[SOLVED] DAC Marketing and IMD (Also, Amps are too powerful rant)

[Vamp898]

Wait, so your OP is complaining about specs of the DAC chip used on a DAC: and your issues with hearing hiss is the volume of your source (your PC)? This doesn't seem to be an issue with pots when it comes to different analog stages, or chip specs with an external DAC.

Sorry if i caused confusion so to clear it up a bit.

Due to lowering the Volume on the PC, i was not testing at -80db, i was testing at an _much_ lower volume. I was giving the DAC a Tiny input signal and then was listening to the DAC at -80db. As this is nonsense, i tested again properly with a 0db input signal into the DAC and the issue was gone.

I hope this makes it clear and sorry again for the confusion.

It's very possible on the PC side that there was some added clipping/noise with whatever processing that could have occurred. PCs are notorious for not having apps that are "exclusive" and would be influenced by whatever audio settings you set for Windows (or their own EQ stages).

What's "Live BluRay"? Do you mean watching blu-ray concerts on PC? That also depends on your app, its settings, and if it has an exclusive mode for Windows API.

I am using Linux. Every Modern Linux System uses a Sound System called Pipewire, so all applications are Bit-Perfect, always. There is nothing like WASAPI or different APIs or something like that. There is only 1 System called Pipewire (that replaced everything that existed earlier, quite a few years ago actually) and that one is always bit-perfect (Unless you tell it to not be for whatever reason one might wish).

So when i listen to a 96kHz file in Firefox, it will be played bit-perfect to the DAC. Everything is bit-perfect, always, no matter which application.

So when watching a 96kHz Live BluRay, it doesn't matter which application i use, all of them will send bit-perfect 96kHz to the DAC (What the DAC detects correctly and shows on the Display).

But for the sake of reference, i am watching all my Bluray with mpv https://mpv.io/ because that one has the highest image quality. As for sound, like all other applications that play sound on Linux, its bit-perfect.

It has the best GPU Decoder Engine, the best scaling algorithms (Most BluRay i own are FullHD, it helps a lot watching them on 4K, even though the GPU is mostly at 70~90% load ) and it supports real HDR on all displays (no matter if its an HDR Display or not).

@gregorio

Where did you get that background? It’s a particularly bad idea as amps will typically have best performance at around 60%, a good amp will have a wide range of optimum performance, say 20% - 80%, few will even be good at 100%, let alone best. At a low gain setting though it might be dependant on the topology, so again, how did you determine best performance at 100%?

1. Official Measurements from Topping
2. Independent Measurements from ASR
3. Measurements i did myself

It has the best THD+N and SNR at 100% Low Gain, so i target that. I usually do not follow "Rule of thumbs", 60% Volume means something different for every amp. some might perform perfect there, others not.

My CS43131 Dongle (DAWN PRO) even performs best at 100% Volume High Gain. I check the measurements (or make ones myself) to see in which range the gear performs best and then use it in that range, no matter what that might be.

Or better said, i try to use it in that range (I do not want to get deaf, so i rather accept the loss in sound quality to protect my ear).

Then it’s the wrong amp for the task.

I was looking for an Amp that can drive my R70xa without issues, has ~0 Ω output impedance, flat frequency response at lowest volume and <0.5µV Nosie. This Amp does have all that. Even with the SE846, there is no audible noise (even at higher volumes), its perfectly flat at all volumes and it has <0.1Ω output resistance.

The only downside is the analog Volume control, but i could not find another Amp that have all the features i want with less power. Most Desktop Amps i looked into had even more power (carzy right? Who needs all that Power?).

I owned an TA-ZH1ES before which i had to use at -99.5db, so its already an improvement

That’s mad, why on earth would you deliberately reduce the resolution/dynamic range of your DAC by a factor of 10,000? That’s got to be close to a record for the worst gain-staging in history, even for a consumer, but I thought you said you were an engineer or did some audio engineering?

I really hope I’ve misunderstood something and this isn’t actually your “real world use”?

There is no other way to reach that low Volume. I tried several DACs and Amps and nothing i found was able to do that without going to such low volumes.

There is no other solution to this problem, hence i use the solution i have. But unrelated to that, as i shown with measurements, its still inaudible even at higher volumes, and it is especially inaudible when listening to music at ~5db.

 

[Davesrose]

I am using Linux. Every Modern Linux System uses a Sound System called Pipewire, so all applications are Bit-Perfect, always. There is nothing like WASAPI or different APIs or something like that. There is only 1 System called Pipewire (that replaced everything that existed earlier, quite a few years ago actually) and that one is always bit-perfect (Unless you tell it to not be for whatever reason one might wish).

This doesn't make sense to me. If your Linux OS driver setup was "direct" "bit-perfect" (IE doing no processing), then there shouldn't be any difference in sound if you set the OS volume to 0 or 100. Yet you say your issue was solved by setting it to 100%.

Being curious, I've googled about Linux and direct audio. Google AI (as reliable as it may or may not be) it seems you may need to cofigure the OS for direct audio:

To set audio to "direct" or bypass processing on Linux, you need to configure PulseAudio to use the desired output device directly, bypassing any virtual sinks or other processing. This can be done through PulseAudio's graphical settings or by using the command line.

Method 1: Using PulseAudio Volume Control (Pavucontrol)

1. 1. Install Pavucontrol:
   If you don't have it, install it using your distribution's package manager (e.g., sudo apt install pavucontrol on Debian/Ubuntu).
   
   
2. 2. Open Pavucontrol:
   Launch the application.
   
   
3. 3. Select Output Device:
   In the "Playback" tab, choose the specific output device (e.g., your headphones, speakers) that you want to use for direct audio.
   
   
4. 4. Configure "Direct Output" (if available):
   Some configurations might have an option to force direct output or bypass processing. Look for settings related to "simultaneous output," "virtual sink," or similar options. If you want to output to multiple devices at the same time, you might need to enable a virtual output.
   
   
5. 5. Test:
   Play some audio and verify that it is playing directly through the selected device.
   

Method 2: Using Command Line (Pactl)

1. 1. List Sinks:
   Use the pactl list sinks command to see a list of available output devices (sinks).
2. 2. Identify Target Sink:
   Note the name of the device you want to set as the default.
3. 3. Set Default Sink:
   Use pactl set-default-sink &lt;Your_Device_Name&gt; to set the device as the default.
4. 4. Example:
   If your desired device is named "alsa_output.pci-0000_00_00.2.hdmi-stereo," the command would be pactl set-default-sink alsa_output.pci-0000_00_00.2.hdmi-stereo.
   

Important Notes:

• PulseAudio and ALSA:
  PulseAudio is the sound server, while ALSA provides the underlying hardware drivers for soundcards.
  
  
• Virtual Sinks:
  Virtual sinks are often used for routing audio to applications or other devices. Setting the default output to a virtual sink bypasses direct output to your physical device.
  
  
• Startup Applications:
  You can add the pactl set-default-sink command to your startup applications to ensure the default output device is set on every login.
  
  
• Different Distributions:
  The specific steps might vary slightly depending on your Linux distribution (e.g., Ubuntu, Fedora, Manjaro).

 

[redrol]

ASIO exists on windows and beyond that, who cares. Bit perfect means nothing anyhow.

 

[Vamp898]

This doesn't make sense to me. If your Linux OS driver setup was "direct" "bit-perfect" (IE doing no processing), then there shouldn't be any difference in sound if you set the OS volume to 0 or 100. Yet you say your issue was solved by setting it to 100%.

You still have volume control. So let me further explain: It is bit-perfect unless you use the volume control. Then it is bit-perfect except for the input signal size.

It would be horrendous to not be able to control the volume as 99% of the people who use Linux, change the Volume on the Desktop, so yes, there is volume control. You could disable it, but i use it too as the DAC is not the only thing i access.

Being curious, I've googled about Linux and direct audio. Google AI (as reliable as it may or may not be) it seems you may need to cofigure the OS for direct audio:

That information is for the PulseAudio Sound System, this Sound System no longer exists. That was used before PipeWire (<=2017)

But if people would still use it, these instructions would not work either. But i think that is because Gemini did not understand what direct audio means or just simply doesn't know how it works on Linux.

The ALSA System in the Linux Kernel outputs the Audio, i think that is what is called a dirver on Windows? As it is limited in features a modern userspace Linux System has there are also so called Sound Servers are sitting on top of ALSA which provides all the features you might want like fancy resampling with different quality settings, dither settings, it even has a graphing api. So that might be the APIs in Windows? Maybe? Not sure. Windows is so complex and every new Version introduces new tools that exist alongside old ones for dedaces but stop working anyway so why have them at all... ANYWAY

DAW like Ardour https://ardour.org/ utilize that graphing feature but you can use it with every application that exists, so you can connect every stream to everything For example my Mic is detected as stereo, but it is mono (i only use the left channel), so i connect the left output channel to the L/R input channel of, for example, Teams for Meetings. Fancy stuff like that. Or some YouTube Videos mess up audio where you only hear it from the left channel or have channel impalance, i then cross-connect the L/R Channel to Mono so i hear it centered.

PipeWire can downmix/upmix audio, again, depending on what you tell it to do. The default settings when you setup nothing, depending on your Distribution, do downmix everything to 44.1/48k (to save resources as nobody needs more than 48k). You can control the quality of the downmix, but you can also tell it to not downmix at all or only downmix for certain resolutions and so on. I set my Pipewire to not do any resempling, its one single setting, its as simple as that.

But there is no system like on Windows where you have different drivers with different features and APIs and depending what the application uses, it gets resampled or not and stuff like that.

There is one single driver, ALSA (since 1998), that supports bit-perfect audio. There is one Sound Server on top of it called Pipewire and all applications (no matter what API they might use to access Sound on Linux) use Pipewire (Pipewire replaced everything that existed before including Userspace ALSA, PulseAudio, JACK and so on) and that one does what you tell it to do.

So there is only one driver and one "api" on Linux, and these support bit-perfect audio _and_ still support volume control (Hard to believe but yes, bit-perfect audio with volume control exists since 1998, just not in Windows. Maybe this breaks bit-perfectness depending on how you see it)

 

[Davesrose]

You still have volume control. So let me further explain: It is bit-perfect unless you use the volume control. Then it is bit-perfect except for the input signal size.

It would be horrendous to not be able to control the volume as 99% of the people who use Linux, change the Volume on the Desktop, so yes, there is volume control. You could disable it, but i use it too as the DAC is not the only thing i access.

But then it's not "direct" as far as having a full audio output in which there's no processing on the Linux side. If you move the volume and it changes the audio, then by definition, it's not direct. Honestly driver issues is a big reason why I don't like Linux. I now have more computers which have secure setups in their BIOS that prevents a lot of Linux distros.....but in the past I've used Mint and Ubuntu. First time I installed Mint as a dual boot from Windows for work, it was with a new Dell laptop-and the audio drivers always sucked for my new high end laptop. The audio was terrible unless I raised the volume to 150%, and then it ruined my laptop speakers (I literally had to have them replaced). I then did audio through a BT speaker.

But there is no system like on Windows where you have different drivers with different features and APIs and depending what the application uses, it gets resampled or not and stuff like that.

There is one single driver, ALSA (since 1998), that supports bit-perfect audio. There is one Sound Server on top of it called Pipewire and all applications (no matter what API they might use to access Sound on Linux) use Pipewire (Pipewire replaced everything that existed before including Userspace ALSA, PulseAudio, JACK and so on) and that one does what you tell it to do.

So there is only one driver and one "api" on Linux, and these support bit-perfect audio _and_ still support volume control (Hard to believe but yes, bit-perfect audio with volume control exists since 1998, just not in Windows. Maybe this breaks bit-perfectness depending on how you see it)

But you're still saying that changing OS volume does something with your audio, so your signal is not "direct" or "bit perfect". To change the digital output to have a different volume: that means it was processed to attenuate the whole audio level.

 

[Vamp898]

But then it's not "direct" as far as having a full audio output in which there's no processing on the Linux side. If you move the volume and it changes the audio, then by definition, it's not direct. Honestly driver issues is a big reason why I don't like Linux. I now have more computers which have secure setups in their BIOS that prevents a lot of Linux distros.....but in the past I've used Mint and Ubuntu. First time I installed Mint as a dual boot from Windows for work, it was with a new Dell laptop-and the audio drivers always sucked for my new high end laptop. The audio was terrible unless I raised the volume to 150%, and then it ruined my laptop speakers (I literally had to have them replaced). I then did audio through a BT speaker.

But you're still saying that changing OS volume does something with your audio, so you're signal is not "direct" or "bit perfect". To change the digital output to have a different volume: that means it was processed to attenuate the whole audio level.

I would not say it is processing the signal, that sounds like creating and changing it. It is creating the signal with a different maximum amplitude. You're shifting 0db and then create the signal with that. It's basically the same thing the DAC does when changing the volume, but I get what you mean.

You obviously loose SNR and increasing distortion in the DAC by doing so. At 0db it is bit-perfect however (What i use if i don't accidentally change it)

When i do TRPG sessions, i often change the volume auf my Audio Player to match the volume of the player voices and i can't count how many times i forgot to change it back wondering, i have to crank up the volume at the amp its somehow a common theme for me.

So it was not Linux fault, not the DACs fault, it simply was my fault for feeding a tiny input signal into the DAC^^

However i get much better IMD values than the measurements from Topping and/or ASR. They both say at -60db Volume you are already at an IMD of >-70db.

This is the result of my D50III set at -60db Volume

I would say this is <=-85db and not >-70db. Just to make sure i did nothing wrong, this is how i set the DAC and obviously, you can barely see the IMD at all as it is masked by the SNR of my DI which has to fight hard at that low volume.



Not sure why my results are so much better. If anyone wants to analyze themself, here is the output file
https://nextcloud.ikaros.space/s/mrSFz2HZCDWXEYn

I also tried setting the DAC to 0db and give it an -60db Input Signal, but the result remains (mostly) unchanged

So i am not sure how exactly they got that value and/or what they did to trigger that, but i can not reproduce that with my DI

 

[Davesrose]

I would not say it is processing the signal, that sounds like creating and changing it. It is creating the signal with a different maximum amplitude. You're shifting 0db and then create the signal with that. It's basically the same thing the DAC does when changing the volume, but I get what you mean.

Well I understand your setup might be "bit perfect" when it comes to no other EQ settings being applied when you output to a digital USB. But it's not "direct" as far as no processing occurring at all with the digital output (it means there would be no change in volume with what level you set in OS). Volume attenuation itself can have a great effect. I've never really relied on a computer as my sound source (now it's either a HDD for a DAC or my phone running Apple Music), but I've always found issues with trying to setup HTPCs giving me "direct" bitstream audio for movies.

 

[Vamp898]

Well I understand your setup might be "bit perfect" when it comes to no other EQ settings being applied when you output to a digital USB. But it's not "direct" as far as no processing occurring at all with the digital output (it means there would be no change in volume with what level you set in OS). Volume attenuation itself can have a great effect. I've never really relied on a computer as my sound source (now it's either a HDD for a DAC or my phone running Apple Music), but I've always found issues with trying to setup HTPCs giving me "direct" bitstream audio for movies.

I never had issues so far except for being to stupid to setting the volume back at 0db.

There is a tool called pw-top (PipeWire Table of Processes) that shows you every sound stream, at which rate it is running, delay, if any transmissions errors occurred and so on.

Because i am paranoid i tend to not trust things and rather measure myself to see the actual result but everything so far was fine, until the IMD mistake that, again, was on my said.

In my personal experience, i can trust my PC a lot more than most audio devices. I've had DACs that randomly changed their roll-off filter when re-plugging them, i am not kidding. Back then i thought something must be wrong with my PC or my setup until i found out, others have this issue too and it occurs with every setup.

I can not remember that i ever found any audio issue at all and it was my PC. When setting it up for the first time 3 years ago or so, i directly plugged my DAC into the USB and noticed a lot of noise. That was the only time but i could fix this with an DSD Isolator. Since then, it operated as it should.

Usually i notice issues before the DAC by just checking the output of pw-top or just the Interface of mpv itself. Here you can see the Audio Stream Information inside mpv (As you can see, it outputs directly to pipewire and nothign else and its s32 96kHz



As it is outputting to Pipewire (like all other applications too) i can see it in pw-top and see if there is any resampling going on (For example i can see that VirtualBox has opened an Stream but is not sending/receiving any data)

And i can see that in qpwgraph too

If there would be anything going on that would change the signal (EQ, resample, whatever) it would showup in either pw-top or qpwgraph. For example when i enable an EQ, i can see the EQ sitting in between mpv and the D50 III and could manually disconnect it or re-connect or whatever.

 

[gregorio]

Would that not depend on the way the digital volume control is implemented in the DAC? A separate digital volume control chip after the DAC chip, employing a switched resistor ladder network wouldn't reduce resolution surely, albeit that you still lose a little bit S/N ratio due to one or two extra semiconductor junctions?

Are you talking about an analogue volume control? You can’t have a “digital volume control after the DAC chip” because after the DAC chip you only have an analogue signal. Clearly I must be misunderstanding something.

I owned an TA-ZH1ES before which i had to use at -99.5db, so its already an improvement
There is no other way to reach that low Volume.

That still makes no sense to me. Reducing the digital level by 15dB or so, even 20dB, should have little/marginal effect but 80dB, that would destroy the SNR. If a reasonable digital level requires the amp to be set at 1% - 3%, clearly it is the wrong amp for the job.

I would not say it is processing the signal, that sounds like creating and changing it. It is creating the signal with a different maximum amplitude.

You’re talking about a digital output, in which case the maximum amplitude of the signal being created never changes. What changes is the data that signal represents. If you lower the digital level, that is processing, the digital data must change and therefore it cannot be bit perfect.

I should maybe bow out because what’s being discussed makes no sense at all, so I must be misunderstanding the discussion.

G

 

[Vamp898]

That still makes no sense to me. Reducing the digital level by 15dB or so, even 20dB, should have little/marginal effect but 80dB, that would destroy the SNR. If a reasonable digital level requires the amp to be set at 1% - 3%, clearly it is the wrong amp for the job.

You’re talking about a digital output, in which case the maximum amplitude of the signal being created never changes. What changes is the data that signal represents. If you lower the digital level, that is processing, the digital data must change and therefore it cannot be bit perfect.

I should maybe bow out because what’s being discussed makes no sense at all, so I must be misunderstanding the discussion.

G

As said, its an 9 Ω 114 db/mw IEM. If i want to listen at 10db, i need 0.0000005986 V (0.00000000003 mW), i do not know any Amp that can do that, something like that doesn't exist, pretty sure. The magnetic fields my refrigerator emits are strong enough to drive them when i am in an 30cm range...

I did quite a lot of research and i tried an CS43131 Dongle before (without any op-amps, just the built-in Class H), so an 50$ Dongle on 3.5mm (<=1 Vrms at around 10 Ohm or below) and even that one i had to use at -90db Low Gain.

If you're listening at an SPL of 10db, it doesn't matter if you crush the SNR, even an SNR of 15db would be already inaudible.

 

[gregorio]

If you're listening at an SPL of 10db, it doesn't matter if you crush the SNR, even an SNR of 15db would be already inaudible.

True. Assuming you’re an adult, at 10dBSPL there is no audible signal or noise and therefore SNR is indeed irrelevant. But then if you’re effectively listening to silence, I still don’t understand this thread or what difference a digital level on a DAC could make.

G

 

[2leftears]

Are you talking about an analogue volume control? You can’t have a “digital volume control after the DAC chip” because after the DAC chip you only have an analogue signal. Clearly I must be misunderstanding something.

Ok, my bad wording. I meant digitally-controlled analog volume control, which colloquially I think most would call a digital volume control especially if they are controlled with two volume up/down buttons. They have an analog input, analog output, and something like a digitally multiplexed resistor ladder network to affect the attenuation. Or something like the Texas Instruments PGA2311P (datasheet here), or a simple DIY demo circuit example like the one below, for which I'm sure digitally-controlled integrated volume control chips exist also:

I think (again, colloquially) that most people would call these digital volume controls even though they control the volume of the analog signal. They do not have the drawback of losing resolution like bit-shifting or integer rescaling in the digital domain would.

 

[Vamp898]

True. Assuming you’re an adult, at 10dBSPL there is no audible signal or noise and therefore SNR is indeed irrelevant. But then if you’re effectively listening to silence, I still don’t understand this thread or what difference a digital level on a DAC could make.

G

I am 35 now and according to the last hearing test i took which is roughly 1.5 Months ago, my hearing threshold is, depending on frequency (of course) between -5db and 5db.

As long my hearing did not heavily degrade in those 1.5 Months, i should be able to hear BGM at around 10~20db (for which i need -80~-70db on the DAC).

About the digital level on a DAC

According to this measurement, the lower the dBFS, the higher the IMD. But i can not verify that with my D50III.

Maybe i am misunderstanding this measurement, but if not, my D50III performs significantly better than what is shown in this graph

 

[gregorio]

I meant digitally-controlled analog volume control, which colloquially I think most would call a digital volume control especially if they are controlled with two volume up/down buttons.

I thought that must be what you meant but vamp seemed to be talking specifically about the digital level in his DAC rather than the analogue volume on his amp, because he gave his amp volume in percent apparently after his DAC in dB.

I am 35 now and according to the last hearing test i took which is roughly 1.5 Months ago, my hearing threshold is, depending on frequency (of course) between -5db and 5db.

No, that is not your hearing threshold! Hearing tests do not measure dB SPL (sound pressure level) thresholds, they measure dB HL (hearing level), which is an entirely different weighted scale designed to measure hearing levels relative to normal hearing, not relative to the threshold of hearing! 0dB HL is equal to normal hearing (technically a range of -10dBHL to 15dBHL), not 0dB SPL which is beyond normal hearing for an adult. 25dB HL is usually the threshold for clinical hearing loss (clinically diagnosed deafness), although different standards/calibrations exist in different regions.

No wonder I’m confused, if you’re using unrelated dB scales all over the place!

G

 

[bigshot]

Why do people use equipment with pathological impedance? Just go buy something that works.