[Lowfer] Sound card measurements (long)

[Lyle Koehler]

I will cross-post this to the LowFER reflector and soft_radio discussion
group because there has been interest in the subject on both forums.

I tested three different sound cards by feeding the same 10 kHz signal to
both left and right channels of the sound card's Line input, and recording a
short .wav file in CoolEdit at a 48 kHz sampling rate. I then used the
zoom-in feature of CoolEdit to look at a few cycles of the signal and
determine the delay between zero crossings on the left and right channels.
Of the cards I tested, ironically the only one that has input delay problems
is the one I thought was my "good" card; a Creative SoundBlaster Live card
that is integrated on the motherboard of a Pentium III 900 MHz computer.
This is a motherboard that was also built (or at least marketed) by
Creative. The SoundBlaster Live has a delay of approximately 14 microseconds
between the left and right channel inputs, which corresponds to a phase
shift of about 50 degrees at 10 kHz. Other cards I tested were a Crystal PnP
on an IBM 366 MHz machine, and a Crystal WDM on a 266 MHz Dell notebook. Not
all notebooks have stereo line inputs, but this one does. I couldn't see any
delay between left and right channels on the two Crystal sound cards.

To test the output delay, I again used CoolEdit to generate 10 kHz tones
with zero phase shift in both left and right channels, and looked at the
output on a dual-channel scope. The SoundBlaster Live card may have had a
very slight delay between output channels; on the order of a microsecond.
Neither of the Crystal cards showed any delay.

Unfortunately the machines with the usable sound cards are too slow for
software defined radio. So I "rescued" an old Yamaha OPL-3 ISA card from a
retired computer and plugged it into the Pentium 900 MHz machine. Supposedly
you can run two sound cards under Windows (I'm using Windows ME), but I
didn't get that far. During the boot sequence, messages came up showing that
the computer had recognized the OPL-3 card. Then the computer said "updating
ECSD" (I think I remember the letters correctly) and froze. For a while,
even after pulling out the ISA sound card, the computer wouldn't do anything
at all. I don't know what was wrong, but eventually things started working
again. I plugged the ISA card back in and went into the BIOS and disabled
the on-board sound card, after which the boot sequence proceeded normally.
Windows detected the card and had no problem installing the drivers. It also
re-installed a bunch of things like the floppy drive controller that should
not have needed re-installing, and came up with some other weird messages
which I ignored. The input test on the OPL-3 showed no delay between
channels, but the output has a delay of about 22 microseconds, or around 80
degrees at 10 kHz. I generated another test file with 1 kHz tones and
verified that it's a constant time delay.

The conclusion is that some sound cards will work with software defined
radio, and some won't. And I don't know if price is any guide to
performance.

This month's QST has a brief writeup on the FlexRadio software defined
transceiver as a "New Products" item on page 81. The FlexRadio web site at
http://www.flex-radio.com/ lists several sound cards that are supported. I
assume that these sound cards don't have the delay problem, although I would
like to know for sure before actually buying one. Some of the "cards" in the
FlexRadio list are external devices, connected via the USB port, which I
hope would eliminate the conflicts on my computer. By the way, two of them
are currently listed as rebate specials at Circuit City
(www.circuitcity.com). The Creative MP3+ is $40 after rebate, and the
Creative Extigy is $90 after rebate.

Lyle, K0LR