As the title asks, has anybody ABX'ed their motherboard's onboard sound output with a PCI or USB soundcard? Were you able to tell the difference? Was one much better than the other? Just to be clear, I'm talking about stereo music listening on headphones.

The reason I ask is as follows. I've been listening music using "audiophile" headphones for a long time. After my last pair succumbed to old age last year I started searching for a new pair. In my search I discovered the suggestion (on this and many other forums) that my motherboards onboard sound was holding back my headphones. So I went on a quest to upgrade my sound card using a USB(coax for the 0404 USB) powered device.

To put things succinctly, apart from subtle (don't think I could pass a blind test identifying them) changes I cannot hear any difference between all the hardware I tried apart from some low level noise (I cannot hear this noise when music is playing) from the motherboard's output. My motherboard doesn't have any cracks, pops, hiccups, or any of the stuff people typically complain about.

I'll list the equipement I've played with. These have all been combined in various ways so I won't list all the combinations. Kubuntu 8.04. Asus K8N-E motherboard (almost four years old now) with a onboard Realtek ALC850. MP3s of varying bitrate upto 320kbps (lame insane). Edirol UA-1EX. EMU 0404 USB. CMOY amp(9V and 18V with OPA2227). Headroom Total Airhead amp. Behringer 1202FX mixer. Denon D2000/D5000 headphones. Grado SR225. Audio Technica AD900 headphones. DT880 headphones.

I just cannot hear any "sound quality" improvement. If I were to do a blind test I'm confident only the low level noise from the motherboards onboard sound would give that away. What's going on? As far as I can guess, it is one of the following.

1). My motherboards onboard sound is much better than I've been led to believe.
2). People are full of it (perhaps the same as option 1).
3). I need to see an audiologist.

I don't mind if option 3 is right answer.

Just to be clear, I did not do a blind test on any of this equipment. I just volume matched and listened that way.

On board is often fairly good to very good, but it varies quite drastically from system to system.

Some reviewers use RMAA on sound cards and motherboards. THis gives you a pretty good idea how good each device is.

Reviewers have run RMAA on the ALC850 many times on a lot of difference boards. It was a popular chipset at the time. The results I've seen vary from average to good on all tests. The RMAA results when I ran it were in line with the other results I've seen for that chipset.

But I'm not really all that concerned with measurements at this point. Nowadays almost everything measures better than the ALC850 chipset. I'm curious as to if people have heard a difference in some sort of controlled testing.

The ability to tell a difference is also limited by the frequency range of your hearing. From my experience, more expensive soundcards tend to make difference mostly at deepest bass and highest treble (and usually have better noise rejection).

HD-Audio is much better than AC'97 sound quality wise.

I think it's mostly outdated knowledge about on-board sound that lead people to believe it must be avoided.

I have tried now and it is a PITA to do anything close to ABX regarding two soundcards. I set up two instances of media player classic looping a sample and set them up to use the two cards (Realtek HD on Asus P5K and Asus Xonar DX). I fairly synchronized the playback position and volume. Then I dived under the table and switched the outputs manually. It was not really an ABX test (I would have to have someone else crawl the floor, switch the outputs and write down the results).
I have Senn. HD 215 headphones (closed) and the difference I heard was next to nonexistent. When I focused on the treble and bass, I would say that Xonar goes a bit deeper (to infrasonic where you more feel than hear the sound) and I couldn't really tell the difference at treble (this might be due to considerable delay when switching the outputs, however). Regarding noise, I haven't heard any audible noise coming from Realtek (even when playing digital silence) but the room was not completely silent.
Then I just listened to some of my favourite pieces through Realtek (no direct comparison). I was fairly surprised that the sound didn't feel any different. There is perhaps a very slight timbre difference in metallic sounds (guitar strings, hihat) but it might be just my imagination.
Overall, I would be more than satisfied with Realtek, I couldn't spot any problematic areas.

Nothing really enlightning here in my post, but willing share my thoughts.

According to my experiences (no abxing though) I've come to conclusion that my realtek ALC880 and ALC888 chipsets give sound that is good enough. Apart from the resampling problem (which is another story ) the sound and quality of my onboard soundcard is perceptually the same as my DVD player's audio out. Again, I can't verify if there really is difference though, but to me it's not a problem if there is, because in a normal room environment (noise floor somewhere around 40dB, an air conditioner humming in the next room) the possible higher noise floor of the realtek chipsets is the only difference that i'm able to hear. Other than that, truly and totally satisfied with the quality. In addition, I haven't trained myself to hear the most subtle differences, what is more important to me, is the ability to play my music in a way which isn't annoying or distrating my concentration from the music itself. I hope I haven't made a TOS 8 violation here.

Regarding to the ABX part, even though it's purely my opinion with the equipment I'm having access to, true ABX test of two different soundcards in two different systems or a soundcard vs DVD player is somewhat difficult to accomplish. Somewhat intriquing question however.

Thanks for doing all that Martel. Your impressions mirror my own. I do not have most of the original equipment listed above so I ran a far simpler test.

Well I've spent some more time with RMAA. I don't know if the how valid the results are because I was using my motherboards built in stereo in. As you said Martel, the frequency graph collapses (slope downward sharply) at the low and high end. I then added a 18V Cmoy and the graph was much more flat. There was still rolloff at the ends, but it was probably out of what I could hear at the high end and beyond what headphones can do at the low end. I did the Cmoy RMAA loaded and unloaded.

What did I hear as far as using the CMOY vs directly into the sound card?

Speaking as an audiophile focused on details, I think there was something subtle.

Speaking from a wider point of view, if there were any perceived differences they were gone between the second or two it took me to switch the headphone jack. I'm planning to build a small switching box so this is easier to look at, but I say that in a blind test they will sound the same.

And just to remove idea the headphones may be too poor for these tests, I'm using the Denon D5000. On a certain forum people have complained the headphone is too revealing of flaws, and so at least one person does not use the headphone because of that. I'm not claiming what I just said is true, I'm just trying to say that the headphones are capable of revealing small changes/details.

So the differences were subtle or nonexistent. However, now that I've run RMAA tests, I feel like getting a better sound card

I noticed this same thing by accident when listening to some test tones. The lowest tone I could hear with the onboard card was the 30hz tone. With the M-Audio I could hear the 20Hz tone quite easily.

This is an old thread that I started, but I've done some more tests since my first post. Perhaps some audio skeptics will find the information here useful.

Most of my tests up this point have been rather haphazard. I never controlled it in any serious way. Since the first post in this thread I've maintained my curiosity in DACS and dedicated headphone amps because of all the people who say they are required and how they hear "night and day" differences.

I recently purchased a HT Omega Claro Halo. To those of you who are not aware of this card, it is a stereo PCI sound card with a dedicated TI headphone opamp (TPA6120A2). The opamp seems to have a good reputation. I contacted HT and they let me knows it runs off the +12V and -12V rails. My assumption is that the headphone amp can swing 24volts. HT states that the headphone amp will work with headphones up to 600 ohms.

I wanted to compare the headphone out of this card to my motherboards onboard sound. To do this in some sort of a controlled manner, I put together a small setup with two input jacks (for each sound card), a toggle to switch between the two inputs, and an output jack for headphones.

Then, using a 1khz signal, I volume matched using a multimeter. I tried the test using MP3s, CDs and FLAC files. In all cases, I played the same song through both sound cards simultaneously and I tried my best to get it so they played synchronously. I also ran the test at different volumes (always matched).

Anybody care to guess the end result before reading the next paragraph?

It was difficult for me to detect differences. If the music was lined up just right, I could toggle between the two sound cards without any major audio indication that I switched. I realize that this wasn't a true blind test, but after toggling the switch a few times I had usually forgotten what card went with what direction of the toggle switch.

With a pair of 25 ohm Denon D5000s, there was a subtle filling in of the low end when using the Claro Halo as opposed to the onboard sound. It was easier to hear when the bass was artificial or generated and when there was a portion of music with mostly bass frequencies. Once music came in from the mid range and high end, it was much more difficult to detect the difference in bass. I really had to focus to hear it.

With a pair of 600 ohm AKG K240s I could not tell a difference at all.

The differences I heard are mirrored in RMAA tests. Loaded with either headphone, the Claro Halo headphone out frequency response is ruler flat till about 17K. With my onboard sound there is some bass rolloff in the audible range when loaded with the 25 ohm Denon D5000. When using the 600 ohm K240 with the onboard sound, the frequency response is mostly straight (although it looks a little drunk) till about 17K.

The Claro Halo measures much better than my onboard sound. Yet, apart from computer noise when music was no playing, the differences between the two are very subtle with low impedance headphones. It's common practice to recommend a dedicated amp for low sensitivity high impedance headphones, but I could not verify such a recommendation during my tests. All the things (body, bass, life) that are supposed to be missing with "improperly" amplified 600 ohms AKGs were present with both my onboard sound and the Claro Halo. More than anything else, the Claro Halo was able to put out a much higher volume.

I suppose it is possible that my hearing at the low and high end is not that great. If so, this test is flawed in that sense. Maybe I'm actually too deaf to be considered a serious audiophile. My wallet will be lighter for it

The biggest potential issue with motherboard sound IIRC is the power supply rejection, and the effect that system loads have on background noise. CD-ROMs are notorious noise sources (I even had an old PCI sound card sound like an airplane at takeoff whenever the CD-ROM drive spin up!). But even the task switching behavior of the CPU can be significantly audible - for instance, if a single process is in a tight run/sleep loop, it can in the worse case scenario wake up every 1ms on Windows. This causes a 1khz square wave on the power rails and is surprisingly easy to hear when no music is playing, with suitably sensitive phones like most IEMs.

Once you get past noise, as you've discovered, you get into finer points of frequency response and distortion where a considerable amount of evidence suggests it is less important than many people think.

Your post prompted me to perform a casual test between the headphone output from the onboard soundcard in the PC (which I had never used before) and that from a 5-6 year old Apple portable which I use occasionally when out of reach of a "proper" headphone socket. The headphones were Sennheiser HD600s which have a relatively large impedance which varies strongly with frequency. The PC motherboard sound chip is a Realtek ALC850 but a quick look did not reveal the audio chip in the Apple portable.

The difference between the two is apparent in a casual test with the bass on the portable being noticeably deficient.

Checking the difference between the PC headphone output and the hi-fi headphone output fed from an external sound card involves a short walk because the noisy PC is not in the same room as the hi-fi. I could not claim to hear a difference but thought there probably was (but one usually does). If there is a difference it is a lot smaller than the difference between the two builtin sound cards.


No. The headphone output from the portable may not be audibly neutral but it does not intrude on my enjoyment of music. If I can reach a "proper" headphone socket than I will use that but will make no significant effort otherwise.