🚀 go-pugleaf

"Paul D." <media@clara.net> wrote in message
news:st0ogtg0h6niro4j7qqn8su7o36gtdpm87@4ax.com...
> "Arny Krueger" <arnyk@pop3free.com> wrote:

> SpectraLab showed a S/N ratio of over 97 dB with a 1Khz tone (not
> 96.3).  You can see from the screengrab that it was taken in real
> time, set to 16 bit with a 48Khz sample rate, used an appropriately
> high resolution FFT window and Blackman analysis.
>
> You explain the error.

SpectraLab is very good software, but it is not a certified, traceable,
laboratory grade measurement system. And ANY measurement system has a degree
of uncertainty. You just need to establish what it is. I can assure you both
it is more than 1 dB.

> There's more!  SpectraLab indicates a S/N ratio of 100 dB with a 20Hz
> tone.  See here :
>
> http://home.clara.net/media/sblive/20HZ-DO.gif
>
> You explain the error.

The FFT window is too large for the signal, so more power is deemed signal,
and less as noise. (yes I know that is it's maximum) SpectraLab also mesures
peak noise not RMS, not sure if any weighting is used.
I would not be unhappy with those results though :-)

> >> For the record I think your tests are a damn sight better than
> >nothing,
>
> I agree with this.. my basic problem with the site is that Kruger
> doesn't keep it up to date and therefore these sorts of issues arise.
> The same thing happened with my Echo Gina.  It *still* says "reference
> card" on the test page which is incorrect, and the results are poorer
> than I obtain with the card.  The same thing with my AWE64 Gold.  And
> this is supposed to be a reference site?

In Arny's defence, he isn't being paid to do this, so I think the
disclaimers apply. If you don't like it ...
I think he admits the card may be capable of better performance, so can we
all call a truce?

Trevor.