View Single Post
Old 11-09-2010, 12:09 PM   #52
yep
Human being with feelings
 
Join Date: Aug 2006
Posts: 2,012
Default

To expand on the above, and to try and keep this thread from tuning into an autistic spec-examination...

All those numbers and measures in Fabian's spec above sound very technical and precise, but they should be regarded as essentially meaningless. Or at least, no more meaningful than a guitar amp that goes up to eleven. Here's why:

Let's say I'm a gear manufacturer making audio interfaces, and I want to publish a spec for max input level. Here are some of the ways I might go about coming up with that spec, short of outright making it up (and there's really nothing to prevent me from doing so, it's not like these are government-regulated measurements of anything):

- Look at the component list and design drawings and calculate the nominal tolerances.

- Take 10 sample units and measure the performance of each, and then set the spec level at either the average, or best-performing one (more to the point, what is the measure of "max"? Where it blows up? Where THD exceeds some arbitrary threshold? Where I think it starts to sound bad? On top of that, what am I even "testing"? A 1kHz sine wave? White noise? Swept sine waves? A Beatles CD?)

- Test the original hand-built design prototype that I made on a breadboard in my design lab (hmm... that couldn't possibly perform differently than real-world production units stamped out in a factory in Thailand made with the cheapest available components, could it?)

- Simply publish the specs of the existing hardware whose design I am attempting to copy (phew! that was easy! I was afraid I might have to learn how to use an oscilloscope!)

- Publish the design spec that the unit was SUPPOSED to meet, according to the product design criteria (IOW, marketing/management says we want a device that meets X spec. I design a device and say it should meet that spec, so that's the spec we publish. Saves a lot of trouble that way).

- Guess.

Maybe your device manufacturer wouldn't possibly do those things. Okay... so, where do their specs come from? Actual tests of your individual production unit? If so, what did they test? What does "max" mean?

My point is, don't go rushing for your manual to see if your numbers are bigger than Fabian's numbers, because those numbers are meaningless unless they are measured results of your specific unit in your studio with your real-world program material. It's like comparing how high the numbers on guitar amplifier volume knobs go.
yep is offline   Reply With Quote