I have to absolutely agree that you shouldn't trust specs, there are far to many games marketing departments can play to make things look better than they really are. However, I think its a bit strong saying they are useless. Specs are just a datapoint. One that you have to take with a huge grain of salt. At the very least you can count on one rule: very few companies products exceed their specs. That alone is useful information.
A part of the reason for my large post was to make the point that there are way to many things that manufacturers shy away from specing. The S/N ratio of most converters is going to be far better than most people can get out of their system. I can't even meet the noise floor of my old Layla converters. 60 cycle hum in my system is at about -86db and I worked very hard to get it that low. Most modern converters have low enough THD that noise from EMI and hum are going to dominate. What I believe you have to worry about are things manufacturers don't talk about.
Many low cost converter don't even pass simple tests. Sonusman gave one story, although there are those that dispute it, it is typical of the few stories I've read about a handful of "budget" converters. I've heard of other people recording a sine wave swept across the frequency spectrum and found unstable frequency response below 100 Hz (not on a Delta converter). How can this be if they all use the same A/D converters? The answer is simple: there's more in the box than just one component. AN A/D covnerter usualy needs to be fed by an amplifier. The design of a balanced input amplifier with 115db S/N ratio, -100db THD, with no phase shift and flat frequency response is far from trivial. It is full of tradeoffs, often involving cost vs performance.
One of the most popular components used in many of the 110-120db S/N converters is the CS5396/97 made by Crystal Semiconductor. It costs $27 in quantities of 20 or less. So if the AD converter itself is less than $27 a channel, where's the rest of the cost of those high end converters that run $300 and up per channel? Its in all the stuff around the converter: a jitter free clock references, highly stable PLLs, high quality analog components, and a good quiet power supply. There are literally thousands of places to cut corners or even outright screw up the design of those parts of the system. Don't ever think that just because company A uses one of the same components as company B, their products are of the same quality. If designing were that simple I'd be a millionare many times over.
Anyway, I think specs are interesting. Obviously, beause I posted a bunch. The responses here have been useful. I especially wanted to thank Sonusman, as I was hoping to get his view on Apogee vs Lucid (I don't know why I didn't just come out and ask).
A part of the reason for my large post was to make the point that there are way to many things that manufacturers shy away from specing. The S/N ratio of most converters is going to be far better than most people can get out of their system. I can't even meet the noise floor of my old Layla converters. 60 cycle hum in my system is at about -86db and I worked very hard to get it that low. Most modern converters have low enough THD that noise from EMI and hum are going to dominate. What I believe you have to worry about are things manufacturers don't talk about.
Many low cost converter don't even pass simple tests. Sonusman gave one story, although there are those that dispute it, it is typical of the few stories I've read about a handful of "budget" converters. I've heard of other people recording a sine wave swept across the frequency spectrum and found unstable frequency response below 100 Hz (not on a Delta converter). How can this be if they all use the same A/D converters? The answer is simple: there's more in the box than just one component. AN A/D covnerter usualy needs to be fed by an amplifier. The design of a balanced input amplifier with 115db S/N ratio, -100db THD, with no phase shift and flat frequency response is far from trivial. It is full of tradeoffs, often involving cost vs performance.
One of the most popular components used in many of the 110-120db S/N converters is the CS5396/97 made by Crystal Semiconductor. It costs $27 in quantities of 20 or less. So if the AD converter itself is less than $27 a channel, where's the rest of the cost of those high end converters that run $300 and up per channel? Its in all the stuff around the converter: a jitter free clock references, highly stable PLLs, high quality analog components, and a good quiet power supply. There are literally thousands of places to cut corners or even outright screw up the design of those parts of the system. Don't ever think that just because company A uses one of the same components as company B, their products are of the same quality. If designing were that simple I'd be a millionare many times over.
Anyway, I think specs are interesting. Obviously, beause I posted a bunch. The responses here have been useful. I especially wanted to thank Sonusman, as I was hoping to get his view on Apogee vs Lucid (I don't know why I didn't just come out and ask).