T+A new flagship streaming dsd1024 dac

[jabbr]

I’ve considered this approach, and agree. 

1 hour ago, barrows said:

The more affordable solution is to just build the single bit DAC and sell it, and require DSD 256 (and up) input.  While it is a little tweaky, one need not get involved with things like HQPlayer to take advantage of this approach anymore: ROON does fine oversampling to DSD 256, for example, and it is simple, and well accepted by many audiophiles.  This approach is much more affordable, and has the potential to offer improved performance over "full featured" DACs.

Well if we are doing this let’s use DSD512   I guess you could do it in kit form? Aren’t there DSC-1 kits of some type? Are they popular?

 

No I expect this approach would need to be more “high end” and a good workstation is needed for upsampling. So you are already bumping up against the Holo Spring/May and T&A 8...

 

Now if you want to use an Ethernet input, there’s the cost of that interface and support.... Considering the opticalRendu is ?$1200 or so, what would the price of a DAC be?

[jabbr]

13 minutes ago, Miska said:

 

At least the T+A flagship has DSD1024 NAA capable ethernet interface. But I'm especially curious about the new HA 200 headphone amp version that is not nearly as expensive.

 

I’m assuming that your P&F will accept Ethernet +|- Wi-Fi in and audio out, so a “power DAC” (with a radar jamming function built in 😂)

[jabbr]

1 hour ago, barrows said:

Jussi has shown that Holo Audio DSD DACs tend to perform better at 256 than at 512.

Hmmm ... the May? Maybe not so, really?

 

The May has DSD1024 in and they say probably DSD2048, so I expect it would perform optimally at DSD512 but if there are measurements which suggest otherwise then I’d love to see. 

[jabbr]

1 hour ago, barrows said:

In most cases 512 requires a higher speed clock, which means more jitter, as for any given clock quality, jitter goes up as speed increases.

Side note: I worked out the math, and posted here over a year ago that as DSD rates increase, the increased clock jitter is less of an issue (the samples average out and are further away from the audio frequency so the clock is filtered even more effectively away)

[jabbr]

20 minutes ago, vortecjr said:

I would categorize this as a computer and a DAC in the same enclosure. 

 

Arguably any DAC which accepts firmware is a "computer" and DAC in the same enclosure, or even on the same chip for that matter. Typically the USB interface and conversion to I2S/DSD involves some type of FSM/processing. It turns out that essentially all more than trivial FPGA implementions need some soft of processor called a "soft core". Building a SoC such as the Xilinx zynq places an FPGA and ARM cores in the same die ... the RF SoC even includes a few ADC/DAC modules intended to implement SDR (software defined radio) ... but with the FPGA, implementing the Ethernet interface becomes easy because you can load an Ethernet IP module (ie PHY) and the chip itself can interface directly to an SFP module. I have no idea how T+A specifally does this, and Intel also has its own similar chips but that is a straightforward and flexible approach. 

 

Merging also implements this approach, AFAIK, with their "ZMAN" board "Z" for "Zynq"  

 

Enabling NAA is as simple as loading a deb package onto the custom Linux.

[jabbr]

22 minutes ago, vortecjr said:

I was not referring to the T+A and would categorize it similar to the PS Audio network input. A Rendu is somewhere in between a computer mother board with all the bells and whistles and the PS Audio network input. Just understand the Rendu has a USB output, but no keyboard connection, PCIE, sata, or on board video, etc. As much stuff as we excluded it’s a miracle it still works:) 

 

Im still not clear what is impossible about making a device like exaSound 8-channel with USB Audio Class? I streaming multi channel to my test DAC via USB. Exa has a proprietary multi channel driver for Linux for some unique reason particular to their USB interface. I’m just not following you. 

 

Sure, just saying that there is a broad spectrum of implementations. The concept of SoC incorporating IO and processing has become ubiquitous e.g. https://www.mellanox.com/products/bluefield2-overview so that would be extraordinary overkill but literally NAA could run on the NIC -- no one would ever do that would they ... hmm ...  

 

exasound ... could do it if they wanted to, the advantage of using an FPGA is that the solution could be delivered in firmware ... but limitations all have to do with selected hardware e.g. 100m vs 1g vs 10gbe

 

Once USB4 becomes more popular and when we see USB-C input DACs things may change again -- at least then we will be able to better separate the power from data if/when desired.