Sound card?

Why are they useless? I've done distortion measurements with laptop sound cards (whatever was built in). Later when the plot from the agency PFC tests came back it all jibed quite well.

They're all over the place. There are spikes in the TDH plots that are unbelievably narrow.

No it really doesn't. Each one has different characteristics.

SOunds more like a Software issue in the decoder.

I've done DSP (FFT/DFT), with various sound cards, good and crappy one's.

Do you have a plot? Sound more like code errors than a sound card issue. Errors would also have to show up in the raw time domain data and I've never seen it happen. Sometimes, for example at one client, the engineer said that it must be my laptop, that it can't possible come from anywhere in the system. Well, turned out there was this lil' switcher way down in their system that suffered a brief hacking cough episode every time it crossed the DCM/CCM line ...

Take a real good look at the time domain data under various input signals (pulses, test tones etc.). Maybe you had bad luck and really ended up with all the sound card lemons while Jamie and I lucked out and received the good stuff ;-)

hree

. =A0

o with

est one.

ome

I'm probably preaching to the choir here, but you can use "sox" to view the data numerically. That takes any plotting software out of the picture.

I did some spice simulations using PWL source generating the spice input via a C program digesting the sox data. This let me analyze some filtering schemes based on real life data.

Why would different cards act differently? It is something strange though because I see no distortion.

They shouldn't. Are you sure there is nothing else running on the PC that could whack the data flow once in a while? Whenever I attempt noise hunts using the sound card I make sure nothing else is running, WLAN is turned off, etc.

one.

The card with the severe distortion doesn't *look* like there is any distortion. At tens of percent (maybe 50% in the extremes) I should see something ugly. Nope. The Creative card has output modulation that is simply rotten. I can easily see that on an external scope hooked to the same points. It also has something going on on the input side too because it's visible on the stereo traces and not in "phase" with the input or each other.

If the card OSR pointed out looks better I'll push for a better sound card. The budget is a bit tight right now though.

one.

I am afraid something else is going on in that PC. I can't do another scope plot from one of mine right now because that part of the lab is being renovated right now. Restoring concrete this afternoon, oh what fun. But the sound card output signals (any sound card) never looked rotten.

I got it working, for now, though as I said earlier the utility of the THD measurements is highly suspect. I recieved the sound card Steve (osr) recommended so I'll give that a try tomorrow. Since the purpose of this exercise is to come up with a method of trading off the hardware hybrid performance against its THD (the software end of the hybrid function doesn't like distortion in the echo) I really need both. ...by Friday. :-(

What did you do? Archie Bunker gave his old TV a good whack to make it work but I assume you didn't do that ;-)

Ouch. I don't know this company, just the first one that popped up, maybe call them or try their THD routines:

Not quite. I used the Turtle Beach card with isolation transformers. The distortion numbers are still suspect but the amplitude numbers look good.

Sure. I'll give 'em a try. Thanks!

Consumer sound cards suck for a lot of reasons. Many years ago I tried three or four different ones for digitizing LP's (!), they all had some noise or low-level 'tone' issues that I could hear even through the LP noise. I finally spent the money on an Audiophile 2496 card (about $150 back when it was newish), and it made a world of difference. I did 16/44 recordings with it and couldn't tell the recording from the source.

Especially the Creative/Soundblaster/SBLive!/Audigy cards suck. I forget what models do what, but a lot/most of them sample-rate everything to 48k, do whatever processing the soudcard does, then resample it to 44k.

Even if they say "24 bit 96kHz" - they even lost a class-action suit on that claim (it wasn't just a bad 24bit converter, it was actually giving only 16 bits), which got buyers discount/credit coupons for MORE of their crappy gear.

If you're doing anything approaching "serious" work where you want more than 8 bits worth of real audio data, get a "pro-sumer" card like the M-Audio (makes the above-mentioned Audiophile 2496, newer 192, other stuff such as I'm-sure-good USB interfaces) or Echo Audio or others you'd find in a large musical instrument store such as Guitar Center, NOT in a computer store. If you want response to DC, below

The Turtle Beach SRM works fine for that.

Several have different sampling rates (it's pretty easy to do).

Creative is well known to "stretch the truth".

Sounds like audiophoolery to me.

Can't get it to work. Can't read the driver disk and haven't found anything on the web that is recognized.

You guys must really have had a streak of bad luck with sound cards. I never bought a fancy card, always used what was in there. Applications are rather demanding since my job includes hunting down very faint noise sources. The kind where they fly you in no matter what the travel costs are.

Noise issues could always be traced back to other sources, such as the PC's power supply. Many laptops are especially bad in that domain. However, in all cases that could be fixed with isolation transformers or differential amplifiers. The topper was a case where a $$$$$ analyzer could not see it but my Twinhead Durabook laptop showed a smacking spike at 7Hz from a mechanical noise source nobody had thought could possibly interfere. Some jaws dropped. Then disbelief. "Ok, maybe it's the PC doing that." ... "Well, can you slow down that spinning shaft over there?" ... spike moved to 6Hz ... "Oh!"

I bought one of these for yuks. Works fine on the PC, still working out bugs under linux. Mine used a C-Media 106-L.

I haven't flogged it with any instruments, but there was no hiss when I connected ear-buds to the front out. That in itself is a good sign. It has a metal case. Kind of cute. The PCB slides out. Four screws and you have it open.

I didn't even include the Dell laptop at work. We use the sound from that all the time but it's junk (both). The PS noise from that really surprised me. As you say, an isolation transformer cleans it up but I need a second transformer to eliminate all the hum (there is already one on our 2-wire output).

I'm surprised you could see anything down at 6Hz with the PC.