T+a dac 8 dsd

[Solstice380]

@louisxiawei  I have a Lampi Golden Atlantic with Amanero USB using the Win Combo384 ASIO driver on a NUC NAA.   I have thought occasionally that I had heard the static pops but because they are very low level and random I wasn’t sure.  I also listen to music that generally does not have long quiet passages.  

 

I listened to the Harry Potter track 5 times and heard the pops.   Totally random, and low level, but they are there. 

 

Is it the Amanero driver?

[blue2]

On ‎5‎/‎28‎/‎2018 at 12:44 PM, bibo01 said:

Update - I have spoken to Amanero today. Unfortunately, last week he was in London for an important job, so no work was done, but this week he's aiming to release 2004BE firmware for Linux. He is confident that distortion and channel swapping are solved. Apparently, he is going to use one DMA channel only. Previosly he was using two channels and they were going out of sync because on-board cpu was nor able to keep up. After release there will be a round of tests on github. Fingers crossed again.

 

18 minutes ago, Solstice380 said:

@louisxiawei  I have a Lampi Golden Atlantic with Amanero USB using the Win Combo384 ASIO driver on a NUC NAA.   I have thought occasionally that I had heard the static pops but because they are very low level and random I wasn’t sure.  I also listen to music that generally does not have long quiet passages.  

 

I listened to the Harry Potter track 5 times and heard the pops.   Totally random, and low level, but they are there. 

 

Is it the Amanero driver?

This report changes perspective on the problem. If the 'pops' are random then it sounds like the Linux firmware issue @bibo01 describes also occurs on Windows but has just been noticed with DSD512 x 48k. It also might explain why @miska reported a negative test result.

@bibo01 maybe Amanero should use the test file to verify the new firmware and drivers he is working on?

[bibo01]

6 minutes ago, blue2 said:

@bibo01 maybe Amanero should use the test file to verify the new firmware and drivers he is working on?

When he releases a firmware,  it follows always a round of tests by people. Some corrections may be done if it is reported as needed.

[4est]

5 hours ago, louisxiawei said:

Thanks for testing.

 

As I mentioned, speaker system is quite difficult to detect the faint pop since there is a certain distance between your listening spot and speakers. So some speaker users don’t even bother the issue.

 

It’s not like a playback stutter caused by insufficient CPU capability when you upsample non-integer rate like 44.1 -> 48 x 512. It’s more like a static noise pops infrequently hidden at the background while the music flows.

 

I have one friend put his ear right close to the right channel of the speaker and then be able to detect the faint pop.

Maybe it is just me, but if you have to have your head in the speaker to hear it, it must be pretty faint. That said, it seems like a petty gripe about a product that is on the cusp of what is possible. It uses an Amanero. Their firmware has been spotty at these rates and they have been working on it for years now. Should we like it? Of course not, but I also do not think we should make a big issue out of it. As I said up stream, what are the other options for DSD512 for reasonable money that do not use an Amanero?

[louisxiawei]

2 hours ago, Solstice380 said:

@louisxiawei  I have a Lampi Golden Atlantic with Amanero USB using the Win Combo384 ASIO driver on a NUC NAA.   I have thought occasionally that I had heard the static pops but because they are very low level and random I wasn’t sure.  I also listen to music that generally does not have long quiet passages.  

 

I listened to the Harry Potter track 5 times and heard the pops.   Totally random, and low level, but they are there. 

 

Is it the Amanero driver?

Many thanks for testing. It helps. It's very likely the Amanero's problem. 

 

The static pop does occur randomly when playing that harry porter soundtrack, but the pop is somehow predictable with a certain time window, something like I will definitely hear the first pop after 10 seconds of the track. I've listened/tested to this track for 76 times (Roon records it).

 

I also installed the old version of T+A driver, didn't make the pop disappeared. 

 

@4est Yes, although that faint pop is quite audible through some revealing headphones like HD800, it can be ignored by people using speakers or even headphone users listen to dynamic music quite often. But I believe for some certain genre of music with long quiet passages or dead silent background such as solo piano, solo vocal pieces. These static pops are a bit annoying and can be audible through speakers.

 

I agree we shall be more forgiving and give people more time to deal with the problem, but pursuing the best possible SQ without flaw is what this audiophile hobby is all about, which includes stubborn people like me looking for reasons for these insignificant faint pops.

[Miska]

1 hour ago, blue2 said:

This report changes perspective on the problem. If the 'pops' are random then it sounds like the Linux firmware issue @bibo01 describes also occurs on Windows but has just been noticed with DSD512 x 48k. It also might explain why @miska reported a negative test result.

@bibo01 maybe Amanero should use the test file to verify the new firmware and drivers he is working on?

 

Such pops happen with various hardware when the source cannot keep up with latency requirements. UAC has 125 µs (8 kHz)  packet interval. Pop is likely what you hear if one of such packets missed it's deadline. Higher the sampling rate, more data goes into each packet. Drivers have their own buffer between application and the USB hardware. For the Amanero driver, this is non-adjustable fixed size that gets shortest in time at highest sampling rate which is of course 48k x512 in this case. In addition, there's a little bit of buffer also at the DAC side between USB and the output, but usually not much more than one or two UAC packets worth. It is also possible that the USB interface MCU in some cases misses it's own internal deadline for delivering input to output in timely manner.

 

It is somewhat different from stutter that happens when you run out of CPU power and longer portions (many ms) of audio miss a deadline. You can have a latency problem even if the CPU is not loaded at all. It can be also partially related to CPU load such way that latencies grow as the CPU load gets higher (this depends on driver implementation and how OS does scheduling). Latency problems can be caused by other drivers disabling interrupts for too long time. For example in the past one typical culprit have been WiFi drivers when they periodically perform SSID scan or change the the session encryption key (either every N bytes or K seconds). So your computer hardware, the OS setup and installed drivers have an impact on this.

 

I just finished fighting on such latency issue on microRendu NAA at DSD512 rates where it depended even on network traffic patterns, so only happened in certain cases. Since this is not Linux, unfortunately in this case with Amanero ASIO driver the fix would need to happen inside the ASIO driver and I cannot do much about it.

[Miska]

2 hours ago, Solstice380 said:

Is it the Amanero driver?

 

Likely either the driver, or the firmware...

[Miska]

3 hours ago, louisxiawei said:

Of course one can say that my newly built HQplayer PC and Logic CL100 NAA both have issue now. What are the odds? 

 

Both still have the same ASIO driver and the same DAC firmware...

 

Since you have a CL100 and iFi micro iDSD BL, how about downloading my bootable NAA image from here:

https://www.signalyst.eu/bins/naa/images/

Boot it up (from USB memory stick or SDcard), default buffer time in HQPlayer and try playing the same test to the iDSD. It will exceed tech specs of the DAC chip, but in any case it has been working for me. I've also used Pro-Ject Pre Box Digital S2 for testing the same thing. Both use XMOS USB interface and support 48k x512 DSD on Linux.

 

I have also newer 3.5.4 NAA images built, but it may take some days or a week to get those out (I want to verify some driver support things first).

 

[juanitox]

8 hours ago, Em2016 said:

 

Can we have a photo please ?

 

[louisxiawei]

1 hour ago, Miska said:

 

Both still have the same ASIO driver and the same DAC firmware...

 

Since you have a CL100 and iFi micro iDSD BL, how about downloading my bootable NAA image from here:

https://www.signalyst.eu/bins/naa/images/

Boot it up (from USB memory stick or SDcard), default buffer time in HQPlayer and try playing the same test to the iDSD. It will exceed tech specs of the DAC chip, but in any case it has been working for me. I've also used Pro-Ject Pre Box Digital S2 for testing the same thing. Both use XMOS USB interface and support 48k x512 DSD on Linux.

 

I have also newer 3.5.4 NAA images built, but it may take some days or a week to get those out (I want to verify some driver support things first).

 

Thanks for the advice and will report back after I finish the test.

[4est]

1 hour ago, louisxiawei said:

Many thanks for testing. It helps. It's very likely the Amanero's problem. 

 

The static pop does occur randomly when playing that harry porter soundtrack, but the pop is somehow predictable with a certain time window, something like I will definitely hear the first pop after 10 seconds of the track. I've listened/tested to this track for 76 times (Roon records it).

 

I also installed the old version of T+A driver, didn't make the pop disappeared. 

 

@4est Yes, although that faint pop is quite audible through some revealing headphones like HD800, it can be ignored by people using speakers or even headphone users listen to dynamic music quite often. But I believe for some certain genre of music with long quiet passages or dead silent background such as solo piano, solo vocal pieces. These static pops are a bit annoying and can be audible through speakers.

 

I agree we shall be more forgiving and give people more time to deal with the problem, but pursuing the best possible SQ without flaw is what this audiophile hobby is all about, which includes stubborn people like me looking for reasons for these insignificant faint pops.

I know that at one time I heard very faint pops/clicks using an Amanero. They were alleviated by a firmware change in my Amanero. There are quite a few variations of the Amanero firmware btw.

[Solstice380]

8 hours ago, Miska said:

 

Such pops happen with various hardware when the source cannot keep up with latency requirements. UAC has 125 µs (8 kHz)  packet interval. Pop is likely what you hear if one of such packets missed it's deadline. Higher the sampling rate, more data goes into each packet. Drivers have their own buffer between application and the USB hardware. For the Amanero driver, this is non-adjustable fixed size that gets shortest in time at highest sampling rate which is of course 48k x512 in this case. In addition, there's a little bit of buffer also at the DAC side between USB and the output, but usually not much more than one or two UAC packets worth. It is also possible that the USB interface MCU in some cases misses it's own internal deadline for delivering input to output in timely manner.

 

Thanks for the explanation.  Makes complete sense.  I’ll test tonight with 128 and 256 to see if the pops go away.   

 

I think i understood from a while back that there are some limitations in this version of the Amanero board (not sure if T+A put in the equivalent) such as memory or processor that was causing difficulties getting the 512 for Linux and Win ASIO in the same firmware.  I’m hoping the new firmware for Linux is close and solves all problems.  ⚡️ ?

 

I also think I heard that he is working on a new board.  Anybody know anything about this?

[sydlow]

22 hours ago, emcdade said:

Also, has anyone tried Roon’s new DSD 512 upsampling vs. HQplayer?

 

 

Yes, I prefer it over HQPlayer at DSD256 and DSD512

[Quadman]

Roon at 512 does not suck, but HQP at 512 xtr 2s mp beats it,  just more there. If you didn't compare you wouldn't know.

[blue2]

18 hours ago, Miska said:

 

Such pops happen with various hardware when the source cannot keep up with latency requirements. UAC has 125 µs (8 kHz)  packet interval. Pop is likely what you hear if one of such packets missed it's deadline. Higher the sampling rate, more data goes into each packet. Drivers have their own buffer between application and the USB hardware. For the Amanero driver, this is non-adjustable fixed size that gets shortest in time at highest sampling rate which is of course 48k x512 in this case. In addition, there's a little bit of buffer also at the DAC side between USB and the output, but usually not much more than one or two UAC packets worth. It is also possible that the USB interface MCU in some cases misses it's own internal deadline for delivering input to output in timely manner...

@Miska is the fact that the incoming DSD data stream is interrupted potentially quite serious as we could have a residual DC output, or is there likely to be some recovery logic in the Amanero interface? This could imply that although the 'pops' are barely audible they could damage your loudspeakers over time?

[Miska]

1 minute ago, blue2 said:

@Miska is the fact that the incoming DSD data stream is interrupted potentially quite serious as we could have a residual DC output, or is there likely to be some recovery logic in the Amanero interface? This could imply that although the 'pops' are barely audible they could damage your loudspeakers over time?

 

Usually DSD interfaces play silence pattern when such happen, this causes a slight pop due to small discontinuity in the waveform. If there would be sudden large DC, you would hear really loud "bang". And usually DACs have also DC block in the chain, either through decoupling capacitor or through a DC servo (which still doesn't block the loud bang-transition though). And another one in pre-amp and third one in the power amp.

 

So don't worry. You are much more likely to have low-frequency/DC type of loudspeaker problems with vinyl playback where that can happen unless your RIAA has aggressive subsonic filter.

 

[bibo01]

New update: I talked to Amanero and he found the cause for distortion at DSD256/512 and tinny sound. It was due to clock communication at high frequency over time. However, he found a method to overcome it - the receiver, differently from Windows, is going to constantly send back to the source a sort of reference signal to sync up to, telling exactly how many bits are needed for the next packet. He will test new firmware on T&A Dac too before releasing. As usual, further tests will follow.  

I am rather optimistic.

[louisxiawei]

On 5/30/2018 at 9:22 PM, Miska said:

Since you have a CL100 and iFi micro iDSD BL, how about downloading my bootable NAA image from here:

https://www.signalyst.eu/bins/naa/images/

Boot it up (from USB memory stick or SDcard), default buffer time in HQPlayer and try playing the same test to the iDSD.

@Miska I've done the test. Connecting CL100 as NAA to the iFi micro iDSD BL, boot up from the linux NAA image. Did exactly the same tests as before and I detected NO pop at all when upsampling 48K based file (48/24, 96/24, 192/24) to DSD512 x 48 using various filters (non-2s or 2s filters). 

 

BTW, can I adjust the IFI buffer size on linux NAA image? Want to try out if I reduce the buffer size will cause any pop noise.

 

Another thing I want to comment about the "faint pop" on DAC8 DSD is: It seems to appear only on the right channel.

[Solstice380]

2 minutes ago, louisxiawei said:

Another thing I want to comment about the "faint pop" on DAC8 DSD is: It seems to appear only on the right channel.

 

Same here on the Lampi.

 

[Miska]

8 hours ago, louisxiawei said:

@MiskaBTW, can I adjust the IFI buffer size on linux NAA image? Want to try out if I reduce the buffer size will cause any pop noise.

 

Yes, you can adjust it within the range allowed by "Buffer time" setting in HQPlayer.

 

8 hours ago, louisxiawei said:

Another thing I want to comment about the "faint pop" on DAC8 DSD is: It seems to appear only on the right channel.

 

That could indicate towards firmware, but it is really hard to know because everything from the ASIO driver to the Amanero output is a "black box" - something goes in and something comes out, but one doesn't know exactly what happens inside. So Amanero could probably better speculate the reasons.

 

[maggio]

On 5/29/2018 at 3:30 AM, louisxiawei said:

...

I've attached a 48/24 track, and would like to invite the DAC8 DSD users to try it out. In this track, you will probably spot a few "faint pop" sound in the first 20 seconds of the track on your "right channel".

If the problem does exist widely among DAC8 DSD users, then we can be confident enough to report this issue to T+A.

Any help would be appreciated.

 

 

03. Hedwig's Theme (From “Harry Potter And The Sorcerer's Stone”).flac

 

No pops here! ☺️

poly-sinc-xtr-mp-2s - ASDM7 - DSD512 x 48

Intel I7-7700

Server : Windows Server 2016 (Lite)

NAA: Windows Server 2016 (Lite)

 

[emcdade]

In your T+A experiences, which approach  do you think will sound better:

 

Native PCM—SMS-200 with Sbooster—T+A DAC8 DSD

 

DSD512–Windows PC with smps—Matrix X Hi pcie usb card — iso regen with Sbooster—T+A DAC8 DSD

 

????

[juanitox]

4 hours ago, emcdade said:

In your T+A experiences, which approach  do you think will sound better:

 

Native PCM—SMS-200 with Sbooster—T+A DAC8 DSD

 

DSD512–Windows PC with smps—Matrix X Hi pcie usb card — iso regen with Sbooster—T+A DAC8 DSD

 

????

definitly  DSD VS PCM

PC with LPS power supply-  iso regen  or INTONA with separate LPS- DAC8 DSD

 

 

[m3lraaHnevetS]

16 hours ago, emcdade said:

In your T+A experiences, which approach  do you think will sound better:

 

Native PCM—SMS-200 with Sbooster—T+A DAC8 DSD

 

DSD512–Windows PC with smps—Matrix X Hi pcie usb card — iso regen with Sbooster—T+A DAC8 DSD

 

????

PCM...I used to own the T+A DSD and did 512 for a while but to me PCM sounds more natural with finer detail, decay and air.

(especially using a better DAC like the Pink Faun I use now)

[juanitox]

15 hours ago, m3lraaHnevetS said:

PCM...I used to own the T+A DSD and did 512 for a while but to me PCM sounds more natural with finer detail, decay and air.

(especially using a better DAC like the Pink Faun I use now)

 

yes it"s true that even with the DAC8DSD PCM sounds more 'hifi" with crisp and finer detail , but fpr vinyl lover DSD512 has a sort of natural flow and analog sound that is very addictive ( some will say a blurring effect)  .  it"s not a contest of Better or worst but two different way to enjoy the music .

if i had to buy another dac i will not buy rhe T+Adac8 if it had PCM only..