EVGA Nu Audio

[4est]

3 hours ago, Miska said:

 

My card has been sitting in the box for two months, waiting for me to have time to do something about it...

 

I guess this is as far as you can get at the moment. Although you likely get just as good latency figures with something like RME ADI-2.

 

 

Why would it need to be same application? You can stream Tidal/Qobuz through Roon+HQPlayer or something else. And Netflix through Netflix application. For TV I would assume HDMI 2, either through normal copper or one of the new optical extenders (for projector).

 

For similar use case I've been using newest NUC's with HDMI 2.0 connected to 4K TV and then external DAC. Video playback goes through whatever (from files through VLC) and audio playback through HQPlayer. For audio-only playback latency doesn't matter and for video use cases it works fine. If you want, you can also put audio with video through HDMI and then you don't really have sync problems in practice. And YouTube won't give you DSD audio anyway, so that's not an issue.

 

Challenge is if you want YouTube and DSD upsampling together. That can be done with HQPlayer for example, but then you need something like VLC to deal with latency synchronization between the two.

 

Yes, thank you. You are correct of course about the multiple programs, I worded it poorly.

[rickca]

On 9/20/2019 at 5:27 AM, Miska said:

Does WASAPI Exclusive work correctly

Yes.  When Qobuz is set to WASAPI Exclusive, videos with audio on sites like CNN don't play and you get an error message saying 'something went wrong during native playback'.  Note that my Nu Audio card is set as the default audio device because it's the only audio device in my system.  'Allow applications to take exclusive control of this device' is enabled in the Advanced tab of Speaker Properties.

 

However, when I set Qobuz to use ASIO, audio from CNN gets played together with the Qobuz track.  So, what does this mean?  Clearly, ASIO is not getting exclusive control.  So does this mean that ASIO is going through the Windows mixer and I'm not getting bit perfect output unless the source material matches the sample rate and bit depth set in Speaker Properties?  If I am stating this correctly, please confirm and I will relay this message to EVGA support.  If this is true, then I have to conclude that the EVGA ASIO driver isn't working properly.

 

In Device Manager, the Nu Audio shows up as Device USB\VID_3842&PID_3100\00000030.

 

 

[Miska]

1 hour ago, rickca said:

However, when I set Qobuz to use ASIO, audio from CNN gets played together with the Qobuz track.  So, what does this mean?  Clearly, ASIO is not getting exclusive control.  So does this mean that ASIO is going through the Windows mixer and I'm not getting bit perfect output unless the source material matches the sample rate and bit depth set in Speaker Properties?  If I am stating this correctly, please confirm and I will relay this message to EVGA support.  If this is true, then I have to conclude that the EVGA ASIO driver isn't working properly.

 

Sounds like WASAPI Exclusive is working correctly, but ASIO is having unexpected behavior. ASIO should essentially work the same as WASAPI Exclusive, except that you get things like native DSD support which WASAPI doesn't have.

 

So I'd say there is something going wrong with the ASIO driver.

 

Have you tried what happens if you play CNN or something while playing DSD natively through the ASIO driver? That could be used as a suitable case for pointing out wrong behavior.

 

[Miska]

OK, I finally got this card installed and measured. And I have to say this is quite awesome device and even more so as internal sound card at very reasonable price.

 

I installed the card first on a Windows desktop for short while to update it's firmware. Then I placed it in my powerful HQPlayer Embedded server with i7-8086K CPU and Seasonic Platinum 660W PSU. More info about the server here:

 

Measurements were almost boring to do, audio band performance is practically identical with both 8x PCM rates (352.8/384k) and DSD256 (also 48k DSD seems to work correctly). Audio band performance seems to be mostly defined by the opamps, by swapping the opamps to other models, it could be possible to improve measurement results further. For most DSD measurements, ASDM5EC modulator was used, but seems like the results would have been a little bit better with ASDM7EC (based on last tests I did). But I didn't start redoing anyway.

 

For the measurements I even used 5 meter long XLR microphone cable because I didn't want to move things. Didn't seem to have adverse effects, which is good! Note that during the measurements the server was running HQPlayer Embedded, poly-sinc-ext2 filter and EC modulators for DSD output, so CPU and RAM under high load on the same machine!

 

Unfortunately I couldn't test the card at 44.1k PCM rate, it just goes nuts / crashes and needs cold reset! (for me this is anyway non-issue)

 

So here's 1k THD test:

Most THD is second and third harmonic, the harmonic series is quite short which is good from audibility perspective.

 

Here's 1 kHz at -60 dBFS level:

This is clean, the small blip at around 66 kHz at -130 dB level is possibly from the PSU. Also note that there's no 8 kHz USB packet noise visible, although the card uses USB internally (PCIexpress USB adapter + UAC2 on board).

 

IMD test:

Looks normal, likely mostly defined by the opamps (AD8056 + OP275).

 

Then one of the unfortunately nowadays forgotten measurement I have started to do again. TIM distortion test:

For interpretation; the four high spikes visible here are the test signal, everything else is extra that shouldn't be there. This is in line with the other results. Some spikes are a little bit lower with DSD, this is one where probably ASDM7EC would have been even better. This is function of device's wide band performance.

 

Then ASR-style multi-tone test:

 

And here's Jtest24:

This is perfectly fine result!

 

 

If we then look at wider band that hints towards the TIM result differences too, 0 - 22.05 kHz sweep from 44.1/32 source file, at 5 MHz bandwidth.

352.8/32 PCM input:

DSD256 input:

With PCM you can see first image around 352.8 kHz sampling rate, at DSD256 this is absent since HQPlayer runs proper filters to x256 rate. So here you have a notable difference.

 

If we then zoom in to 500 kHz bandwidth we can see the same better.

352.8/32 PCM input:

DSD256 input:

There you can see the difference between PCM and DSD inputs, image around 352.8k PCM sampling rate vs no image with DSD (all depending on digital filters used to create the DSD).

 

[lucretius]

10 hours ago, Miska said:

Measurements were almost boring to do, audio band performance is practically identical with both 8x PCM rates (352.8/384k) and DSD256 (also 48k DSD seems to work correctly).

 

By "48k  DSD", I take it you mean DSD1024? Thanks.

[Miska]

4 hours ago, lucretius said:

By "48k  DSD", I take it you mean DSD1024? Thanks.

 

No I mean DSD at multiples for 48 kHz base rate. IOW, 3.072/6.144/12.288 MHz in addition to the usual 44.1 kHz base rates of 2.8224/5.6448/11.2896 MHz.

 

Unfortunately most DACs on the market today don't support DSD at multiples of 48 kHz base rate, although it is only matter of switching the clock same way as is already done for PCM. So they have gone through some extra effort to not support it!

 

[Miska]

Since the first image is down about -85 dB, it makes maximum reconstruction accuracy for PCM 44.1k family equivalent of about 14-bit resolution. This is quite good result for PCM mode. (practically same as Pro-Ject PreBox S2 Digital)

 

[the_bat]

I understand from the designer that the measured performance (nonlinearity and noise wise) can be improved  by swapping out the AD8056 for another op-amp, such as an OPA1612.

 

The AD8056 was chosen on the basis of its sound quality, after trying several options.  It’s interesting that it can achieve respectable electrical performance with relatively low open loop gain (71dB vs 130dB plus), hinting at good native linearity.

[Miska]

3 hours ago, the_bat said:

I understand from the designer that the measured performance (nonlinearity and noise wise) can be improved  by swapping out the AD8056 for another op-amp, such as an OPA1612.

 

The AD8056 was chosen on the basis of its sound quality, after trying several options.  It’s interesting that it can achieve respectable electrical performance with relatively low open loop gain (71dB vs 130dB plus), hinting at good native linearity.

 

Good that they left easy option for swapping the opamps and documented it. Many motherboard manufacturers do the same for their better motherboards too. If one wants to get (even) better measurement results.

 

AD8056 is also high speed, maintaining fairly low distortion even at high frequencies, which is important for the first analog filter stages where there's quite a bit of high frequency noise present.

 

At some point I may try with something, I have a lot of different opamps from my own projects, so no need to even buy anything for that. Just too busy to do it in near future.

 

[EVGA_Lee]

On 9/21/2019 at 10:33 AM, rickca said:

Yes.  When Qobuz is set to WASAPI Exclusive, videos with audio on sites like CNN don't play and you get an error message saying 'something went wrong during native playback'.  Note that my Nu Audio card is set as the default audio device because it's the only audio device in my system.  'Allow applications to take exclusive control of this device' is enabled in the Advanced tab of Speaker Properties.

 

However, when I set Qobuz to use ASIO, audio from CNN gets played together with the Qobuz track.  So, what does this mean?  Clearly, ASIO is not getting exclusive control.  So does this mean that ASIO is going through the Windows mixer and I'm not getting bit perfect output unless the source material matches the sample rate and bit depth set in Speaker Properties?  If I am stating this correctly, please confirm and I will relay this message to EVGA support.  If this is true, then I have to conclude that the EVGA ASIO driver isn't working properly.

 

In Device Manager, the Nu Audio shows up as Device USB\VID_3842&PID_3100\00000030.

 

 

Hi rickca,

 

Small world.  I was directed here by our friends at Audio Note who saw Miska's post on the measurements.  However, I also recognize the issue you're reporting, since I was the PM our CS checked with.  There's a lot more context and information here in this thread about your issue, so I made an account here to let you know that I'll follow up with our Audio team to check into your situation and see what we can do about it.

[rickca]

1 hour ago, EVGA_Lee said:

I made an account here to let you know that I'll follow up with our Audio team to check into your situation and see what we can do about it.

Thank you so much for joining the AS community and following up on this issue.

[rickca]

On 9/26/2019 at 3:23 PM, EVGA_Lee said:

I made an account here to let you know that I'll follow up with our Audio team to check into your situation and see what we can do about it.

Any progress?

[EVGA_Lee]

15 hours ago, rickca said:

Any progress?

 

Hi rickca,

 

I sent you a PM.  To make a long story short, ASIO appears to work fine in other applications, such as foobar2000 or jriver.  More testing with Qobuz is needed at this point.

[Rugby]

It does not work right with Roon playing DSD.  Have contacted evga tech support on the issue for months.  Once DSD is played, Roon will not release the device. 

[4est]

On 10/10/2019 at 11:19 PM, Rugby said:

It does not work right with Roon playing DSD.  Have contacted evga tech support on the issue for months.  Once DSD is played, Roon will not release the device. 

I don't know if my issues were the same as yours, but this card sounded great- for one night! I was really excited at how good this thing sounded using Roon>HQPlayer 11.2MHz DSD. On the second day the it wouldn't make a peep, and hasn't since despite my efforts. Sadly I think I will have to return it. I simply cannot seem to get the drivers to install after uninstalling them as per instructions. I'll gladly try again if anyone has some suggestions. It seems to fail after preinstall is sufficient. the more window info seems to indicate failure at creating desktop icon or something similar. EVGA_Lee?

[rickca]

1 hour ago, 4est said:

I'll gladly try again if anyone has some suggestions.

Did you reboot after uninstalling?

 

Are you sure the driver isn't installed?  Even if it didn't create an icon, the driver might be installed.  If running Windows 10, check add/remove programs and see if EVGA Nu Audio Driver 4.66.0 is in the list.  If it is, make sure nothing is muted in the EVGA GUI and that you have the intended playback device (speakers/headphones) selected.

 

I don't use Roon so I can't help you there.

 

[4est]

2 hours ago, rickca said:

Did you reboot after uninstalling?

 

Are you sure the driver isn't installed?  Even if it didn't create an icon, the driver might be installed.  If running Windows 10, check add/remove programs and see if EVGA Nu Audio Driver 4.66.0 is in the list.  If it is, make sure nothing is muted in the EVGA GUI and that you have the intended playback device (speakers/headphones) selected.

 

I don't use Roon so I can't help you there.

 

Thanks for your reply. Yes, I rebooted. There are some files for the driver, but the device still isn't recognized and is not available.

[EVGA_Lee]

On 10/19/2019 at 2:34 PM, 4est said:

Thanks for your reply. Yes, I rebooted. There are some files for the driver, but the device still isn't recognized and is not available.

If you don't mind, please send me a PM with what's going on.  I don't mind helping you out here, but I'd rather not try to take the thread too much further off-topic.  If the OP doesn't mind, however, we can work on it here.

[Miska]

On 10/19/2019 at 9:01 PM, 4est said:

On the second day the it wouldn't make a peep, and hasn't since despite my efforts.

 

When the card crashed on me, I had to do a cold reboot (power off). And since there are quite a lot of capacitors, short power down wasn't enough...

 

Note you should really power down the computer completely from power switch in the back or pulling the plug. Soft power off from the front panel power button may not be enough. (PSU still provides sleep power)

 

I just want to emphasize about this if someone else encounters such situation, to avoid unnecessary extra hassle. No need to do anything else.

 

[4est]

2 hours ago, Miska said:

 

When the card crashed on me, I had to do a cold reboot (power off). And since there are quite a lot of capacitors, short power down wasn't enough...

 

Note you should really power down the computer completely from power switch in the back or pulling the plug. Soft power off from the front panel power button may not be enough. (PSU still provides sleep power)

 

I just want to emphasize about this if someone else encounters such situation, to avoid unnecessary extra hassle. No need to do anything else.

 

Thank you Miska, I have done that repeatedly including removing the card between software (un)installs, but you are right to mention that.

[4est]

5 hours ago, EVGA_Lee said:

If you don't mind, please send me a PM with what's going on.  I don't mind helping you out here, but I'd rather not try to take the thread too much further off-topic.  If the OP doesn't mind, however, we can work on it here.

PM sent, good point and we can post the solution if there is one.

[rickca]

7 hours ago, EVGA_Lee said:

If the OP doesn't mind, however, we can work on it here.

I have no objection to you and @4est working on it here on this thread.

[rickca]

@4est did you ever get this resolved?

[4est]

18 hours ago, rickca said:

@4est did you ever get this resolved?

Sadly no. I sent it back. I really wish it had worked out, but I was only willing to put so much effort into my AV/background system. I got frustrated in part because my ASRock MB seems to have a faulty RAM slots that would cause havoc when inserting the card. I am past due for an upgrade! Thanks for asking Rick.

[Miska]

I'm quite happy with the card on Linux, with a few caveats. But for firmware upgrades it needs to be placed in a Windows machine...