T+A new flagship streaming dsd1024 dac

[asdf1000]

18 hours ago, OE333 said:

As ethernet is using asynchronuos transfer mode and thus uses one of the 3 USB inputs - the other two are available as external inputs.

 

Thanks so much for the info and the photo.

 

So the answer is yes, that the ethernet input still goes through a USB connection? So things like 8kHz USB packet noise still applies? Even if they are very very low levels, they still apply in the technical sense?

 

If I am wrong, please correct me. But if I'm correct, what is the reason/s USB must still be in the chain/process?

 

 

[OE333]

Yes, you are right, the signal path is Ethernet -> USB -> DSD.

 

We use USB at this stage because it is asynchronous. Keeping the signal path asynchronous as long as possible means that the synchronous D/A clocking comes into play only at the very last stage of the signal chain. If we would use I2S or DSD with synchronous clocking at an early stage to transfer the signals to the DSP this would lengthen the synchronous part of the path introduce more jitter. 

 

8kHz packet noise is not a problem with this converter because we have a complete galvanic isolation between digital and analog sections. So any kind of digital noise does not influence the converter and the analog output sections.

 

Here photo of a prototype D/A converter board. Maybe you can see the white line which shows the boundary between digital and the isolated converter/analog circuit parts...

 

[asdf1000]

5 minutes ago, OE333 said:

Yes, you are right, the signal path is Ethernet -> USB -> DSD.

 

We use USB at this stage because it is asynchronous. Keeping the signal path asynchronous as long as possible means that the synchronous D/A clocking comes into play only at the very last stage of the signal chain. If we would use I2S or DSD with synchronous clocking at an early stage to transfer the signals to the DSP this would lengthen the synchronous part of the path introduce more jitter. 

 

8kHz packet noise is not a problem with this converter because we have a complete galvanic isolation between digital and analog sections. So any kind of digital noise does not influence the converter and the analog output sections.

 

Here photo of a prototype D/A converter board. Maybe you can see the white line which shows the boundary between digital and the isolated converter/analog circuit parts...

 

Greatly appreciate the detailed response. And the DAC p0rn 😁 (photo of inside).

 

Fantastic engineering (again).

 

[Ben-M]

On 4/8/2019 at 8:37 PM, OE333 said:

 

... even though a bit off topic in this thread:

There is no "old" or "new" Amanero board - in fact there is only one Amanero USB receiver design. Most other companies just use the Amanero standard module. T+A decided to place the whole Amanero hardware directly on the DAC  PCB to get shorter signal paths - otherwise everything is 100% compatible to the Amanero original design.

 

Please believe me, we try what we can to push Amanero and we give all support that we can (like paying license fees, testing, testing testing, helping Amanero with hardware etc.) but we can not help with the development of the receivcer firmware itsself because it is closed source....

 

For the new SD/SDV3100HV devices with their DSD1024 capability (which is out of reach for the Amanero design) we decided to design our own UAC3 compliant proprietary USB receiver solution. This solution supports native DSD1024 for both Windows and LInux. The necessary Linux patches were made available by us and we hope that they will be included in mainline kernel soon. For the time beeing anybody interested in the patches please send me a PM.

When I was researching about the issue beforehand and later when I was discussing it with Domenico from Amanero, that was what I understood. I guess I was incorrect and I appreciate the courteous way that you have corrected me about it. 

 

If there is a time that I can afford the new SDV3100HV, I will enjoy giving it a shot. I am still happy with the DAC8DSD's sound quality, but it's lack of modern Linux compatibility is not a pleasant thing to deal with in my system. I think the approach that you've taken with the SDV3100HV's USB and ethernet implementations will serve you very well into the future. I'm very happy to see your company taking a different approach this time around and I hope the SDV3100HV meets a warm welcome in the market. Hopefully I can join those new buyers sometime later.

[OE333]

Dear Ben,

there are ways to use native DSD256 with DAC8DSD and Linux. I know that some members of this forum do this with very good success.

I would like to propose to continue the DAC8DSD related questions in the DAC8DSD thread.

[Schedel]

Are there plans to incorpurate this technology in the much more affordible R-series?

MP 2000 mk3?

Billy

[OE333]

... the R series already has all elements of this technology i.e. T+A True-1-Bit DAC, extremely precice re-clocking, digital/analog isolation, discrete analog electronics without OP-Amps etc.

 

The difference is that the R-series supports a maximum DSD rate of 48k*512. True-1-Bit processing @ DSD1024 is really extremely complicated and costly and will remain an exclusive feature of the SDV3100HV / SD3100HV models.

 

[StreamFidelity]

Hi OE333,

 

As a T + A fan, I've missed a high-quality streaming DAC in the HV segment for some time now. Nice that there is one now.

 

Currently I enjoy listening to the Holo Audio Spring DAC contrary to the usual opinion in oversampling (OS) DSD mode: Either PCM or DSD input over all sampled DSD approach to digital to analog converter.

 

That sounds very good to my ears. Some manufacturers also go this way and convert every signal (including PCM) into DSD. For example EMM Labs DAC V2 or PS Audio DirectStream DAC. Optionally also in Lumin DACs.

 

Is this option also possible in the SDV3100HV / SD3100HV?

[Sunya]

Since the SD/SDV3100 now support PCM up to 768kHz it means you switched to AKM from BB; is it the AK4497 or the very new AK4499?

 

Also since you don't use the chip built in digital filters, do you upsample the signal in the DSP to 705.6/768kHz?

[barrows]

@OE333, thank you for posting the details of T+As very impressive new DAC here, your contribution to the forums is appreciated!  As you appear to be a direct representative of T+A, please add something to your signature indicating your relationship to the company.  Again, I am very happy to be able to have you post here, and share details of the product!

 

I suspect that this DAC will offer extremely high sound quality, and am impressed at the details of the engineering.  Many here on theses forums have longed for a DAC which could support high DSD rates over a direct Ethernet input, and it appears that T+A has done a very good job of isolating the Ethernet processing section from the D/A section and analog output circuitry.  As the rest of the engineering of this product appears to be very well implemented, can you also confirm a completely separate power supply for the Ethernet input section (with it's own transformer secondary?)

Again, I am very impressed with the details if this new DAC and suspect that those who can afford it will be very, very happy! 

[StreamFidelity]

I heard the new streaming DAC at King Music in Berlin. If interested, my report (in German).

 

Report on T+A SDV 3000 HV

[OE333]

On 4/12/2019 at 5:01 PM, barrows said:

@OE333, thank you for posting the details of T+As very impressive new DAC here, your contribution to the forums is appreciated!  As you appear to be a direct representative of T+A, please add something to your signature indicating your relationship to the company.  Again, I am very happy to be able to have you post here, and share details of the product!

 

Yes, good point. I have added a signature.

I want to point out, that I am here in this forum being an audiophile myself.

I am happy if I can share my knowledge and give some insights into T+A technology and developments and I offer my help to members of this forum in all questions regarding T+A devices...

 

[OE333]

On 4/11/2019 at 12:54 AM, Sunya said:

Since the SD/SDV3100 now support PCM up to 768kHz it means you switched to AKM from BB; is it the AK4497 or the very new AK4499?

 

Also since you don't use the chip built in digital filters, do you upsample the signal in the DSP to 705.6/768kHz?

 

I respect AKM as a very capable manufacturer of very good sounding DACs.

In the SDV3100HV  TI (formerly BurrBrown) PCM1795 are used. These chips can be used at higher rates than 384 if the internal upsampling filters are bypassed. Bypassing the internal filters makes sense in our case because we use our own proprietary T+A upsampling algorithms. All upsampling is done is done in the DSP.

[OE333]

On 4/10/2019 at 4:42 PM, StreamFidelity said:

Hi OE333,

 

As a T + A fan, I've missed a high-quality streaming DAC in the HV segment for some time now. Nice that there is one now.

 

Currently I enjoy listening to the Holo Audio Spring DAC contrary to the usual opinion in oversampling (OS) DSD mode: Either PCM or DSD input over all sampled DSD approach to digital to analog converter.

 

That sounds very good to my ears. Some manufacturers also go this way and convert every signal (including PCM) into DSD. For example EMM Labs DAC V2 or PS Audio DirectStream DAC. Optionally also in Lumin DACs.

 

Is this option also possible in the SDV3100HV / SD3100HV?

 

Conversion to ultra-high rate DSD can be done in many ways but only the very best algorithms will bring an optimal performance. The computing power needed for such algorithms is extreme and can be best achieved with very powerful PCs (and GPU co-processing). So we leave the task of conversion/upsampling to specialized programs like HQPlayer and concentrate on doing a best possible D/A conversion with the ultra-high rate DSD signals delivered from the Audio PC. Also it is only possible with a PC based upsampling to support the vast number of algorithms as for example HQPlayer does.

This approach is also very future-proof as both the upsampling software and the PC hardware can easily be upgraded when new algorithms get available and more powerful processors come to the market. Such easy upgrade is almost impossible if the conversion is an integral part of the DAC.

For PCM upsampling things are somewhat different. Here less computing power is needed and the upsampling can well be integrated - like it is done in the T+A DACs. But also with PCM it is possible to do the upsampling externally and send PCM 768 directly to the DAC completely bypassing the internal processing.

 

All this does not mean that there is no DSD processing done in the DAC. With the T+A True 1 Bit Converter each incoming DSD bit is converted by 16 balanced converter stages resulting in up to 3.1 billion conversions per second. This converter structure performs a filtering function and reduces high frequency noise  just like additional oversampling would do. So in the SDV 3100HV a DSD1024 input stream results in fs*1024*32 conversions per second and the outcome is a very clean analog signal...

 

 

 

 

[ferenc]

On 4/8/2019 at 3:44 PM, OE333 said:

Hi Em2016,

 

I can confirm that jitter performance is the same for all formats and rates. The D/A converter is running completely on clock signals generated by the master clock oscillators located directly in the converter section.

 

Currently only the NAA protocol is supported on the ethernet module but this could easily be changed. For ethernet there is a separate single board computer module and it would be possible to implement other protocols as well.

During development I have also tested with roon bridge on that SBC and it also worked perfectly well but of course limited to DSD512 which is currently the maximum rate supported by roon.

As we wanted to have the best performance we decided for Signalysts NAA as we think this gives the best quality and highest data rates currently available.

 

But recommendations for other protocols are welcome...

 

 

 

 This is quite a bit an endorsement for @Miska  . He deserves it. 

[Cortes]

@OE33,

 

 the ingeneering looks really impressive, as looking at a Ferrary, and also the features. But, if one closes the eyes, do you have any proof ( some style of double blind test) that an average listenner is able to discern  the $26k T+A  from a good $1k o $500 DAC?.

 

 

[OE333]

13 minutes ago, Cortes said:

@OE33,

 

 the ingeneering looks really impressive, as looking at a Ferrary, and also the features. But, if one closes the eyes, do you have any proof ( some style of double blind test) that an average listenner is able to discern  the $26k T+A  from a good $1k o $500 DAC?.

 

 

 

 

proof ?  Well: MEASURE & LISTEN

 

The phsyscist in me says MEASURE !

By measurements we can prove that no engineering faults were made and that technically everything works as it should.

But of course a technically correct and well engineered device can only be the necessary basis for a good sounding eqipment.

 

So the audiophile in me says LISTEN !

The problem with listening tests is that all such tests are more or less subjective. In almost all cases none of the listeners was present when the recording was made - so how can be judged if equipment A or equipment B is closer to the original ?

The only thing we can do is listen and find out what we like best.

And then of course there is the final question "Is it worth the money ?"

I think there is no objective answer to this question - this question can only be answered very subjectively by everyone for her-/himself...

[StreamFidelity]

@Cortes

 

I want to perform this unequal test, but not double blind test.😎

 

My DAC story is a bit unusual. When comparing a Linn Klimax DSM and the T+A MP 3000 HV, I came to the decision to sell the Linn. Not because the device was bad, but because it did not sound better in my ears than the MP 3000.

 

Because I once wanted to have a R2R DAC I bought the Armature Asterion Holo Spring DAC. With this device I discovered my new preference for DSD, which the MP 3000 can not convert. The Spring DAC from France costs 1.6 k € and has replaced my MP 3000.

 

My setup currently looks like this:

 

 

With the HQPlayer I convert everything in DSD 256. Theoretically synonymous DSD 512, but I suspect a bottleneck in the USB interface (XMOS U208) from Spring DAC. I am very curious, what I can push over the SD 3100 and how it sounds. I'll get the SD 3100 HV in a few weeks.

 

[matthias]

8 minutes ago, StreamFidelity said:

 

My DAC story is a bit unusual. When comparing a Linn Klimax DSM and the T+A MP 3000 HV, I came to the decision to sell the Linn. Not because the device was bad, but because it did not sound better in my ears than the MP 3000.

 

May I ask which Klimax DSM, the latest with Katalyst architecture?

These Klimax Katalyst accept DSD up to DSD128.

 

Matt

 

[StreamFidelity]

Yes, it was the Linn Klimax Katalyst. I had an upgrade done then. By the way a service, which T+A would be good to face. 😉

 

I did not hear any DSD at the time.

[OE333]

@StreamFidelity an upgrade for the MP3000HV is currently under development....

[StreamFidelity]

@OE333

 

That would be great! Is there information about the specifications?

[OE333]

7 minutes ago, StreamFidelity said:

@OE333

 

That would be great! Is there information about the specifications? 

 

 

... not yet - but official information will be available soon.

 

[guiltyboxswapper]

48 minutes ago, OE333 said:

@StreamFidelity an upgrade for the MP3000HV is currently under development....

 

When you mean upgrade, do you mean a new product replacing the MP3000HV?  Is it likely to have the same ethernet input as the 3100?

[OE333]

On 5/3/2019 at 6:42 PM, guiltyboxswapper said:

 

When you mean upgrade, do you mean a new product replacing the MP3000HV?  Is it likely to have the same ethernet input as the 3100?

 

No, with upgrade I mean an upgrade of existing MP3000HV to bring most of the features of the successor model MP3100HV to the older models. This upgrade will mean a complete exchange of the digital processing board and the streaming module.

 

Details about the MP3000HV upgrade program will be announced in the near future by T+A.