TR Forums

Recently I remember sluggo's post on this thread/

sluggo wrote:

The laptop's output is biased off of (at least) a 3.3V supply, so a clipped line level output is not possible. Besides, the laptop's volume controls are attenuators, so they can't produce a clipped output.

Well it seems most laptop's volume control (including mine) is actually Windows master volume control, so it seems what applies to laptop volume control also applies to Windows master volume control. So of Windows volume control are attenuators, do they yield unity gain (0db) at 100% volume? And if it's not the case, what at volume percentage does it yield unity gain?

It seems it also depends on the sound card/codec being used. How about Realtek? More specifically, Realtek ALC892 used in Gigabyte H77N-WIFI mobo? Where could I find such information?

After Googling a little, I found the manual for Realtek ACL658 instead of ACL892. The Analog Performance Characteristics table says the Master Volume (FRONT/SURR/CENT/LFE) attenuation range has the minimum value of 0dB and the maximum value of -94.5dB. Huh? Isn't that supposed to be the other way around? But fine, let's assume the minimum and maximum value in the manual actually goes the other way around, which means the maximum attenuation range is 0dB. It means 100% volume control is safe, doesn't it? (ie no clipping) So it means the master volume control outputs unity gain (0dB at 100%), am I correct?

But then again, is the maximum gain value the same (0dB) for all Realtek version?

Thanks.

Attenuation means "diminish", so the way it's stated makes sense to me. Ie, the max the signal can be diminished by is 94.5 dB, giving you this amount of volume range. This is normally how digital volume control works in codecs/DACs. So no, there is no clipping per se here. What you are concerned about is the resulting output voltage of the analog output stage from the DAC. For some amps, it could be that minimizing the attenuation of the digital signal may create an output stage voltage that induces some amount of distortion, either in the output stage itself, or in producing a signal that is a bit high voltage-wise for what ever amplifier you are plugging into the analog output (ie, headphone jack) on your motherboard. In such situations, attenuation of the volume might be somewhat beneficial.

The main volume control of Vista/7/8 seems to yield 0db at 100%. I leave Windows' and foobar's volume levels at 100% on all three of my computers. My laptop has an older Realtek codec; the desktop has Asus' "shielded" implementation of the ALC898 (which they call SupremeFX III); and the HTPC uses Nvidia's HDMI audio. So far I have not heard any distortion/clipping out of any of them, even when playing obnoxiously loud albums.

WinXP seemed to work a little differently. I recall hearing distortion out of a SB Audigy 2ZS card when the volume was set at 100%.

In the pro audio and conference/presentation world which I sometimes encounter, laptops/media players/devices with "consumer" audio outputs are often set to around 75% maximum volume before playback, in the unfortunate circumstances that they have to be connected to a pro grade mixing system. This is normally due to perceived audio quality issues. Pro mixers have a VERY wide range of gain on their inputs.

This is my understanding.

If you use an external USB DAC the volume is locked at 100% which makes sense that this would be 0dB.

Kreshna Aryaguna Nurzaman wrote:

Recently I remember sluggo's post on this thread/

sluggo wrote:

The laptop's output is biased off of (at least) a 3.3V supply, so a clipped line level output is not possible. Besides, the laptop's volume controls are attenuators, so they can't produce a clipped output.

Well it seems most laptop's volume control (including mine) is actually Windows master volume control, so it seems what applies to laptop volume control also applies to Windows master volume control. So of Windows volume control are attenuators, do they yield unity gain (0db) at 100% volume? And if it's not the case, what at volume percentage does it yield unity gain?

It seems it also depends on the sound card/codec being used. How about Realtek? More specifically, Realtek ALC892 used in Gigabyte H77N-WIFI mobo? Where could I find such information?

After Googling a little, I found the manual for Realtek ACL658 instead of ACL892. The Analog Performance Characteristics table says the Master Volume (FRONT/SURR/CENT/LFE) attenuation range has the minimum value of 0dB and the maximum value of -94.5dB. Huh? Isn't that supposed to be the other way around? But fine, let's assume the minimum and maximum value in the manual actually goes the other way around, which means the maximum attenuation range is 0dB. It means 100% volume control is safe, doesn't it? (ie no clipping) So it means the master volume control outputs unity gain (0dB at 100%), am I correct?

But then again, is the maximum gain value the same (0dB) for all Realtek version?

Thanks.

I assume you're talking about an analog-though signal, so I"ll ignore digital sources for now. Yes, in the PC world, the windows volume control is applied to the final stage attenuators in the codec (that is, after the mixer). Following those attenuators is usually a mute/buffer, and then straight to the coupling caps for the line out.

"Minimum" attenuation means no attenuation of the source. A 250mV rms sine wave in yields a 250mv rms sine wave out. So far so good.

However, there are no analog sources in the PC. Everything not named Line In arrives at the codec as bits, which means there's a DAC, which means an amplification stage prior to the mixer. Different codec manufacturers take differing views on how this amp is to be implemented. Some set it up so that a full-range digital signal yields 0dB at the input to the mux. Others, knowing that there's usually 3-6 dB of headroom in the source, turn the amp up to 11 so as to stake a claim on greater dynamic range.

That's why I usually tell folks to set their windows volume control no higher than about 80%. Let the CMOS line stage go gently about it's business, and let your juicy sweet bipolar amp have it's way with your speakers.

Thanks for the replies, many interesting points being made. Anyway, so yes, now I understand that - because of the sound chip's DAC amplification - the 0dB is usually achieved around 80%.

Anyway, is it true that it's undesirable to reduce volume digitally? I read many opinions that say it's better to set Windows master volume at fixed amount, then use the external preamp's master volume control to change loudness, because the former is digital while the latter is analog. It's said that - due to reduced bit depth - reducing volume control in software (like Windows master volume) reduces SQ.

I'm thinking about connecting my would-be HTPC's (Gigabyte H77N-WIFI) multichannel analog output directly to power amplifiers instead of connecting its S/PDIF or HDMI output to an A/V receiver. (I'll be using 4.0 anyway, so I won't need too many power amplifiers.) That way, the signal is purely analog - it won't be compressed by DTS Connect or Dolby Digital Live.

The drawback is I'll have to use Windows master volume control instead of a receiver's master volume control, and of course, the former is software/digital and the latter is hardware/analog.

So which is worse: bit depth reduction resulted from decreasing Windows volume control, or the DTS:C/DDL compression? Does using multichannel analog output outweigh the better SQ of using pure analog connection?

Yes, using the Windows volume control will reduce the effective bit depth of the output signal. But unless you're routinely needing to set it really low (e.g. less than 30%) or have a crappy DAC, you probably won't notice the difference.

just brew it! wrote:

Yes, using the Windows volume control will reduce the effective bit depth of the output signal. But unless you're routinely needing to set it really low (e.g. less than 30%) or have a crappy DAC, you probably won't notice the difference.

Well since it's used as master volume control, there are times when I need to set it really low.

How crappy the DAC of the Gigabyte H77N-WIFI's Realtek audio chip, by the way?

Kreshna Aryaguna Nurzaman wrote:

just brew it! wrote:

Yes, using the Windows volume control will reduce the effective bit depth of the output signal. But unless you're routinely needing to set it really low (e.g. less than 30%) or have a crappy DAC, you probably won't notice the difference.

Well since it's used as master volume control, there are times when I need to set it really low.

How crappy the DAC of the Gigabyte H77N-WIFI's Realtek audio chip, by the way?

The DAC itself is probably fine. But the implementation and analog stages as it exists in a Gigabyte motherboard is going to put it near the bottom of the barrel. You can't pretend that DAC chips exist in a vacuum. As such, the loss in resolution from reducing the Windows master volume will be negligible to the quality reduction due to distortion/crosstalk introduced by the analog stages.

cynan wrote:

The DAC itself is probably fine. But the implementation and analog stages as it exists in a Gigabyte motherboard is going to put it near the bottom of the barrel. You can't pretend that DAC chips exist in a vacuum.

Yes, when I say "DAC" in this sort of context I mean the motherboard vendor's complete implementation, not just the chip itself. IOW, all the stuff you would be replacing if you used the digital out and an external DAC.

Realtek DACs (the chips themselves) are "good enough" on paper for most purposes. But fidelity will be very dependent on how much care the motherboard designer has taken with isolating the post-DAC analog traces, and filtering the power being fed to the DAC to prevent noise from being injected via the power rail. (Poor choice of DC blocking caps can also affect frequency response, but this is unrelated to the S/N issue we are discussing.)

cynan wrote:

As such, the loss in resolution from reducing the Windows master volume will be negligible to the quality reduction due to distortion/crosstalk introduced by the analog stages.

I personally have no idea how good or bad Gigabyte's DAC implementations are these days. But historically, Asus has been pretty good for the past decade or so, with other brands ranging from decent down to unacceptable (suitable only for driving tinny little in-monitor speakers or $10 earbuds). MSI's onboard DAC implementations tended to be particularly grim, IIRC.

Bottom line (as you've indicated) is that you want to keep the signal level well above the analog noise floor. If you're needing to dial the master volume level way down all the time, then you've got too much amplification in the signal path after it leaves the PC (or your headphones are really sensitive). Turn down the volume on your amplified speakers or external amp, and/or get an in-line volume control for your headphones.

Edit: And on the bit resolution issue... because of the way human hearing works, each perceived doubling/halving in volume level corresponds to approximately 10dB. With digital audio, each additional bit of resolution corresponds to 6dB of S/N ratio. So if you have the master volume set such that things sound 1/4 as loud as they are at 100%, you're down about 20dB, and have effectively lost somewhere between 3 and 4 bits of resolution. Your 16-bit DAC is now essentially a 12 or 13 bit DAC!

just brew it! wrote:

Yes, when I say "DAC" in this sort of context I mean the motherboard vendor's complete implementation, not just the chip itself. IOW, all the stuff you would be replacing if you used the digital out and an external DAC.

I was responding more to the OP's question of how good the DAC in the Gigabyte MB potentially is.

just brew it wrote:

Edit: And on the bit resolution issue... because of the way human hearing works, each perceived doubling/halving in volume level corresponds to approximately 10dB. With digital audio, each additional bit of resolution corresponds to 6dB of S/N ratio. So if you have the master volume set such that things sound 1/4 as loud as they are at 100%, you're down about 20dB, and have effectively lost somewhere between 3 and 4 bits of resolution. Your 16-bit DAC is now essentially a 12 or 13 bit DAC!

I suppose this may be a justification for 24 bit digital interfaces and DACs. As far as I know, Windows will support up to 32-bit depth audio. Although, the benefit of going beyond 24 bits is debatable as 32-bit depth seems to be only useful for digital audio post processing (so that you end up with a result that has something at least as good as "pure" 24-bits of dynamic range).

I see. Am I correct to assume a typical A/V receiver's DAC is better than a typical motherboard's DAC? As such, using S/PDIF and let the A/V receiver do the decoding would result in cleaner sound than directly connecting the mobo's multichannel analog into power amps, wouldn't it?

Kreshna Aryaguna Nurzaman wrote:

I see. Am I correct to assume a typical A/V receiver's DAC is better than a typical motherboard's DAC? As such, using S/PDIF and let the A/V receiver do the decoding would result in cleaner sound than directly connecting the mobo's multichannel analog into power amps, wouldn't it?

Yes, and yes.

excession wrote:

Kreshna Aryaguna Nurzaman wrote:

I see. Am I correct to assume a typical A/V receiver's DAC is better than a typical motherboard's DAC? As such, using S/PDIF and let the A/V receiver do the decoding would result in cleaner sound than directly connecting the mobo's multichannel analog into power amps, wouldn't it?

Yes, and yes.

Alright then.

Oh, wait, even after taking into account the audio compression in DDL and DTS:C?

If you use multichannel PCM (for content that is not natively encoded in DD or DTS) then you shouldn't have to worry about DDL or DTS:C.

For example, I have a 6-year old Denon receiver connected to a HD 4850 from 2008 via HDMI. I just select the HDMI audio source from Windows Playback Devices, make sure it is configured to to 5,1 channels, and then my receiver automatically switches back between multichannel PCM and DTS/DD depending on whether the content and media player is encoded and setup to output PCM or DD/DTS.

cynan wrote:

If you use multichannel PCM (for content that is not natively encoded in DD or DTS) then you shouldn't have to worry about DDL or DTS:C.

For example, I have a 6-year old Denon receiver connected to a HD 4850 from 2008 via HDMI. I just select the HDMI audio source from Windows Playback Devices, make sure it is configured to to 5,1 channels, and then my receiver automatically switches back between multichannel PCM and DTS/DD depending on whether the content and media player is encoded and setup to output PCM or DD/DTS.

For games that support DirectSound3D, is HDMI the only thing necessary to enable multichannel PCM?

For example, let say, WarCraft III. The game supports DirectSound 3D, but the game is sufficiently old that I doubt it has specific support for HDMI.
(1) when you use the sound card's multichannel analog output, you will hear 3D sound from the game.
(2) when you use the sound card's SPDIF output, you will hear 3D sound from the game only if your sound card support Dolby Digital Live and DTS: Connect.
(3) when you use HDMI output, will you hear multichannel PCM from the game?

Does a game need to specifically support HDMI sound output (multichannel HD) to yield multichannel PCM sound?

No, HDMI output is up to Windows, not the games.

If an application supports multichannel directsound (as Warcraft 3 appears to: see post on this page) then multichannel PCM, at least over HDMI, should work with any half decent receiver.

S/PDIF does not support uncompressed multichannel PCM due to lack of bandwidth. So go for HDMI if you can.

morphine wrote:

No, HDMI output is up to Windows, not the games.

cynan wrote:

If an application supports multichannel directsound (as Warcraft 3 appears to: see post on this page) then multichannel PCM, at least over HDMI, should work with any half decent receiver.

Flying Fox wrote:

S/PDIF does not support uncompressed multichannel PCM due to lack of bandwidth. So go for HDMI if you can.

Thanks for the replies. So as long as the game supports DirectSound3D or the likes, then it doesn't matter, does it? Whether the multichannel sound information is channeled through multichannel analog, Dolby Digital Live, DTS Connect, or multichannel PCM, is decided by Windows and the sound hardware (i.e., whether the hardware has HDMI output, whether the hardware has S/PDIF output and DDL or DTS:C, and the likes). The game itself only care about sending multichannel sound information to DirectX. As such, a game as old as Hostile Waters: Antaeus Rising can use multichannel PCM for its DirectSound3D multichannel sound, provided it can run on a Windows version (XP, 7, etc) that has the proper driver for the said HDMI hardware (integrated sound chip, discrete sound card, etc).

Am I correct so far?

So here's a question: where does a motherboard's HDMI output belong to? The built-in sound chip (Realtek, VIA, etc), or the integrated graphic?

I'm going to use either ASUS P8Z77-I-DELUXE or Gigabyte H77n WIFI. Both are mini ITX motherboards, both have built-in HDMI out, and both have integrated Intel HD graphics which I'm not going to use. I'm going to use discrete video card instead (a GeForce GTX 280, to be exact). Can I still use the motherboard's built-in HDMI output if I'm using discrete video card?

Kreshna Aryaguna Nurzaman wrote:

So here's a question: where does a motherboard's HDMI output belong to? The built-in sound chip (Realtek, VIA, etc), or the integrated graphic?

In every board I've seen, the integrated graphics.

As far as I can tell, you can still use the HDMI out just for sound - though I haven't tested it personally yet. But it looks to be that way because the HDMI out appears in Windows as just another sound device.

morphine wrote:

Kreshna Aryaguna Nurzaman wrote:

So here's a question: where does a motherboard's HDMI output belong to? The built-in sound chip (Realtek, VIA, etc), or the integrated graphic?

In every board I've seen, the integrated graphics.

As far as I can tell, you can still use the HDMI out just for sound - though I haven't tested it personally yet. But it looks to be that way because the HDMI out appears in Windows as just another sound device.

I see, well I hope so.

The other alternative is to use the video card's HDMI output (it's composite with DVI and you can use DVI-to-HDMI adaptor (which is nothing but a splitter, it seems) for the purpose. However, the motherboard feeds the video card with S/PDIF header, so I'm not sure if I can have digital multichannel PCM that way.

I remember in the XP days I would keep Master Volume & Application volume @ ~80% (2 page downs from full). That would give me the loudest volume while maintaining fidelity. As others have said, not an issue anymore.
Side note, you can still get clipping if you use VLC's >100% volume settings.

Flying Fox wrote:

S/PDIF does not support uncompressed multichannel PCM due to lack of bandwidth. So go for HDMI if you can.

SPDIF via TOSLINK has the bandwidth for it, but most software still thinks SPDIF is still stuck in its ancient days of 3Mbps and just won't let you output multichannel PCM to it.
7.1-channel DTS-HD MA at 24/96 is just 24Mbps.
TOSLINK is now 125Mbps, more than enough for 8 channels at 24/384, or 32/192 (but sadly, a tiny bit short for 8-channel 32/384).

Usacomp2k3 wrote:

I remember in the XP days I would keep Master Volume & Application volume @ ~80% (2 page downs from full). That would give me the loudest volume while maintaining fidelity. As others have said, not an issue anymore.
Side note, you can still get clipping if you use VLC's >100% volume settings.

Well, the problem is actually lowering the volume from Windows master volume control. If I use Windows master volume control as the master volume control for my system, then I have to lower it sometimes - and losing bit rate while doing so.

jihadjoe wrote:

Flying Fox wrote:

S/PDIF does not support uncompressed multichannel PCM due to lack of bandwidth. So go for HDMI if you can.

SPDIF via TOSLINK has the bandwidth for it, but most software still thinks SPDIF is still stuck in its ancient days of 3Mbps and just won't let you output multichannel PCM to it.
7.1-channel DTS-HD MA at 24/96 is just 24Mbps.
TOSLINK is now 125Mbps, more than enough for 8 channels at 24/384, or 32/192 (but sadly, a tiny bit short for 8-channel 32/384).

What modern equipment (players, receivers, motherboards, sound cards, etc.) actually support this latest TOSLINK spec? At least the mainstream industry is pushing "uncompressed PCM = HDMI", partly because of the convenience of the single cable, and partly because of HDCP protection.

So, can we use the mobo's built-in HDMI output connector for sound while using discrete video card?