$50 Sound Cards Impress Versus Integrated Audio

crookedvulture writes "Most PCs have audio integrated right on the motherboard. There's much to be gained from upgrading to a discrete sound card, though. This look at a couple of sub-$50 sound cards from Asus explores what can be found at the budget end of the spectrum. In blind listening tests, both cards produced better sound than an integrated solution. They also offered superior signal quality, but neither had an impact on gaming performance. The days of hardware-accelerated game audio seem to be behind us, with developers handling positional audio processing in software."

Comments

Distance from the power supply

[itamblyn]

Most of the improvement is likely due to increased distance between the amplification circuits and the noisy AC/DC power supply.

Re:Distance from the power supply

[ChrisMP1]

A decent ATX power supply really isn't bad. Distance from the noisy digital circuitry on the motherboard is important, though.

Re:Distance from the power supply

[TheGratefulNet]

sometimes the circuit (board) does this work.

case in point, ESI juli@ (hate the at-sign. how stupid!)

this card is lab grade (test gear quality) in its a/d and d/a. some people use it for RMAA audio gear testing. not kidding! this is a low noise floor that you won't normally find on internals cards, yet it IS AN INTERNAL card! blows my mind ;)

also supports balanced and unbal i/o as well as 'easy' i2s and very easy coax spdif i/o.

it needs a full height slot and generally is pci-only even though some new pci-e version is supposed to be out soon.

keep it in mind: if you find yourself needing to test audio amps, preamps, dacs, etc - the ESI julia card is about as good as it gets for under $1k or even higher. amazing for audio guys. stupid for gamers but we are not talking about gaming at all.

Re:Distance from the power supply

[TheGratefulNet]

atx supplies are switchers. switchers are HORRIBLE.

you have to do a lot of filtering to make switching supplies sound good in audio. most good audio companies go out of their way to use analog (linear) regulator style supplies.

once you get noise in at very high freq's, its very hard to kill that later. really.

one reason usb to spdfi 'wins' is that the digital power supply matters very little and there is no analog signal to care about until your stereo/amp.

Re:Distance from the power supply

[Smauler]

Where do graphics cards that pump audio out fit in here? I only noticed my card (gtx460) was putting out sound when I changed my monitor to my TV, didn't connect the sound, and it made me jump out of my skin.

I've since just used the dvi-hdmi cable for everything - not bothered with the motherboard sound.

Re:Distance from the power supply

[ChrisMP1]

The switching frequency is well above the limit for audio frequencies, and a good quality PSU will do plenty of filtering. USB to S/PDIF is good for a few reasons: 1) The signal is kept digital, and either differential, balanced or optical for as long as possible. This makes it hard to pick up noise. Digital-to-analog conversion should be done as late as possible because digital signals are very noise-tolerant. 2) Optical connections eliminate grounding issues (less of a problem if you obey #1, anyway). 3) S/PDIF equipment tends to be built to higher quality because it's considered somewhat "high-end" (or, at least, more than a cheap motherboard audio chip). 4) D-A conversion is done far away from the noisy CPU and data buses. It has nothing to do with the SMPS.

You should hear the sound from the vacuum tube guitar amplifier I built. The high tension (300V) is generated by a crappy 50kHz switching boost converter I designed and built myself. The noise on the supply is absolutely awful - 1Vp-p even with only a light load. You can't hear it, though, because it's 50kHz. And that's really low frequency for a SMPS.

Re:Distance from the power supply

[TheGratefulNet]

with some amps, though, they are very wideband and you don't WANT high freq garbage up there. some good phones amps go flat to 100khz and higher. they want to more than more than cover the audio range and they don't 'like' protection or LP filtering.

so, that means you have to care. if you are a source box in front of their amp, you HAVE to care. just giving you some free advice.

keep the high freq stuff out of the audio chain and the guy in the next stage will thank you.

Re:Distance from the power supply

[ChrisMP1]

Well, yes, that is true, it can be bad to drive a big amplifier with high frequency noise. Very bad, in fact - the excess load can do anything from distorting the output to causing extra heat dissipation and killing the transistors. (My guitar amplifier was meant to drive a speaker directly, and it had a low-pass filter between the gain and output stages. The tube was just for the clipping distortion sound; the output stage was a BJT class AB push-pull with a much cleaner power supply.) I was talking more in the context of this article, though - most people who own an amplifier with flat response to 100kHz are not going to be bothering with $50 sound cards, or for that matter even considering integrated audio.

Though, a couple questions (despite some vacuum tube stuff, audio is only a minor interest for me) - 1) What exactly do you mean by "don't like LP filtering"? I can't imagine how a low-pass filter could cause a problem in this case, especially if you just attenuate about 20-30kHz and up. 2) WHY do people even bother with an amplifier that has flat response to 100kHz?? That seems a bit excessive, unless you're playing music for your pet bat... Is there an advantage I'm not aware of?

Re:Distance from the power supply

[epyT-R]

they act as a sound device with only a digital out. some cards include a cable to wire to your sound board's spdif out instead. There are no analog components for the system noise to interfere with (barring egregious digital noise that creates too much jitter).

Re:Distance from the power supply

[Anonymous Coward]

[Citation needed]. Seriously.

Re:Distance from the power supply

[ChrisMP1]

As we all know, "audiophiles with trained ears" are not at all prone to misinformed bullshit regurgitation, confirmation bias, and elitism.

You don't say...

[zonker]

$50 sound card produces better audio than a 50 cent onboard chip.... You don't say.

Re:You don't say...

[MrL0G1C]

Actually their RightMark audio analysis's don't show this at all, frequency response, THD, noise, are all so close between devices that a human wouldn't be able to tell the difference between them, the ignoramuses at TechReport however don't know how to read the graphs/understand the limit of human hearing and came to erroneous conclusions.

Their section on different peoples opinion of the various audio devices does not state the result of the blind listening test and so is useless, why even bother with tests that are not blind?

Re:You don't say...

[gazbo]

I think what's most telling about their analysis is on the 96kHz plots. The Realtek consistently drops to nothing around 20kHz, and yet apparently that didn't mean anything to them other than "look how well these results fit with out hypothesis". Anyone who actually knew something about didgital audio would think "either I've set this up wrong or the drivers/hardware are bust, because this thing is blatantly stuck at 44.1kHz".

The only other thing to be gleaned from the graphs is that running at 96kHz is pointless because the supposedly better cards' performance FUCKING SUCKS past 20kHz.

Re:You don't say...

[makomk]

Anyone who actually knew something about didgital audio would think "either I've set this up wrong or the drivers/hardware are bust, because this thing is blatantly stuck at 44.1kHz".

Or it's just got a low-pass filter with the cutoff set at 20 kHz which can't be disabled. You need one for proper signal reconstruction at the 44.1kHz sample rate, and it's not like most people are going to notice that their onboard sound can't actually output frequencies above 20 kHz in its 96kHz sampling mode.

usb to spdif, then to your home stereo

[TheGratefulNet]

there, done in one.

cheap cmedia usb sound dongles (not all dongles are cmedia, in fact most are not so you have to shop carefully) and also the burr brown PCM series all do a decent job of converting 44 and 48k audio (including dvd audio downmixed to 2.0) to spdif.

everyone's avr, today, has opto in. the sound card dongles send out usb audio over opto to spdif-in of your home stereo. if its 5.1 or newer, it will accept opto just fine. (aka toslink).

nothing else to care about, pretty much. let your stereo (or DAC) do the heavy lifting. usb audio is the way to go (for future, use UAC2, usb audio class 2 which works fine with linux and some hacks on windows at 24bit and 192k, personally verified to be bit-perfect).

Re:usb to spdif, then to your home stereo

[Tyler Eaves]

I remember when I obsessed over such things. These days I mostly don't worry about it and spin vinyl. Sounds better anyway.

Re:usb to spdif, then to your home stereo

[TheGratefulNet]

these days, i'm not sure ANY onboard sound forces a resample.

soundcards 'offer' freqs to the sound system. back in the dinosaur days, they would offer only 48k and the o/s would have to resample.

but xp and win7 all deal with 44/48 split just fine. linux always has.

onboard audio (if its native spdif via jacks or headers) is usually bit perfect. its there to give 5.1 and even 7.1 digital out. the days of speaking 'only 48' ended 5yrs ago or more (I forget). a long time, at any rate (lol, any rate!)

If I was spending $50

[slackware 3.6]

it would be on a cheap video card before a sound card. I never bought any sound cards after they started puting them on the mobo. Sure the sound could be better but I have a stereo for playing tunes and if i'm playing games at night I'm using headphones anyway. A better soundcard is a non-issue for most users.

Re:If I was spending $50

[Smauler]

Many video cards now put audio out too - I've got an old DVI to HDMI cable from the back of mine, and it's spitting out audio happily.

Re:If I was spending $50

[lowlymarine]

The problem is that if you spend $700 on a video card today, you will get the same Vsync-capped performance you would have gotten out of a $350 card, and six years from now, as you suggest it should last, you would have a brick that can't handle anything remotely modern while the hypothetical other guy would only have a three-year-old $300 card that beats your $700 six-year-old card into the ground.

Six years ago, the Core 2 Duo X6800 and GeForce 7950GX2 were the top-of-the-line parts, costing a grand and $700, respectively. Within two years, both were getting clobbered by parts that cost half as much. Today that $1700 combination wouldn't even be competitive with a $75 A6-3650.

Sound Blaster is dead

[Anonymous Coward]

I dumped my Sound Blaster Live! 5.1 for Realtek ALC888 and never looked back. I never heard such clarity until I went with this onboard chipset.

Also, I tried to compare a $99 Razer Barracuda AC-1 with the ALC888, all I could make out is that the back channels get more bass than the front. I'm guessing there's a placebo effect in play here with the little DSP cards...

$50? Try a $250.

[X0563511]

I have this for studio purposes, but this thing sounds beautiful.

If I chose to, 96khz 24-bit. 2-in, 2-out, SPDIF support if I chose to use it. (technically 4 in 4 out, but that's mono.)

solder 3 wires and get good toslink out

[TheGratefulNet]

a DIY I wrote on how to open up a cm102 (cmedia usb audio dongle) and find the 3 solder pads you need to connect in your own toslink (TOTX) opto transmitter for your home stereo:

http://farm5.static.flickr.com/4086/5052505190_07d7ec5903_b.jpg

http://farm5.static.flickr.com/4148/5052506250_c71b26586a_b.jpg

it was just that simple. there was already an onboard cheap-o toslink sender but I prefer the standard square block style.

the TOTX part is a dollar or so at digikey or mouser (suppliers). the usb dongle is $15 or less, often much less. make sure its cmedia and cm102. it will work very well then.

usb powers it and you know its working when you get the red light out of the toslink end ;)

I'm not sure it passes dts or dolby digital but its fine for 44.1 cd audio (and mp3) as well as 48k dvd downmix to stereo 2.0

Re:Hypothesis Number Two: Is There a Cyber Stalker

[Anonymous Coward]

Do you have a newsletter?

Not really impressive review

[gweihir]

These are not audio people and did not have an audio-expert look at their write-up. Why. They got the very well known very-low-cost / not-very-good audio OpAmp NE5532 P wrong as NE55329. No audio-expert would make that mistake. It is not a number, it is an identity that experts immediately recognize.

I have to say that this puts a big question-mark on the whole test for me.

Re:Not really impressive review

[MrL0G1C]

Their analysis of the RightMark audio benchmarks are also complete fail, they do not seem to either understand the Decibel scale / they can't read a graph / they don't understand the limits of human hearing or how Psychoacoustics fit in with the graph.

http://en.wikipedia.org/wiki/Psychoacoustics

For instance:

check out the Xonar DGX in the frequency response plot. Again, it starts dropping at lower frequencies than the competition

A less than 1/2 a Hz difference in a 10,000Hz frequency is not going to be audible, stupid people.

And:

Also worth noting: the higher noise and distortion exhibited by the integrated audio in several of the graphs.

The only way you'd actually hear any of the noise is if your amp+speakers where out-putting at ~150Db but not playing any sound (noise only), unlikely I think.

As for the distortion level, it's far too small to be humanly perceptible.

Re:Avoid Asus AT ALL COSTS

[spauldo]

Just a guess, but I'd say it's because there's stories just like yours for every manufacturer out there*.

Case in point: hard drives. Ask 20 people what hard drives they've had trouble with, and you'll find they pretty much average out as all of the companies having issues. I use Seagate, but you'll find a lot of people here who swear they're the worst drives on the market.

Another example: T-Mobile. I had nothing but trouble with them. They would, for no reason, forgo the automatic draft from my Visa and then shut me off for non-payment. Their customer service was horrible. When I called them at the end of my contract and told them I wanted my service cut as soon as the contract was over, the sales drone threatened me that if I didn't pay the final bill, they'd sue me. I hadn't said anything about it up to then, so this was just out of the blue. (Of course, I knew he was full of shit, and intentionally didn't pay my final bill because of it.)

With all that, T-Mobile has an excellent reputation for customer service and very few people I know have issues with them. Go figure.

The article was about the difference between soundcards and integrated sound, and just happened to use Asus cards for the testing. Your last paragraph was on topic, but the rest of your post wasn't.

* There are a few manufacturers that have earned widespread derision, like PC Chips for its fake cache chips or SCO for judicial douchebaggery. Asus isn't anywhere near that level.

Re:Discrete sound

[Gadget_Guy]

Discrete sound...Is farting silently

No, that is a soundless indiscretion.

Re:can that possibly matter with optical digital c

[TheGratefulNet]

distance generally won't matter MUCH on toslink cables.

yes, jitter matters but only at very high end. your avr system, even if its a $2k monster, is still never going to be 'good enough' to care about picosecond level jitter in the spdif stream.

toslink does blur the digital audio and it changes timing slightly, randomly. that's jitter.

shorter cables do less damage.

at least on plastic toslink, there's no concept of reflections. back-energy does not happen on toslink. but it DOES happen on coax/spdif. shorter does matter here. think of it this way: you send a signal to the far end of the coax (again, not opto, but coax) and it sends most of its energy there but reflects back some. that takes some time to travel along the cable back to the start. it then bounces back again, along with new energy from the last pulse of the transmitter. this goes back and forth and blurs the 'location' or timing of the 1's and 0's.

now, if your DAC system fully and completely locally (!) reclocks, you are fine. if not and if it DEPENDS on the timing of each and every 1 and 0, it would 'dump out' the 16bit audio word at the wrong time since one of those 'clock edges' was off by a bit, due to the reflection blur. it happens but its test-equip level, not 'wow, that sounds horrible' level. very subtle but once you have $10k-class spkrs (etc) you CAN hear blurring of the timing.

long answer: but in real world, you don't care about cable length in digital audio. best to stay 6' or 10' or less. if you go farther, you MAY want to consider a bridge (like data link bridge; fully receives spdif stream then recreates it via a receiver/transmitter combo; not a repeater but a full recieve, digest, regurg, retransmit pair of rx/tx chips).