Neutron HiFi DAC V1 Review

[neutroncode]

By the way, the NConfigurator can‘t set THD-compensation state when applying profiles, it ignores this parameter according to the console log. I've reported the details on the Neutron forum.

This bug is now fixed in NConfigurator 1.8.7:

https://forum.neutroncode.com/viewtopic.php?f=10&t=7436

To make profile recognizable - apply it first, then when device is restarted, select preset again and press Save button. With this you will update preset record with latest format changes.

Also released Firmware 62 which in cooperation with NConfigurator 1.8.7 will no longer reset device when THD optimizing preset is changed. Many thanks to @jkim for noticing that Balance is disabled in case of Voltage Mode (No Load or Custom THD preset), 62 supports balance in any output mode now.

 

[Hifihedgehog]

Yes---I measured THD vs output of the new compensation settings (will post later when I get a chance).
Here is how to interpret each THD compensation setting:

• Low Impedance (< 50ohm) & Low Output (< 0.5V) Optimized for reasonably sensitive (low-impedance, too, in most cases) headphones & IEMs fed with < 0.5V signal most of the time.
• Low Impedance (< 50ohm) & High Output (>= 0.5V) Optimized for demanding situations in which low-impedance phones are driven hard (e.g., boosted bass, somewhat insensitive headphones), requiring high-level (> 0.5V) signals often times.
• High Impedance (>= 50ohm) Optimized for high-impedance (> 50 ohm) headphones.
• No Load (DAC/Preamp, >= 1kOhm) Optimized for connection to an external device such as a headphone amp or active speaker system.

For the IEM you have, the Low Impedance & Low Output mode will work best. With that IEM you will NOT ever need a higher level than 0.5V even with some EQ boost in bass.

Do you happen to have the results handy?

 

[neutroncode]

Firmware 64 + NConfigurator 1.8.8 are out!

Firmware 64 release notes:

https://forum.neutroncode.com/viewtopic.php?f=10&t=7453

NConfigurator 1.8.8 release notes:

https://forum.neutroncode.com/viewtopic.php?f=11&t=7454

Key new features:

Adaptive Loudness Compensation (ALC) DSP:
This update introduces a new useful DSP effect - Adaptive Loudness Compensation (ALC). Of course, if you are in DSP world and using EQ for sound correction. You can google what this effect does, but briefly, it is a kind of 2-band EQ which adjusts gain of its Low/High shelving filters based on device volume (the volume you set in the OS for the device). DAC V1's implementation offers Adaptive mode (auto adjustment) and Manual (fixed gain, does not depend on device volume). The idea of the effect is to compensate the loudness of low and high band frequencies at lower device volume (human hearing perceives loudness of various frequencies differently, see ISO 226:2003 or latest ISO 226:2023 for the derived curve).

The idea to add ALC DSP to DAC V1 as well as kind help with parametrization, optimization and debugging of the effect behavior was provided by @jkim! @jkim thank you very much for that, now we all have a new DSP effect in DAC V1

Luckily, MCU of DAC V1 is powerful enough to handle ALC + Crossfeed at max 384 kHz PCM, thus combination of these both DSPs becomes super useful for listening with a headset. Of course, there are other DSP combinations with ALC DSP.

Digital Headroom:
This new option adds possibility to select 2 volume modes when DSP is active:

1. None - default mode, it adjusts DAC chip's analog volume control first up to -26 dB, then adjusts DAC's chip digital volume control further. As a result this mode gives best SNR due to analog volume attenuation first. Due to absence of any headroom in digital domain before DSP is applied (PEQ) the use of Preamp for EQ/FRC is required of course if your music is generally normalized (most commercial music). NConfigurator 1.8.8 adds Preamp Mode → Auto to make it easier for you, just activate Auto and do not bother what changes you made to EQ bands.
2. Optimal -18 dB mode. This mode was proposed by @jkim as a way to simplify PEQ management. The volume is first adjusted in digital domain before DSP is applied, then goes analog volume on DAC chip after -18 dB is reached, then finally DAC chip's digital volume. If you are listening to the music at lower volume than DSP is amplifying then you do not need to use Preamp as it is compensated by the digital volume applied before DSP is processed.

Preamp Mode → Auto:
This is a convenience feature of NConfigurator 1.8.8 which can be found in DSP tab → EQ or FRC DSP. If Auto is activated then NConfigurator will calculate the frequency response of the PEQ and adjust the Preamp precisely. It is a guarantee of the overflow/overload absence. If you are using default Digital Headroom → None mode then make sure to activate Auto for a hassle-free PEQ management.

Voltage mode:
The THD presets behavior is changed. Now they are all, except Custom (Current mode), forcing Voltage mode which allows [-26, 0] dB analog volume control on DAC chip (see Digital Headroom description). Many thanks to @jkim for deriving new THD coefficients for these presets which improve SNR for various loads! Due to changes in THD presets, go to Advanced tab and set preset to None and then to your favorite mode.

If DSP is not active Voltage mode is still beneficial for a sensitive headset because it allows to lower analog volume first down to -26 dB. As a result we lower noise floor of internal amp too, thus making sensitive headset or other connected equipment happy (less noise - "darker" silence).

 

[OHtaru]

Firmware 64 + NConfigurator 1.8.8 are out!

Firmware 64 release notes:

https://forum.neutroncode.com/viewtopic.php?f=10&t=7453

NConfigurator 1.8.8 release notes:

https://forum.neutroncode.com/viewtopic.php?f=11&t=7454

Key new features:

Adaptive Loudness Compensation (ALC) DSP:
This update introduces a new useful DSP effect - Adaptive Loudness Compensation (ALC). Of course, if you are in DSP world and using EQ for sound correction. You can google what this effect does, but briefly, it is a kind of 2-band EQ which adjusts gain of its Low/High shelving filters based on device volume (the volume you set in the OS for the device). DAC V1's implementation offers Adaptive mode (auto adjustment) and Manual (fixed gain, does not depend on device volume). The idea of the effect is to compensate the loudness of low and high band frequencies at lower device volume (human hearing perceives loudness of various frequencies differently, see ISO 226:2003 or latest ISO 226:2023 for the derived curve).

The idea to add ALC DSP to DAC V1 as well as kind help with parametrization, optimization and debugging of the effect behavior was provided by @jkim! @jkim thank you very much for that, now we all have a new DSP effect in DAC V1

Luckily, MCU of DAC V1 is powerful enough to handle ALC + Crossfeed at max 384 kHz PCM, thus combination of these both DSPs becomes super useful for listening with a headset. Of course, there are other DSP combinations with ALC DSP.

Digital Headroom:
This new option adds possibility to select 2 volume modes when DSP is active:

1. None - default mode, it adjusts DAC chip's analog volume control first up to -26 dB, then adjusts DAC's chip digital volume control further. As a result this mode gives best SNR due to analog volume attenuation first. Due to absence of any headroom in digital domain before DSP is applied (PEQ) the use of Preamp for EQ/FRC is required of course if your music is generally normalized (most commercial music). NConfigurator 1.8.8 adds Preamp Mode → Auto to make it easier for you, just activate Auto and do not bother what changes you made to EQ bands.
2. Optimal -18 dB mode. This mode was proposed by @jkim as a way to simplify PEQ management. The volume is first adjusted in digital domain before DSP is applied, then goes analog volume on DAC chip after -18 dB is reached, then finally DAC chip's digital volume. If you are listening to the music at lower volume than DSP is amplifying then you do not need to use Preamp as it is compensated by the digital volume applied before DSP is processed.

Preamp Mode → Auto:
This is a convenience feature of NConfigurator 1.8.8 which can be found in DSP tab → EQ or FRC DSP. If Auto is activated then NConfigurator will calculate the frequency response of the PEQ and adjust the Preamp precisely. It is a guarantee of the overflow/overload absence. If you are using default Digital Headroom → None mode then make sure to activate Auto for a hassle-free PEQ management.

Voltage mode:
The THD presets behavior is changed. Now they are all, except Custom (Current mode), forcing Voltage mode which allows [-26, 0] dB analog volume control on DAC chip (see Digital Headroom description). Many thanks to @jkim for deriving new THD coefficients for these presets which improve SNR for various loads! Due to changes in THD presets, go to Advanced tab and set preset to None and then to your favorite mode.

If DSP is not active Voltage mode is still beneficial for a sensitive headset because it allows to lower analog volume first down to -26 dB. As a result we lower noise floor of internal amp too, thus making sensitive headset or other connected equipment happy (less noise - "darker" silence).

Genius idea! It looks like DRE but not realtime, which prevents from distortion like "cirrus hump". I drew the graph according to your description, notice DAC takes over the digital volume control when analog volume reaches -26dB:
Edit: Confirmed the minimum analog volume is -24dB, not -26dB

BTW, why not always use pre-DSP volume control? If so, the digital headroom is unlimited as long as the volume is low enough.

 

[OHtaru]

Firmware 64 + NConfigurator 1.8.8 are out!

Firmware 64 release notes:

https://forum.neutroncode.com/viewtopic.php?f=10&t=7453

NConfigurator 1.8.8 release notes:

https://forum.neutroncode.com/viewtopic.php?f=11&t=7454

Key new features:

Adaptive Loudness Compensation (ALC) DSP:
This update introduces a new useful DSP effect - Adaptive Loudness Compensation (ALC). Of course, if you are in DSP world and using EQ for sound correction. You can google what this effect does, but briefly, it is a kind of 2-band EQ which adjusts gain of its Low/High shelving filters based on device volume (the volume you set in the OS for the device). DAC V1's implementation offers Adaptive mode (auto adjustment) and Manual (fixed gain, does not depend on device volume). The idea of the effect is to compensate the loudness of low and high band frequencies at lower device volume (human hearing perceives loudness of various frequencies differently, see ISO 226:2003 or latest ISO 226:2023 for the derived curve).

The idea to add ALC DSP to DAC V1 as well as kind help with parametrization, optimization and debugging of the effect behavior was provided by @jkim! @jkim thank you very much for that, now we all have a new DSP effect in DAC V1

Luckily, MCU of DAC V1 is powerful enough to handle ALC + Crossfeed at max 384 kHz PCM, thus combination of these both DSPs becomes super useful for listening with a headset. Of course, there are other DSP combinations with ALC DSP.

Digital Headroom:
This new option adds possibility to select 2 volume modes when DSP is active:

1. None - default mode, it adjusts DAC chip's analog volume control first up to -26 dB, then adjusts DAC's chip digital volume control further. As a result this mode gives best SNR due to analog volume attenuation first. Due to absence of any headroom in digital domain before DSP is applied (PEQ) the use of Preamp for EQ/FRC is required of course if your music is generally normalized (most commercial music). NConfigurator 1.8.8 adds Preamp Mode → Auto to make it easier for you, just activate Auto and do not bother what changes you made to EQ bands.
2. Optimal -18 dB mode. This mode was proposed by @jkim as a way to simplify PEQ management. The volume is first adjusted in digital domain before DSP is applied, then goes analog volume on DAC chip after -18 dB is reached, then finally DAC chip's digital volume. If you are listening to the music at lower volume than DSP is amplifying then you do not need to use Preamp as it is compensated by the digital volume applied before DSP is processed.

Preamp Mode → Auto:
This is a convenience feature of NConfigurator 1.8.8 which can be found in DSP tab → EQ or FRC DSP. If Auto is activated then NConfigurator will calculate the frequency response of the PEQ and adjust the Preamp precisely. It is a guarantee of the overflow/overload absence. If you are using default Digital Headroom → None mode then make sure to activate Auto for a hassle-free PEQ management.

Voltage mode:
The THD presets behavior is changed. Now they are all, except Custom (Current mode), forcing Voltage mode which allows [-26, 0] dB analog volume control on DAC chip (see Digital Headroom description). Many thanks to @jkim for deriving new THD coefficients for these presets which improve SNR for various loads! Due to changes in THD presets, go to Advanced tab and set preset to None and then to your favorite mode.

If DSP is not active Voltage mode is still beneficial for a sensitive headset because it allows to lower analog volume first down to -26 dB. As a result we lower noise floor of internal amp too, thus making sensitive headset or other connected equipment happy (less noise - "darker" silence).

The co-work of volume control and pre-amp is screwed up.

I found that my DSP pre-amp was set to -12dB, and the part of the volume control above 12dB was ineffective. If the pre-amp is set to -30dB, then more than half of the volume bar becomes useless. The volume control seems to be incorrectly set to start taking effect only when it is lower than the pre-amp value.

Moreover, the pre-amp slider also doesn't work properly when the volume control is below 0dB. At low volumes, more than half of the pre-amp slider is ineffective. After many tests, I concluded that:

The volume control works normally only when the pre-amp is at 0dB, and the pre-amp slider works normally only when the volume is at 100%.
The final volume depends on the lower one between the pre-amp and the volume control. The higher one is always ineffective.

 

[neutroncode]

The volume control works normally only when the pre-amp is at 0dB, and the pre-amp slider works normally only when the volume is at 100%.
The final volume depends on the lower one between the pre-amp and the volume control. The higher one is always ineffective.

It is a new behavior of the volume in relation to Preamp. For example, let's set EQ 1kHz band to for example +10 dB then we effectively request amplification on +10 dB. The user volume control will start decreasing volume of the device below -10 dB level (when this peak is passed). This condition makes user volume control loose for the amount of Preamp applied. Maybe it feels too experimental and confusing based on your comments.

 

[OHtaru]

It is a new behavior of the volume in relation to Preamp. For example, let's set EQ 1kHz band to for example +10 dB then we effectively request amplification on +10 dB. The user volume control will start decreasing volume of the device below -10 dB level (when this peak is passed). This condition makes user volume control loose for the amount of Preamp applied. Maybe it feels too experimental and confusing based on your comments.

This is really confusing. And this behavior has caused me trouble: the volume of all music has increased. This forces me to use a small part of the lower end of the volume bar, making it very awkward to adjust the volume.

To give an example:
-Previously, I had a pre-amp of -12dB. I usually used 20% volume, and at that time, the DAC showed -24dB, resulting in a total attenuation of -12 -24 = -36dB, which made the volume just right.
-But with this new behavior, when I use 20% volume, there is only -24dB of attenuation. The volume is 12dB louder than before, making it unbearable for me to listen to.

Additionally, this behavior is a fool-proofing, which is also inconsistent with the intuition of the signal chain. The new firmware already has a new feature called auto pre-amp to prevent clipping, so what is the necessity of this approach?

Moreover, I don't think the two volume modes is a good idea. I suggest merging the two to optimal mode, while discarding the digital volume of the DAC chip. As long as the volume control is pre-DSP, there can be as much headroom as needed with the volume decreasing (see the picture below). Why create two modes to confuse users even more?

My suggestions for the signal chain:
volume control (digital volume) ----> pre-amp ----> DSP ----> DAC chip (analog volume)

This achieves headroom while maintaining good compatibility. Users' previous DSP solutions can still be used, the volume will not change compared to before, and it will not cause confusion for users.

I hope you can adopt my suggestions. If you have any questions, PM me and feel free to discuss.

 

[stefanovesa]

@neutroncode Hi, can you explain the difference between 24 bit and 32 bit mode in the OS driver?
Thanks

 

[jkim]

Moreover, I don't think the two volume modes is a good idea. I suggest merging the two to optimal mode, while discarding the digital volume of the DAC chip. As long as the volume control is pre-DSP, there can be as much headroom as needed with the volume decreasing (see the picture below). Why create two modes to confuse users even more?

This is exactly what Qudelix does in its 2V mode, a simple yet effective way of digital headroom management. It also gives a user freedom to use Preamp in EQ if he likes to.

 

[neutroncode]

@neutroncode Hi, can you explain the difference between 24 bit and 32 bit mode in the OS driver?

Could you please clarify what mode do you mean (sample bits selection?)? DAC V1 does not have specialized driver but all modern OSes support USB Audio 2.0.

 

[neutroncode]

This is exactly what Qudelix does in its 2V mode, a simple yet effective way of digital headroom management. It also gives a user freedom to use Preamp in EQ if he likes to.

In Qudelix app, do you mean Headroom option? It gives me 3 variations: 0 dB, -6 dB, -12 dB and works like a fixed-level Preamp in my tests, so I get volume decreased on -12 dB if I select -12 dB option. That is the same effect as if Preamp in NConfigurator is set to -12 dB.

My suggestions for the signal chain:
volume control (digital volume) ----> pre-amp ----> DSP ----> DAC chip (analog volume)

Will thoroughly consider again. Thank you for brainstorming and proposal!

 

[stefanovesa]

Could you please clarify what mode do you mean (sample bits selection?)? DAC V1 does not have specialized driver but all modern OSes support USB Audio 2.0.

Correct, I meant for example in MacOS Midi setup. What changes if I choose a 24 bit mode or a 32 bit mode? I can see on the DAC display that the bit depth changes accordingly. I'm asking because some DACs have 32 bit only mode, DAC V1 has both.

 

[neutroncode]

What changes if I choose a 24 bit mode or a 32 bit mode?

Your software can send data stream with a higher precision that leads to a less quantization noise and apparently 32-bit data bears 8 more bits of valid audio data in comparison to 24. If 24-bit stream is available then software can send 24-bit PCM without doing any conversion to 32-bits (although it is quite simple in case of Int -> Int), if source is 24-bit FLAC, for example. In case of 32-bits, if software is doing some DSP and it is done with double precision then 32-bit would deliver more valid data to the DAC chip.

 

[d1m0n]

Super nice! ALC is basically what I suggested a while back too. Thank you very much for implementing it.

@neutroncode I have some questions for better understanding how and when to use this new functions properly.

1. Do I need care about digital preamp adjustment in any way when using ALC w/o EQ as DSP or is its handled by implementation?
2. In ALC tab: What is the "Reference Gain" referring to. Do I need to set my aprox. normal listening level where the correction is 0 and the corrections kicks in when I go lower with the volume. Or is it the lowest listening level where I want the effect be at max. If so what is the effect range where there is no correction anymore?
3. Does Digital Headroom Optimum -18 setting has any effect if DSP is set to "off"?
4. When using EQ. I don't fully get it when it better to use Digital Headroom "none" + "Auto" or manual preamp value OR just set Optimum - 18bd?
5. Do I understand the Optimum -18db function correct. According the description above I don't need a preamp setting with EQ as long as my listening volume level is below the EQ amplification, which would be basically the necessary preamp value if I do it the "classic" way, right? If I for example would normally need -9db preamp for my EQ I would be fine with jut Optimum -18db active as long as I don't go with my volume louder than -9bd level according to the V1 display?

Sorry for the many questions. Hope your answer will be of any help for others too.

 

[OHtaru]

Will thoroughly consider again. Thank you for brainstorming and proposal!

It's more about careful consideration than brainstorming. I think a principle for firmware updates is to think twice before action. Users don't want to be caught off guard by updating to a "beta version". Or, release it as a genuine beta version before it's mature. There's a "Development version" option in NConfigurator, isn't there?

Users are happy to see the release of new features, but the process should be smooth enough and compatible with previous versions. In other words, the DSP configuration I use daily shouldn't have issues like increased volume or changed behavior of the pre-amp gain slider after a firmware update.

The Neutron HiFi DAC V1 is a great platform. It deserves powerful features and a well-considered signal flow, but I think it takes time.

 

[neutroncode]

Super nice! ALC is basically what I suggested a while back too. Thank you very much for implementing it.

Sorry, missed your proposal within active discussion probably.

2. In ALC tab: What is the "Reference Gain" referring to. Do I need to set my aprox. normal listening level where the correction is 0 and the corrections kicks in when I go lower with the volume. Or is it the lowest listening level where I want the effect be at max. If so what is the effect range where there is no correction anymore?

Technically, it is the volume level of DAC V1 at which ALC kicks in. Above this level effect does not do anything. Assume that at 0 dB of DAC V1's volume level corresponds to 80 dB SPL Mastering Level. Reference Gain applies correction to the level at which Mastering Level is perceived, for example by setting Reference Gain to -6 dB will cause ALC DSP to start modifying sound when you reach DAC V1's volume level of -7 dB and lower.

3. Does Digital Headroom Optimum -18 setting has any effect if DSP is set to "off"?
4. When using EQ. I don't fully get it when it better to use Digital Headroom "none" + "Auto" or manual preamp value OR just set Optimum - 18bd?
5. Do I understand the Optimum -18db function correct. According the description above I don't need a preamp setting with EQ as long as my listening volume level is below the EQ amplification, which would be basically the necessary preamp value if I do it the "classic" way, right? If I for example would normally need -9db preamp for my EQ I would be fine with jut Optimum -18db active as long as I

It is being fully revised to remove any further confusion. For now just use None and use Preamp Mode -> Auto with your EQ config.

Users are happy to see the release of new features, but the process should be smooth enough and compatible with previous versions. In other words, the DSP configuration I use daily shouldn't have issues like increased volume or changed behavior of the pre-amp gain slider after a firmware update.

Issues happen during the development, but I understand your concerns. Development version option in NConfigurator is not for betas, it is to provide OTA version to the user during the development process. There is no need to poke the developer, just kindly ask for a previous fw version to rollback to the desired behavior if you can't wait for issue resolution.

 

[d1m0n]

Technically, it is the volume level of DAC V1 at which ALC kicks in. Above this level effect does not do anything. Assume that at 0 dB of DAC V1's volume level corresponds to 80 dB SPL Mastering Level. Reference Gain applies correction to the level at which Mastering Level is perceived, for example by setting Reference Gain to -6 dB will cause ALC DSP to start modifying sound when you reach DAC V1's volume level of -7 dB and lower.


It is being fully revised to remove any further confusion. For now just use None and use Preamp Mode -> Auto with your EQ config.

Thx a lot for clarification. Will do that.

 

[neutroncode]

Firmware 65 is out and fixes discussed issue with Preamp and user volume. Many thanks to @OHtaru for testing the pre-release candidate.

Release notes:

https://forum.neutroncode.com/viewtopic.php?f=10&t=7457

 

[neutroncode]

NConfigurator 1.8.9 is released.

Release notes:

https://forum.neutroncode.com/viewtopic.php?f=11&t=7458

It removes Digital Headroom from the DSP tab in favor of standard volume behavior when Analog gain fires first resulting in lowering the noise floor of the internal amp too. It adds more DSP combinations with a new ALC DSP.

For experts, Digital Headroom is still supported by firmware and NConfigurator's console command, as well as by Profile:

NConfigurator --dsp_headroom -18
or
NConfigurator --dsp_headroom ANY_HEADROOM_VALUE

Missed in previous posts, here is ALC DSP in action (-4 dbFS on a plot corresponds to 0 dB volume of DAC V1, volume is decreased in 4 dB steps up to -24 dB) - low and high frequencies are getting amplified when device volume is decreased: