Neutron HiFi DAC V1 Review

[OHtaru]

ASIO driver is not provided but you can use FlexASIO (https://github.com/dechamps/FlexASIO) configured to use Exclusive WASAPI. WASAPI backend is provided by PortAudio (https://github.com/PortAudio/portaudio/) and it is currently very stable and allows low-latency operation.

Could you clarify why you need ASIO for USB DAC - low latency, bit-prefect output, specific formats, software supports only ASIO?

Thanks for your reply. 感谢你的回复。
I want ASIO for low latency, because I use MIDI keyboard to play virtual instruments in Studio One which supports ASIO for low latency only. And Exclusive WASAPI doesn't let other applications play audio, so it's not a perfect solution. Is there any plan for ASIO support?

 

[dmitrykos]

Speculating a little... I suspect Apple reason(s) are the same as the Mozilla Foundation: it is potentially opening a backdoor into the web browser, as the system doesn't really know what kind of USB device is connected and whether it could be malicious...

There is no need to expose USB to web API (on Android NConfigurator is accessing HID via OS API, not web API), Apple could just allow the use of IOKit framework for apps. Entitlement for the app is required anyway, so Apple normally reviews app on case by case basis and does not allow questionable apps. Android device's security doesn't suffer from allowed USB access because user will have to allow USB device access by the app in GUI explicitly anyway, so it doesn't happen silently under the hood and everybody happy To my view, large corporation is too slow in its decisions and these decisions are influenced primarily by its top management's vision. Android OS historically was initially very open, I do not know whether it was the main idea (to make it popular) but it really worked, later Google severely limited it but it still more or less OK.

It's such a shame that I can't find my perfect dac/amp right now.
Something with hardware PEQ, bluetooth and enough power to drive HD800 with EQ effortlessly. Am I asking that much?

Having PEQ and, in general, DSP support, such as one present in DAC V1, implies usage of some general purpose MCU as USB controller and writing own firmware to handle all the stuff - USB interface, DSP, configurability and etc. It also drags writing configuration software which would control DAC and configure PEQ/DSP. All this work is high cost in terms of complexity, time and money because software support is spread over the long time, long after the actual device model release (unless producer decides to release software once and drop it completely in order to cut costs). For example, DAC V1's code base related to firmware and NConfigurator is huge and I did not expect, initially, that it would grow to such size and complexity over the time.

Not all producers risk to choose this way either from the competence point of view (firmware/software development is a hard task) so from the business feasibility aspect, therefore majority is using some ready to use controller with firmware developed by IC vendor - you modify device descriptor string and voila. Therefore you won't find many portable USB DACs with onboard DSP, with configuration software and which are being updated on a regular basis. As a result the selection is limited - one device has this or that feature but missing another...

 

[dmitrykos]

Thanks for your reply. 感谢你的回复。
I want ASIO for low latency, because I use MIDI keyboard to play virtual instruments in Studio One which supports ASIO for low latency only. And Exclusive WASAPI doesn't let other applications play audio, so it's not a perfect solution. Is there any plan for ASIO support?

I see, what lowest latency is tolerable for you in milliseconds? I believe FlexASIO must be capable of around 20 ms with WASAPI Shared.

 

[d1m0n]

Hello @dmitrykos,

I have one last suggestion as future improvement if it is feasible at all of course.
Is it possible to implement a "loudness" function as a DSP?
The idea behind it would be to take into account the effect of the Fletcher-Munson curves.

Possible scenario: The user defines at the NConfigurator DSP tab his normal listening volume level + the quietest volume level + a desired max bass and treble elevation gain for the lowest level.
This gain then varies automatically between max on the quietest level and below and none on the normal and above.
It would be even more perfect if one could combine that with the PEQ instead of FRC.

Thank you very much
BR

 

[CedarX]

Hello @dmitrykos,

I have one last suggestion as future improvement if it is feasible at all of course.
Is it possible to implement a "loudness" function as a DSP?
The idea behind it would be to take into account the effect of the Fletcher-Munson curves.

Possible scenario: The user defines at the NConfigurator DSP tab his normal listening volume level + the quietest volume level + a desired max bass and treble elevation gain for the lowest level.
This gain then varies automatically between max on the quietest level and below and none on the normal and above.
It would be even more perfect if one could combine that with the PEQ instead of FRC.

Thank you very much
BR

Would a selectable signal/volume-based zoned (3 zones?) compressor/expander be even better?

 

[jkim]

@dmitrykos,

I would appreciate your answer to my question asked here about the distortion performance of ES9219-based DACs. I am very curious.
You may want to post an answer in this thread if you like to. Your call.

 

[dmitrykos]

@jkim replying to your question in CS thread:

On the Distortion of Cirrus Logic CS431xx-Based Devices: A Comparative Review

"This DAC contained in our product has a huge flaw that is in the hardware itself and we cannot fix it, sorry guys" Well, depending on the market's consumer protection laws, that could very well warrant a refund - for consumers, and even the OEM if it's Cirrus at fault.

www.audiosciencereview.com

What would be the primary design factor in harmonic distortion performance of ES9219-based DACs? Your Neutron HiFi v1 has the lowest harmonic distortions I've seen among the DACs adopting ES9219.

DAC V1's harmonic distortions in the review could be much lower if I could revise DAC V1's THD Compensation values earlier than provided sample to @amirm for the review I was relying on the engineer's work with THD but while revising THD Compensation in NConfigurator started playing with ESS's THD Compensation feature more in-depth and found out that DAC V1 can perform much better in this respect, see achieved results in my post:

Neutron HiFi DAC V1 Review

@CedarX I do not know yet. My intention is to test and to check whether it is capable of sufficiently long Convolver. Taking into account that V1's performance allows 40-band 2-channel PEQ the hope becomes more realistic. There is similar task in Neutron Music Player :)

www.audiosciencereview.com

There are numerous factors affecting linearity (less linear -> more harmonic distortions): clean and sufficient power supply (obviously - ultra low noise), correct PCB layout of the circuit surrounding DAC IC (input and output), ultra low phase noise precise clock (SPXO), DAC must operate in a asynchronous Master mode, all components interacting with DAC must be of high quality and etc. Basically, the more ideal is the PCB design (layout, components) surrounding DAC IC, the lower the harmonic distortions will be which can be compensated additionally by THD Compensation feature of ESS DAC chip.

From a measurement point of view, this is a huge difference.
EDIT. By the way I recently measured both revisions of the Qudelix 5K with ES9218P and ES9219C. They measured the same and matched Amir's results.

Qudelix-5K device is not advertising ES9219Q performance, for sure But, that is a flaw of either hardware design (as mentioned above) or software/firmware (which can be fixed). 5K has 2 ES9219Q chips and synchronization of 2 channels becomes very tricky as you have to synchronize clock and digital inputs, additionally you have to make symmetric layout so that signals flow in parallel without excessive time skew. It could be one of the reasons of so low SNR which is far below the chip's performance capabilities but ESS lab has reference design with 2x ES9219Q, so getting good SNR/DNR with such design is technically possible.

I have Qudelix-5K based on ES9218P from Neutron Player's development. I played 48 kHz Sweep tone and recorded with Audacity in order to see how sweep tone looks like:



Spectrogram shows strong harmonic components found in the device's initial review, also frequencies start attenuating after 16 kHz. The waveform of the sweep wave tells that Qudelix developers are either trying to use their custom Oversampling filter (OSF) or it is the effect of the hw layout flaw which is causing such DSP-like effect (attenuation after 16 kHz) if 5K still uses default OSF filter. Highlighted area shows unattenuated frequencies.

For comparison, I did the same measurements with DAC V1 to show the reference output of ESS onboard OSF - Fast Linear (default, attenuates frequencies after 22 kHz at 48 kHz SR):


and Neutron's custom NEUTRON Slow Linear (attenuates frequencies after 20 kHz at 48 kHz SR):


If Qudelix designed their own OSF then those strong harmonics can be as a result of overloading during oversampling stage when 2-nd stage of OSF filter does not compensate gain of the 1-st stage and it can cause these strong harmonics (I was getting such issues during my experiments with custom OSF). If it is custom OSF then 5K's sonic performance can be improved by using built-in Fast Linear filter + THD Compensation and it can be done by a simple firmware update boosting performance of all existing devices.

 

[jkim]

Thanks for your reply.

I dug up some measurements of ES921x-based devices. I suspect that this drastic difference in harmonic products is due to something unique to the implementation of ES9218/ES9219 in a DAC design, which can be done either correctly or sub-optimally. Most likely not due to general design choices associated with other DAC designs.

See the following FFT results of the Shanling UP4 w/ ES9218P.

Except for the high noise floor, which was due to Amir's ASIO driver issue he couldn't resolve at that time, the harmonic distortion products are very similar to those of the Qudelix 5K:

Also, the Hiby FC4 with ES9219C shows essentially the same high harmonics:

In contrast, the following implementations resolved the issue.

Yours first:

THX Onyx with ES9218P:

Fiio BTR7 with ES9219C. Except for somewhat high 3rd product, this looks fine, too:

 

[dmitrykos]

I suspect that this drastic difference in harmonic products is due to something unique to the implementation of ES9218/ES9219 in a DAC design, which can be done either correctly or sub-optimally.

Yes and there are many ways to get it screwed, either on hardware or firmware level. DAC V1's hw revision is 4.1, so there were quite many iterations until target goal was reached. ES9219 needs correct routing (collaboration with ESS engineer is a good idea), separate ultra-low noise power supplies (3 in total), dual ultra-low phase noise XO for best performance in Master mode (with own power supply), careful analysis by the engineer regarding anomalies on DAC pins, final fine-tuning of produced PCBA with THD Compensation (I am very curious to see re-measured performance of DAC V1 with latest firmware with ASR's AP equipment by the way) - and it shall do a trick.

 

[Robbo99999]

Yes and there are many ways to get it screwed, either on hardware or firmware level. DAC V1's hw revision is 4.1, so there were quite many iterations until target goal was reached. ES9219 needs correct routing (collaboration with ESS engineer is a good idea), separate ultra-low noise power supplies (3 in total), dual ultra-low phase noise XO for best performance in Master mode (with own power supply), careful analysis by the engineer regarding anomalies on DAC pins, final fine-tuning of produced PCBA with THD Compensation (I am very curious to see re-measured performance of DAC V1 with latest firmware with ASR's AP equipment by the way) - and it shall do a trick.

Has Amir said he's gonna remeasure it with latest firmware?

 

[dmitrykos]

Has Amir said he's gonna remeasure it with latest firmware?

We agreed that my measurements sufficiently demonstrate the achievement with new changes in THD Compensation in the latest firmware:

Neutron HiFi DAC V1 Review

@CedarX I do not know yet. My intention is to test and to check whether it is capable of sufficiently long Convolver. Taking into account that V1's performance allows 40-band 2-channel PEQ the hope becomes more realistic. There is similar task in Neutron Music Player :)

www.audiosciencereview.com

Though, it would be great to check ES9219 chip's performance (SNR, DNR, and jitter) tested with AP's resolution, I would say for academic purpose.

 

[howard416]

Sorry if this question has been asked already... does the use of PEQ add any latency (more than 1-2 ms)?

 

[dmitrykos]

Sorry if this question has been asked already... does the use of PEQ add any latency (more than 1-2 ms)?

Good question The latency (algorithmic delay) is 0 ms for any number of bands.

 

[howard416]

Whooaa. Awesome, thanks.

 

[Bernard23]

Seems a little underpowered to me. I have a Hidizs S9 Pro Plus (Martha) which outputs at 32 ohms 138 Mw with the single ended 3.5 and 183 Mw with 4.4 balanced. It only costs $90 on Ali. Doesn't have the PEQ, but I get that from USB Audio Player Pro (costs around $15) on my Android phone.

So if you're willing to roll your own a little bit, you can, I think, do better than the Neutron for cheaper.

Similar experience for me, I also own a, Fiio KA17 to use with my Samsung S23 that doesn't have a headphone jack. However, the big limitation of UAPP is that it only works for streaming, it cannot play files saved to your device. The KA17 has a PEQ but it's such a tortuous method to set it up it's almost pointless.
I've not read all 18 pages of this review, but how easy is it to change the PEQ settings? Is it effected via an app?

 

[dmitrykos]

I've not read all 18 pages of this review, but how easy is it to change the PEQ settings? Is it effected via an app?

Yes, via NConfigurator app.

You can check its screenshots in DAC V1 Details -> scroll to NConfigurator's section:

Neutron HiFi DAC V1: Specs, Features And NConfigurator Downloads

Explore the Neutron HiFi™ DAC V1 — a portable, high-resolution USB DAC and preamp featuring ESS ES9219Q, 32-bit/384 kHz PCM, DSD256 support, and integrated DSP for studio-quality, audiophile-grade audio on the go.

neutronhifi.com

There is also a web version of NConfigurator where you can click Demo and play with menus, PEQ's settings are in DSP section (Demo mode is read-only, so changing number of bands has no effect but you will have an idea anyway):

NConfigurator

nconf.neutronhifi.com

If you have Android device then there is NConfigurator, you can get it via Google Play or Huawei App Gallery or Neutron's website and there is also a Demo button to try it:

NConfigurator - Apps on Google Play

Configuration utility for Neutron HiFi USB DAC audio device

play.google.com

 

[Plz virtual Surround 7.1]

Hey, i wanted to ask if it were possible to add ASIO for v1 dac?

Thank you for your Hardwork as always!

 

[dmitrykos]

@Plz virtual Surround 7.1 to my view FlexASIO with Exclusive WASAPI backend will do the trick - support for ASIO-only apps + low latency. Exclusive WASAPI gives 1-3 ms latency which is quite ok. If you use Neutron Music Player for Win32 then you can get direct USB access with native DSD.

 

[OHtaru]

@dmitrykos I have received my DAC today. Nice sound and DSP feature! But I found the DSP for output and AGC for microphone can't be enabled simultaneously. I tried to enable both of them and saved it as profile but after reconnecting the power or applying the profile, one of them will be turned off. Is that normal or is it a bug?

 

[dmitrykos]

@dmitrykos I have received my DAC today. Nice sound and DSP feature! But I found the DSP for output and AGC for microphone can't be enabled simultaneously. I tried to enable both of them and saved it as profile but after reconnecting the power or applying the profile, one of them will be turned off. Is that normal or is it a bug?

Would you please PM me with details which effects you tried to activate and also save profile to a file and attach it to the message. It can be a bug of fw or NConfigurator applying the preset or limitation where playback DSP effect does not allow additional DSP processing (on mic side).