More on sound cards

[Terry Fox]

According to the Flex Radio software, the soundcards currently RECOMMENDED
for use with PowerSDR are:

M Audio Delta 44 (or 66).  4 (6) mono channels in and out, PCI bus with
external connector pod ($139 on the web).  24bit/96kHz.  This is made by M
Audio, who is owned by Avid.  Avid is THE premier film and video editing
company, we bought their broadcast systems by the millions of dollars.  I
would expect their product to be first-rate.  spec Dynamic range is 99db
(A/D) & 103db (D/A), freq response claimed 22Hz-40kHz (96kHz sampling).

PreSonus Fiberbox.  6 in, 10 out mono channels, firewire interface to PC
($299 on the web).  24bit/96kHz.  Claims to have 110db dynamic range, flat
response 10Hz-50kHz.  Little pricey, but appears to be a gret high-end
solution for laptops.

These cards are SUPPORTED by PowerSDR:

Sound Blaster Audigy (PCI), SB Audigy 2 (PCI), SB Audigy 2 ZS (PCI), SB
Exitgy (USB), SB MP3+ (USB), and Turtle Beach Santa Cruz (PCI).

All other cards may be useable, but not known by PowerSDR.  A card that is
not supported may have benefits (higher sampling rate, bit depthr) that
PowerSDR may not be able to use.

The Sound Blaster Audigy series is what I am currently using, with an Audigy
4 PCI (unsupported) and an Audigy 2 ZS Notebook PCMCIA (unsupported).  Both
function under PowerSDR, but with limitations versus their capabilities.

I am thinking of purchasing a Delta 44 and either a SB Extigy or similar USB
audio card.  My thinking is that we are moving away from internal cards (PCI
or other), and moving toward external interfaces.  That reduces the induced
noise at RF, IF, and audio frequencies.  The main reason I would purchase a
Delta 44 is that it seems to be the "standard" of the SDR industry right
now, and a good unit to compare against.  Firewire cards seem out of my
price range, so I am thinking of standardizing on USB at this point.

The Soundblaster Live 24 series is not supported, but others have had some
luck with them.  The email I sent earlier on the tests conducted on the SB
live 24 external (USB) showed that it had fiarly good response for its price
($45 on the web).

I have three Turtle Beach sound cards (Fiji, Santa Cruz, Montego).  While
they are consistantly some of the better cards for their money, they are NOT
100% compatible with Sound Blasters.  I have had several occassions where
audio software would break when I tried to use them.  Therefore, I do not
recommend we standardize on them.

The SB Extigy and MP3+ are both supported cards that use USB interfacing.  I
think they even use 96kHz sampling.  I would like to try one of these.

There are many other USB sound cards out there.  We should be able to try a
bunch of them, and report he results to the SDR community.  Just Creative
alone makes the SB 24 Live, Extigy, MP3+, Audigy 2 NX - all are USB and very
similar.

I still believe that higher sampling and bit depth are important, but only
if accurate.  16 bits is not enough, as a 16-bit A/D often has an error of
the lsb or greater.  If a 24-bit A/D has only 20 accurate bits, then you are
still guaranteed that the 16msb are accurate.  If you buy a 16bit card, you
are guaranteed only 15 bits of accurate data, and probably less.  Again, I
don't agree that 16 bits is the upper limit of achievable accuracy these
days.  I will bet that 18-20 bits is achievable at low cost, across the
audio spectrum at least.  While SDR software may only use 16 bits in the
final analysis, making sure all 16 of those bits are reliable is the name of
the game.

Yes, "A" weighting is an issue, as is any other frequency roll-off (or
roll-up?).  But, if it is implemented consistently and the bits are there,
it can be adjusted for.  If not, you just lost more accuracy.  Another
reason to have 24 bit sampling vs 16 bit.  With only 16 bits you are
immediately throwing away useful bits.  If 20 of 24 bits are accurate, and
the SDR software only uses 16 bits, you are still ahead of the game.

Having a higher sampling rate leads directly to being able to use more of
the RF band being received.  With only 48kHz sampling (and a xtal-controlled
SoftRock), you can only "see" and receive plus and minus 24kHz around the
xtal frequency (48kHz total).  With 96kHz sampling, it doubles your
receivable bandwidth.  Moving to 192kHz sampling, the receivable bandwidth
quadruples.  An important benefit to rockbound systems.  However, the sound
card must be capable of ACCURATELY conveying the amplitude (bits of
resolution) and the frequency, otherwise there will be bad spots within the
receiveable bandwidth.  THIS IS THE REASON WE NEED TO DO THESE TESTS.
24-bit and 96kHz sampling are only good if they are accurate.

We should come up with different card "suggestions" for different price
ranges.  If someone wants to spend up to $50, here are the cards and their
specs.  Between $50 and $100, here is another group.  Between $100 and $150,
a third group.  And above $150 as a fourth group, as an example of how to
break down cost vs performance.

While I don't beleive the reviewers or the card manucafturer specs (at least
for what we want to see), I do believe we can find fairly low-cost cards
that do a much better job than the older 16-bit, 48kHz sound cards.

I suggest we make a list of cards, divide up that list, and purchase and
test these various options.  If a "standard lab" can be created, I wouldn't
mind purchasing a couple cards myself, and shipping them up to the lab for
testing.  Frank: could you, George, and possibly Steve Dove put together a
lab?  I will be glad to help as I can.  If we can document what the lab
testing consistes of, and I can duplicate it here, I can help with testing
as well.

More later.
Terry