Head-Fi.org › Forums › Equipment Forums › Computer Audio › The Xonar Essence STX Q/A, tweaking, impressions thread
New Posts  All Forums:Forum Nav:

The Xonar Essence STX Q/A, tweaking, impressions thread - Page 288

post #4306 of 4958
Quote:
Originally Posted by kingwaffle View Post

Is there a "best" practice regarding the sample rates? I understand music is typically 44.1khz, if I set the Xonar control panel to anything higher music will get re-sampled. Is this particularly a bad thing? Also, if I do change the sample rate to 192khz, am I supposed to also set this sample rate in the "Advanced" tab of the speaker properties? 

 

I have also read that you should set the gain on the headphones to match the impedance, but was curious as to whether the potential damage is brought about if/when the output volume is set too high, or if simply setting the gain alone is enough to do it. I swear the cans sound better with the higher "medium" gain setting in Xonar, but am worried since it suggests "Normal Gain" for my cans (42 ohms).

 

Normally it is best to not resample anything because a./ there is varying quality between different resamplers b./ it is CPU-heavy on a PC. If you don't resample you just know the sound is ok,which is good for the mind!

 

If you choose to resample you must know what you do, and use a good resampler, normally one associated with the player you use and in the form of a plugin. Also, the setting in Windows control panel is only active if you run the Windows sound stack/mixer, which you shouldn't for music, use WASAPI or ASIO - selectable in the player - as they talk directly to the soundcard and should be used if resampling is going on. The new Soundblaster Z/ZxR even require upsampling if you run in the "stereo direct"-mode, which is only active @192kHz. In the Creative case "stereo direct" bypasses the DSP on the card and lets the drivers talk directly to the DAC.

 

So what are good upsamplers/resamplers? If there are any. It's debatable, but according to my own findings so is the standard resampling plugin in foobar2000 not the best out there(there is other plugins that can be used), I think it colors the sound ie no upsampling is normally better. There's better results using the built in upsamplers in uLilith and Musicbee imho. But the best upsampling for PC that I've found so far is the  plugin in Deadbeef-player under Linux which uses the original "Secret Rabbit Code" for upsampling - I think it sounds excellent! Also, under Linux it's essential to use ALSA instead of Pulseaudio for playback. Like it's important to use WASAPI or ASIO under Windows, as those modes bypasses all sound stuff going on under the hood in the Windows OS.

 

For gaming sound you can forget all of the above, just use the normal recommended settings.

 

About the gain on the Xonar-cards, it is just gain and not anything else so it doesn't change any impedance or anything, you can just forget about what impedance is written in the drop menu and use the setting which sounds the best with your headphones. It's not strange if a higher gain actually sounds better, as the HP-amplifier on the Xonars is a small amplifier and all amplifiers sounds better with higher gain as long as it is under the clipping limit. Higher gain normally means lower distorsion and better S/N in the amp, it can amplifiy underlying noise in the source signal but the amp itself works better with higher gain. As I said, use what sounds subjectively the best!

post #4307 of 4958
Quote:
Originally Posted by xeizo View Post
Higher gain normally means lower distorsion and better S/N in the amp

 

In fact, the opposite is true, higher analog gain generally increases distortion and noise, because there is less negative feedback to correct errors, and the input noise voltage is amplified more. The main reason why people think higher gain sounds "better" is simply because it is louder, and louder sound is perceived by humans as having better quality. Casual audio comparisons are rarely performed with accurate level matching. By the way, the Xonar Essence ST/STX have no analog volume or gain control at all, the gain setting simply adjusts the digital volume by -18, -6, or 0 dB from the lowest to highest impedance mode.

post #4308 of 4958
Quote:
Originally Posted by stv014 View Post

 

In fact, the opposite is true, higher analog gain generally increases distortion and noise, because there is less negative feedback to correct errors, and the input noise voltage is amplified more. The main reason why people think higher gain sounds "better" is simply because it is louder, and louder sound is perceived by humans as having better quality. Casual audio comparisons are rarely performed with accurate level matching. By the way, the Xonar Essence ST/STX have no analog volume or gain control at all, the gain setting simply adjusts the digital volume by -18, -6, or 0 dB from the lowest to highest impedance mode.

 

I agree that input noise is amplified more, which I wrote, and I agree that "higher" normally is perceived as "better". That's why there's "loudness wars" going on in the recording studios.

 

But if you see the amplifier itself as a isolated entity so does most if not all amplifiers indeed perform better with higher gain, as long as it doesn't clip. Signal to noise ratio is better and there's lower overall distorsion from the amp itself. Of course no chain is better than it's weakest link, that's why I recommend selection of what setting to use based on taste rather than any thumb rule ....

post #4309 of 4958
Quote:
Originally Posted by xeizo View Post

I agree that input noise is amplified more, which I wrote

 

No, you wrote that noise in the source signal is amplified more relative to other components of the source signal, which is wrong. I meant the input noise of the amplifier itself (which is always there, even with no source connected) is amplified more, and therefore there will be a higher noise voltage on the output. For a given signal voltage on the output, if it can be achieved with low gain + higher volume, then low gain will have a lower noise voltage and thus better SNR.

 

Quote:
Originally Posted by xeizo View Post
there's lower overall distorsion from the amp itself

 

Why do you think that would be the case ? Higher gain in an amplifier reduces negative feedback, and that increases distortion. It can easily be seen on measurements like this one, or those of the O2 (link not allowed, but you can easily find them), that the higher the gain is, the worse the THD+N becomes. Except for some unusual cases, the only advantages of higher gain are:

- higher maximum output voltage if the lower gain cannot already reach the clipping level

- better stability (if the amp is poorly designed and unstable at low gain)

 

But all this discussion is moot here, because, as I already explained, the Xonar STX has digital gain control.

post #4310 of 4958

I guess I should've specified how I listen to my music in Winamp.

 

I typically enable ReplayGain, max the volume out in Winamp, and then listen to my music at what I would consider a normal volume.

 

If I'm understanding what I'm reading, when I enable the "medium" gain in the Xonar control panel I am more or less just off-setting the ReplayGain, correct? I've dabbled in audio editing and such, so I like to think I know what I'm doing but I might be off. 

 

Am I more or less doing a "hard limit" on my music by doing that? If so, that would explain why my music "sounds better", simply because it's louder at the cost of dynamic audio. 

post #4311 of 4958
Quote:
Originally Posted by kingwaffle View Post

Hello. I have been lurking off and on for awhile now. I currently have this sound card (STX Essence) as well as a pair of Audio-Technica ATH-A900X's. Also have a pair of 700's that I'm trying to decide what to do with.

 

I primarily listen to music (mp3s/FLAC & Winamp w/ the WSASPI plugin), but also occasionally dabble in FPS's. (Namely TF2 and BF3)

 

I tried searching the thread, but came up fairly empty as regards to my question:

 

Is there a "best" practice regarding the sample rates? I understand music is typically 44.1khz, if I set the Xonar control panel to anything higher music will get re-sampled. Is this particularly a bad thing? Also, if I do change the sample rate to 192khz, am I supposed to also set this sample rate in the "Advanced" tab of the speaker properties? 

 

I have also read that you should set the gain on the headphones to match the impedance, but was curious as to whether the potential damage is brought about if/when the output volume is set too high, or if simply setting the gain alone is enough to do it. I swear the cans sound better with the higher "medium" gain setting in Xonar, but am worried since it suggests "Normal Gain" for my cans (42 ohms).

 

I realize these questions have been probably asked/answered before. I tried searching but it didn't work all that particularly well. 

 

Almost all DAC's actually perform best at 48KHz sample rate though you would be hard pressed to hear any difference as even at the lower performing sample rates they still performs at a very high level.

 

The Xonar cards all perform less well at 44.1KHz due in large part to increased noise as there is no crystal time reference for that sample rate but once again except in rare occasions you probably wouldn't hear it. There are some IEM's with very high sensitivity where you would likely hear the increase in noise caused by 44.1KHz sample rate.

 

The gain control is simply reducing the bits by 1-3 bits. Reducing volume by 6-18db does exactly the same thing. There in no reduction in quality if you perfectly match the respective volume for each gain setting Both "gain" & volume are handled in the same way. Increasing "gain" does actually limit the number of usable volume steps in the windows volume control depending on your headphones but if you use a program the has it's own volume control this point is moot as many have in excess to 24 bit volume controls.

 

Unless you have separate way to control volume other than through the PC's or programs volume control I highly recommend you do not use ASIO as you will have no volume control. People think that going bit perfect & using an external means of volume control is the "cats meow"  but this is not always true, in fact it is rarely true I have found. This due to inherent losses in devices used to control volume outside the computer. I had a preamp that I used to control volume for a long time but then I took it out of the signal chain & went directly from my computer to the speaker amps & it was like removing a cloud over my music, everything sounded much more there as in feeling as though I was in the studio with the musicians on my best recordings. I will likely never go back to using an external volume control.

 

I recommend using the lowest "gain" setting possible that gives adequate volume as this will give the most volume choices in the widows volume control hence you will be able to find the most perfect volume level for each piece of music more easily. If you use a program such as foobar, go ahead & use it's volume control. Setting to the proper "gain" will prevent accidental damage to your headphones & minimize the risk of hearing loss as well

post #4312 of 4958
Quote:
Originally Posted by stv014 View Post

 

No, you wrote that noise in the source signal is amplified more relative to other components of the source signal, which is wrong. I meant the input noise of the amplifier itself (which is always there, even with no source connected) is amplified more, and therefore there will be a higher noise voltage on the output. For a given signal voltage on the output, if it can be achieved with low gain + higher volume, then low gain will have a lower noise voltage and thus better SNR.

 

 

Why do you think that would be the case ? Higher gain in an amplifier reduces negative feedback, and that increases distortion. It can easily be seen on measurements like this one, or those of the O2 (link not allowed, but you can easily find them), that the higher the gain is, the worse the THD+N becomes. Except for some unusual cases, the only advantages of higher gain are:

- higher maximum output voltage if the lower gain cannot already reach the clipping level

- better stability (if the amp is poorly designed and unstable at low gain)

 

But all this discussion is moot here, because, as I already explained, the Xonar STX has digital gain control.

 

 

Due to the "gain" being digital noise remains at a constant level no matter the "gain" level. Only the top 1-3 bits are cropped & as a result windows only sees 21-23 bit of a 24 bit system & can only be turned up that far. As a result maximum S/N ratio is reduced as noise remains constant. This is fine though as analog noise from the opamps will likely be little more than  the noise output of the DAC anyway. Minus 18db  from 120db is still 102db which is fine unless you are using IEM's with 122db sensitivity at 1mw. 127db is the spec of the DAC itself the headphone amp is speced at either 117db or 120db depending on where you read. If those bits had come off the bottom of the scale they would have been totally inconsequential as minus 18db from 144db is 126dbwhich is still under the 120db spec of the headamp & 2db under the lineout amp but since those bits come off the top they are consequential if only under limited circumstances. Note that the bottom bits do help reduce distortion though at this level you are not going to hear it anyway.


Edited by germanium - 5/6/13 at 12:34am
post #4313 of 4958
Quote:
Originally Posted by germanium View Post

Due to the "gain" being digital noise remains at a constant level no matter the "gain" level.

 

In case it was not obvious enough, I was referring to the effects of analog gain (in response to incorrect information in another post), and specifically pointed it out that it does not apply to these cards.

 

Quote:
This is fine though as analog noise from the opamps will likely be little more than  the noise output of the DAC anyway.

 

Since the noise level is constant, it does not matter for the (digital) volume control if it comes from the DAC or op amps.

 

Quote:
Minus 18db  from 120db is still 102db which is fine unless you are using IEM's with 122db sensitivity at 1mw.

 

It is only 102 dB if you listen to those IEMs at full volume, which is too much even at the lowest gain.

 

Quote:
127db is the spec of the DAC itself the headphone amp is speced at either 117db or 120db depending on where you read.

 

It is 120 dB in the datasheet of the TPA6120, and 117 dB in the specs of the card (where it includes noise from all the other components, and is thus obviously slightly worse). However, it is 117 dB relative to the full scale ~7 Vrms output at 48 kHz sample rate. At 44.1 kHz, it drops to about 110 dB, maybe 111 at best. Let's say we use Shure SE425 IEMs, which have an impedance of 22 Ω at 1 kHz, and need 0.02 Vrms for 90 dB SPL. The noise voltage is then 7 * 10^(-117/20) * 22/(22+10) = ~6.8 uVrms at 48 kHz, and 14-15 uVrms at 44.1 kHz, regardless of the gain or volume. That translates to 20.6 and 27-27.5 dB A-weighted noise SPL, respectively. The latter is definitely audible (I can confirm it is even with full size semi-open headphones that have poor isolation), and the former might be audible with the very well isolating IEMs while no music is playing. Some very efficient full size headphones could also have audible noise at 44.1 kHz.

 

Quote:
18db from 144db is 126dbwhich is still under the 120db spec of the headamp & 2db under the lineout amp

 

Once the level is attenuated by 18 dB, the line out no longer has 124 dB SNR anyway (a number that I have never seen confirmed by any source other than ASUS, by the way, Stereophile measured ~120.5 dB relative to full scale, and I get about 118-120 with the ADC noise factored out, and that is at 48 kHz). The percentage of the quantization noise relative to the total noise does not change, only the signal level is lower.


Edited by stv014 - 5/6/13 at 3:58am
post #4314 of 4958

Interesting discussion, but back to tweaking! biggrin.gif

 

I received the dual OP LME49990 from Frugalphile yesterday, vary fast delivery! Haven't had the possibility to test it yet, anyone else?

post #4315 of 4958
Quote:
Originally Posted by stv014 View Post

 

In case it was not obvious enough, I was referring to the effects of analog gain (in response to incorrect information in another post), and specifically pointed it out that it does not apply to these cards.

 

 

Since the noise level is constant, it does not matter for the (digital) volume control if it comes from the DAC or op amps.

 

 

It is only 102 dB if you listen to those IEMs at full volume, which is too much even at the lowest gain.

 

 

It is 120 dB in the datasheet of the TPA6120, and 117 dB in the specs of the card (where it includes noise from all the other components, and is thus obviously slightly worse). However, it is 117 dB relative to the full scale ~7 Vrms output at 48 kHz sample rate. At 44.1 kHz, it drops to about 110 dB, maybe 111 at best. Let's say we use Shure SE425 IEMs, which have an impedance of 22 Ω at 1 kHz, and need 0.02 Vrms for 90 dB SPL. The noise voltage is then 7 * 10^(-117/20) * 22/(22+10) = ~6.8 uVrms at 48 kHz, and 14-15 uVrms at 44.1 kHz, regardless of the gain or volume. That translates to 20.6 and 27-27.5 dB A-weighted noise SPL, respectively. The latter is definitely audible (I can confirm it is even with full size semi-open headphones that have poor isolation), and the former might be audible with the very well isolating IEMs while no music is playing. Some very efficient full size headphones could also have audible noise at 44.1 kHz.

 

 

Once the level is attenuated by 18 dB, the line out no longer has 124 dB SNR anyway (a number that I have never seen confirmed by any source other than ASUS, by the way, Stereophile measured ~120.5 dB relative to full scale, and I get about 118-120 with the ADC noise factored out, and that is at 48 kHz). The percentage of the quantization noise relative to the total noise does not change, only the signal level is lower.

I was not disagreeing with you, only expanding on it with my own observations. Yet you seem to want to treat everything I say with a certain contentiousness as if I'm somehow disagreeing with you which I am not. I never recommended listening to IEM's at those kind of volumes just exposing the fact the IEM's with sensitivities such as 122db @ 1mw are definitely going to reveal noise in quiet portions of music with this card. This it will do no matter the volume setting as there is no attenuation in the analog domain with this card. While I'm not a big fan of analog attenuation especially when carried out by a separate device outside the computer High sensitivity IEM's is one situation where such may be needed but only if the buffer following the volume control is sufficiently quiet which is not an easy task to accomplish with said IEM's.

post #4316 of 4958

Hi, idk if this has been brought up, but under the mixer tab in xonar audio center, you can choose different volume levels for the left and right channel.

For this slider, it appears that leaving it at default will be bitperfect, and that maxing this slider out will apply a boost (unlike the main volume slider, which should be maxed out if there is a seperate amp).

Is this correct?

 

Lastly, is there a driver that autodetects sampling rates? I can't believe asus didnt give an option for it.

And on that note, ASIO bypasses the resampling done in the drivers, correct? but wasapi and kernel streaming do not?


Edited by mindbomb - 5/7/13 at 11:40am
post #4317 of 4958
Quote:
Originally Posted by mindbomb View Post

Hi, idk if this has been brought up, but under the mixer tab in xonar audio center, you can choose different volume levels for the left and right channel.

For this slider, it appears that leaving it at default will be bitperfect, and that maxing this slider out will apply a boost (unlike the main volume slider, which should be maxed out if there is a seperate amp).

Is this correct?

 

Lastly, is there a driver that autodetects sampling rates? I can't believe asus didnt give an option for it.

And on that note, ASIO bypasses the resampling done in the drivers, correct? but wasapi and kernel streaming do not?

WASAPI does bypass the windows sample rate converter but not the volume control. Unsupported files will not play with WASAPI, in other words 88.2& 176.4KHz will not play with WASAPI on this card.

post #4318 of 4958

yes, the windows resampling is bypassed with wasapi, but the card itself still resamples to whatever is selected in the xonar audio center unless you use ASIO?

post #4319 of 4958

even wtih asio, it will resample to whatever it is in the asus control panel.

post #4320 of 4958

Last night, I did check on my xonar dx's digital output, and it appeared that ASIO disabled the card's resampling whereas wasapi did not.

I assume it is the same for all xonar cards.


Edited by mindbomb - 5/8/13 at 7:30pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Computer Audio
Head-Fi.org › Forums › Equipment Forums › Computer Audio › The Xonar Essence STX Q/A, tweaking, impressions thread