Mo Facta
Farts of Nature
I wrote a reply to necropost from 2005 this morning that got lost because of my stupid mouse. Don't ask. It's highly annoying and prone to make me lose it if I have to explain. So now that I've got my cup of warm Rooibos tea at hand, I'm feeling better and thought I would post a new thread on what I wanted to say.
I want to disseminate some of my thoughts on converters, sample rate and their relation to audio quality.
I see a lot of threads all over the internet discussing the merits and demerits of recording at higher sample rates, quality of prosumer vs high end converters, etc and a lot of this talk is based primarily in the digital domain, totally negating the FACT that there are numerous analogue components in the design of every single piece of digital gear.
To clarify, I see a lot of digital jargon being spewed everywhere about digital theory, Dan Lavy papers being thrown left and right, and a lot of focus being put on the digitized signal itself, as if it's the only thing impacting audio quality. Well, my argument is that the digital aspect of "digital audio" is indeed a small part of what constitutes a quality signal.
Why?
Well, like I said, analogue components are crucial to the design of any converter. Before any signal hits the converter chip it has to go through at least some sort of analogue section involving some sort of analogue amplifier such as an opamp. To me, THIS is where the "sound" of any given converter lies. The rest is 1s and 0s. Of course, there are many other factors that are paramount in converter design, such as a good power supply, a well thought out PCB, a stable and low-jitter clock, etc, among other things, but when you're talking about the aesthetics of a captured signal, most of the mojo comes from the analogue domain, IMO.
Now, I'm not a tech guru, but I am able to grasp a lot of this and coming from an analogue background, it's easy to get a gist of the facts.
The first thing to know is that there are only a few digital audio converter chip manufacturers, the main ones being Asahi Kasei (AKM), Cirrus Logic, and TI (who now owns Burr Brown). These chips are used in 90% of all converters out there from the cheapest to the most expensive. For instance, the SSL Alpha-Link, a high end converter system costing $2000+ uses the Asahi Kasei AK4620B converter chip, which is the exact same chip used in the M-Audio Profire 2626, an interface costing several times less. Additionally, the analogue sections (AD and DA) are based around the JRC 4565 opamp, which is also found all over prosumer designs. They have mediocre specs and have an abysmal slew rate. Why? Because they are cheap. It actually blows my mind that a company with a pedigree like SSL would use such crap and so insult the people that spend all that money for the SSL name. All you have to do is go onto Digikey.com and find higher quality replacements to see why these companies skimp on these components for mass production. The Burr Brown OPA2134 for example, a high quality replacement for the 4565, is about 12 times the price per unit.
This is what makes sense to me.
Lower quality components generally saturate and distort quicker than higher quality components. Some converters - and Massive and Waltz will back me up on this - have been designed to tolerate higher levels or even clipping better than others. The point here is that you get less distortion at higher levels, which is why modern low headroom prosumer devices using these components have actually done us a disservice. If your mixer or preamp or converter clips at +18dBu but it starts producing distortion at +12dBu, it kind of negates having that extra 6dB of headroom, doesn't it?
Distorting low quality components is the road to harshness and has nothing to do with the digital section of the converter. That is why "digital harshness" to me is a misnomer. Often even saturating the front end components produces distortion many dB below clip point. That, IMO, combined with 24-bit audio, is from whence the philosophy of conservative levels was borne. I think people found that if they didn't saturate the analogue section of their converters, they got cleaner, punchier recordings because the harmonic distortion accrued from saturating cheap components wasn't present. Of course, this harmonic distortion due to circuit saturation (from slammed levels) would often muddy up the lower midrange and produce a harsh, "woolly" sound, which is not very musically pleasing. Hence we have standards that have been around for decades like the 0VU/-18dBfs/+4dBu standard. They help us keep our audio clean and distortion free for high fidelity recording.
That's pretty much all I have to say for now.
Cheers
I want to disseminate some of my thoughts on converters, sample rate and their relation to audio quality.
I see a lot of threads all over the internet discussing the merits and demerits of recording at higher sample rates, quality of prosumer vs high end converters, etc and a lot of this talk is based primarily in the digital domain, totally negating the FACT that there are numerous analogue components in the design of every single piece of digital gear.
To clarify, I see a lot of digital jargon being spewed everywhere about digital theory, Dan Lavy papers being thrown left and right, and a lot of focus being put on the digitized signal itself, as if it's the only thing impacting audio quality. Well, my argument is that the digital aspect of "digital audio" is indeed a small part of what constitutes a quality signal.
Why?
Well, like I said, analogue components are crucial to the design of any converter. Before any signal hits the converter chip it has to go through at least some sort of analogue section involving some sort of analogue amplifier such as an opamp. To me, THIS is where the "sound" of any given converter lies. The rest is 1s and 0s. Of course, there are many other factors that are paramount in converter design, such as a good power supply, a well thought out PCB, a stable and low-jitter clock, etc, among other things, but when you're talking about the aesthetics of a captured signal, most of the mojo comes from the analogue domain, IMO.
Now, I'm not a tech guru, but I am able to grasp a lot of this and coming from an analogue background, it's easy to get a gist of the facts.
The first thing to know is that there are only a few digital audio converter chip manufacturers, the main ones being Asahi Kasei (AKM), Cirrus Logic, and TI (who now owns Burr Brown). These chips are used in 90% of all converters out there from the cheapest to the most expensive. For instance, the SSL Alpha-Link, a high end converter system costing $2000+ uses the Asahi Kasei AK4620B converter chip, which is the exact same chip used in the M-Audio Profire 2626, an interface costing several times less. Additionally, the analogue sections (AD and DA) are based around the JRC 4565 opamp, which is also found all over prosumer designs. They have mediocre specs and have an abysmal slew rate. Why? Because they are cheap. It actually blows my mind that a company with a pedigree like SSL would use such crap and so insult the people that spend all that money for the SSL name. All you have to do is go onto Digikey.com and find higher quality replacements to see why these companies skimp on these components for mass production. The Burr Brown OPA2134 for example, a high quality replacement for the 4565, is about 12 times the price per unit.
This is what makes sense to me.
Lower quality components generally saturate and distort quicker than higher quality components. Some converters - and Massive and Waltz will back me up on this - have been designed to tolerate higher levels or even clipping better than others. The point here is that you get less distortion at higher levels, which is why modern low headroom prosumer devices using these components have actually done us a disservice. If your mixer or preamp or converter clips at +18dBu but it starts producing distortion at +12dBu, it kind of negates having that extra 6dB of headroom, doesn't it?
Distorting low quality components is the road to harshness and has nothing to do with the digital section of the converter. That is why "digital harshness" to me is a misnomer. Often even saturating the front end components produces distortion many dB below clip point. That, IMO, combined with 24-bit audio, is from whence the philosophy of conservative levels was borne. I think people found that if they didn't saturate the analogue section of their converters, they got cleaner, punchier recordings because the harmonic distortion accrued from saturating cheap components wasn't present. Of course, this harmonic distortion due to circuit saturation (from slammed levels) would often muddy up the lower midrange and produce a harsh, "woolly" sound, which is not very musically pleasing. Hence we have standards that have been around for decades like the 0VU/-18dBfs/+4dBu standard. They help us keep our audio clean and distortion free for high fidelity recording.
That's pretty much all I have to say for now.
Cheers