks on an oscilloscope) just like what went in.

Imagine a 20k sine wave. It goes up. It goes down. It's a very smooth and rounded wave form. In a 44.1 recording there are essentially TWO SAMPLES to record this wave. That digital sample would be represented as a straight up signal and instantly straight down.

<font size="2" face="Verdana, Arial">After the signal has been converted back to analog, it's the same sine wave it was when it went in.

Example: Draw three vertical lines. The two areas between the lines represent a 40k sampling space (for simplicity I am using 40k instead of 44.1). Now, in those 2 spaces draw a 20k sine wave. It has to go up, round off at the top, go down through 0, continue down, round off at the bottom and then go back up to 0. That would represent one full 20k sine wave. That was four smooth lines that has to be represented by 2 digital square blocks. The voltage or level of those blocks would be an average of the content in that sample.
Again, there is no way to represent that analog signal accuratly.

<font size="2" face="Verdana, Arial">I'm sorry, but you obviously don't understand how all of this really works. I'm not trying to get into a "pissing match" here...there are facts we're talking about, and what you've just stated is wrong. In your example, what's stored are three samples, right? Not blocks, not waves, just three samples. When a D/A converter is fed those three samples it connects them the only way it can...in this case, with the sine wave that was captured in the first place. It can't output square waves because a square wave at 20 kHz by definition contains high-frequency content that's much higher than 20 kHz. That will have all been rounded off by the filters in the D/A converters and what remains is the only thing that can possibly remain at that point...a pure sine wave that is exactly the same as the one that went in.

Are we on the same page now?

<font size="2" face="Verdana, Arial">Now I ask the same question of you.

-Duardo

[Lowfreq]

Duardo - I never said anything about HD/TDM systems PCI cards. They are out of the price point we are discussing here and therfore do not apply. HD/TDM cards are sonically superior & have a low level noise floor that nothing in the lower price line (Mbox, 001, 002) systems can touch.