Oh yeah. No idea about audio applications back then, although no reason they couldn’t have been. They just replace the discrete gates and memory devices that formed the logic and state machines, but I suspect you’re right that audio didn’t use so long ago.
It was a good gig - worked on the ultra high speed telecom shelfs/racks that were pumping 1’s and 0’s around the internet in the early days. When Xilinx and Altera were designing their new products (next-generation of FPGAs), they would actually meet with us and ask what features we needed (how many RAM blocks, what size, how configured, what speed, gate count, topology, etc.), and it wasn’t lip service - they actually listened and catered. That was the advantage of being on the upswing as the Internet came to dominate our lives, haha, sigh.
Audio by contrast is a little less leading edge (slower moving, small market share), but its all good. Doesn’t make design choices and solutions any less interesting or challenging or cool, just that audio companies don’t deal in volumes that can sway FPGA manufacturers’ courses, I suspect. But probably the advancements are much more appreciated by the end users at the consumer level, so that is super rewarding I’m sure. FPGAs were a huge game-changer in letting us move from one-shot ASICs to something that could be reconfigured in the field. That was an amazing advance.
Xilinx (what Ted uses) was a big player back then but I preferred Altera simply for the operating and development environment. The tools were much less sophisticated then (I started in 1994) as you could imagine.
I assume it is not only the fpga which sets the limits of improvements. Also the quality of other components define the weakest link in the audio chain. It is useless to decrease the noise floor in your fpga if other components define the final noise level. So, a waste of effort. I think Ted explained that earlier. So, there is always a moment that you just have to improve on hardware outside the fpga. That hopefully gives plenty of opportunities to improve on fpga software in years after the release of the mk ii hardware. If only money and waf could keep up with the that pace…
That’s all undisputed. The point of discussion was, if the point in time was accidentally now or chosen by commercial reasons. It’s fine for me to accept the latter, I just don’t force to convince myself it was the first.
I think this is normal. For years, Paul and Ted’s approach was a hobbyist approach. We find it odd that we are used to that approach. I bought a second hand last year, so I haven’t benefited from any updates, but I think I should thank you for years of free updates.
On the contrary, you’re benefiting from all the updates, all the improvements Ted has made over the years. All old versions are still available if you’d like to load and compare.
The DS will keep doing what it does at infinitum. It’s not obsolete unless you say it is.
Indeed, I fought this for quite a while with the current update work and finally found a place to put one pipe stage and cut the compile time down from more than an hour to 5 min. Still there just isn’t enough memory for coefficients to significantly improve the upsampling filters in the old FPGA. The rest of the old FPGA has been at 150% usage before optimization to 99% after for years.
Yes there are still two I2S inputs. There are now also two AES3 (XLR) digital inputs as well as the expected RCA, TOSLink and USB. All inputs will be at least as well functioning as their current versions so you won’t go wrong since doing what’s best for the DS in your system will also be very good or best in the Mk II in your system.
I was surprised by the announcement, but it makes sense for PS Audio since otherwise they need to keep a frozen set of tools to work on very old code, code that seriously needed replacing. On the FPGA side, the FPGA that’s in the DS and DS Jr requires a toolset from Xilinx that nominally doesn’t run on Windows 10 (tho you can fake it with a virtual machine running, say, Windows 7.) The new tools from Xilinx support the new FPGAs used in the DS Mk II and TSS, but don’t support the older FPGA used in the Jr and Sr. At this point supporting the older tool sets both for the FPGAs here and also at PS Audio for the display/control processor becomes more and more of a drag on development resources.
Yes. each board has it’s own two windings from a oversized toroid transformer, from there there are multiple sets of regulators to each load. Here’s a post that addresses a slightly different question but also give more information about the power supplies.
That fits to your previous statements and credibility.
That’s comprehensible.
I’m in for the next Ted Smith DAC I can afford. For me, with a later finally multiple tuned DS, I think it would need a later or bigger step than the Mk II some time.
