I really owe you an update on this project. The problem is that I don't know where to start and how to keep it short. And I am currently too pressed for time to make it an essay!
matthias wrote:Fredrik,
from a Source First POV the streamer is much more important than the DAC.
So I am wondering if it is easier to develop a pure streamer with ethernet input and maybe USB output for a DAC. You could concentrate your precious time and efforts on the first parts of the digital chain where it matters most.
Please correct me if I am wrong.
Yes, the digital parts of the streamer are much more important than the DAC for the musical qualities. But in addition, the NAS and Switch (if using ethernet to supply your streamer with the digital dataI) are
at least as important as the streamer. What I initially intended to achieve was an understanding of the mechanisms behind this, but that proved extremely difficult. When talking to others in the HiFi industry, I've drawn the conclusion that nobody else understand them either. But there are patterns and laws which give consistent, repeatable results. So it's practially possible to make a really good digital platform for fetching data and delivering it to a DAC. A massive amount of trial and error is required, as every part of the hardware matters, but it certainly is possible.
Then there's software. And this is where the project goes from a big to an enormous challenge. The firmware which runs the digital platform has a fundamental influence on musicality. The impact is big enough to make excellent hardware sound really bad, or make mediocre hardware sound quite good. So more trial and error is required, and not just of the parts that are involved in handling the musical data. No, every detail of the firmware has an impact. Some a little, some a lot and some things I don't dare mention as you would laugh and question my sanity.
This would all be fine and no nightmare, as I enjoy challenges and don't mind organising lots of trial&error. But the platform needs to adapt to an ever changing environment. Different networks, routers, smartphones and tablets all require updates to the firmware and/or the software. And the apps and streaming service plugins (Tidal, Spotify, Google Play, etc) all require constant updates too. For instance, my programmer made some ingenious work on how to increase quality of streaming material, which later turned out to be unusable when Apple now launches iOS11. So, if we for a second completely forget the musical quality of the streamer, it would still require constant work on updates that keep it working flawlessly. Quality would go up and down with each firmware update - and this I have mocked other companies for - but it would work. Now, add that I want to promise that with each update, your streamer will sound
better than before - and the difficulty increases exponentially. We've already run into bugs that can't be fixed without a drop in sound quality. Painstaking workarounds and compensations have been made, only to later find that no, the previous version with an annoying bug still performs better.
As my focus is on musical quality, not making a swiss army knife, I've tried radically limiting functionality. The network and streaming services are a large part of the difficulties. I know many customers wouldn't want a machine without them, but I still would love it if it gave me a bigger musical thrill than anything else. The challenge here is that I have so far
not been able to make it perform as good as when the data is retrieved from the network. This is quite mysterious, but the results are consistent - there is a very defined path and speed that the data needs to travel for the digital platform to perform optimally. Even if the digital platform is built and tuned for retreiving data from a local memory, it still performs much better when later converted to be used on an optimised network with data storage. And the optimal design of the network and data storage appears to be the same regardless of which digital platform or streamer one uses.
OK, assume we now have an incredible digital platform. It feeds a DAC, which converts the data to analogue music. This can be done in many different ways, both the transfer and the conversion, and the impact on sound quality is big. With a bad transfer to the DAC, you can loose the quality of the signal. I was sceptical to USB at first, but it turned out to be really good if well tuned, so it's possible to have an external DAC and still achieve great results. There are other options as well and in general easier to optimise a DAC that sits very close to the digital feed of the streamer. I am not yet sure which solution can be taken furthest, as the theoretically optimal ways are not always practically the best. There's a give-and-take in that some parameters improve when
separating parts and other parameters improve when
integrating parts. In any case, all of this is more "at home" to me, as it's more analogue in nature and easier to tune, so I'm mostly looking forward to testing and evaluation all the options that spring to mind.
The digital platform, network and data retreival, however, is not easy. I finished testing my 10th prototype last month and since then I've taken a break. Some breakthrough is needed to take this project further. Either a deeper understanding of the digital mechanisms that so strongly affect musicality
or a new idea for how the platform can be organised.