That’s actually a good point though.
PS Audio should sell merch. Id buy a hat and a shirt.
I want a hat too! XXXL for size 8 head.
A year has passed since the last update. I’m very satisfied with the current release but am dying to get the DSD256 Input (plus other improvements)! Could I expect the new update to come out by Christmas?!
I don’t like to say it’s ready before it actually is. Christmas would be nice but I don’t know how much work PS Audio has to do for this next release so I can’t really even guess.
14 Likes
Yes Christmas please
At least there’s still something good in 2020 
2 Likes
Yes… no, wait… if it’s released in 2020, somethings bound to go wrong with it. Maybe best to wait until 2021!
5 Likes
No pressure, Ted. 
Nice. Looking forward to the new update.
What do you hear there that is sonically different from the current one?
That would make it into the other thread…you’re an audiophile, if all you wish for Christmas is a lower noise floor from a guy named Ted 
6 Likes
That’s funny shtuff. 
I have a Aesthetix DAC with eclipse mod. I was hoping to get an update to help the PSA dac. The PSA is now playing second fiddle to the Aesthetix, not to say the PSA is bad. It has been my reference dac since the day I purchased it. Even without I2S the PSA isn’t ahead anymore in my system. Just being honest.
I expect the new update to be even more of the same, reflecting Ted’s continuing efforts to eliminate noise.
The user reports following each release are always telling and informative. There are some great ears in this group.
1 Like
I found the Aesthetix dac to have more air, better focus, spot on center image, wider and deeper sound stage. What I found most amazing is the bass was more natural and less bloated too. As a tube dac that surprised me.
Like you I have been exploring different DACs. I agree the PSA DSD isn’t bad at all. You just found something that works better for you. Aesthetix is a killer company and makes awesome products.
Congrats on the Aesthetix DAC.
I’ll likely post my DS Dac for sale soon.
The other shoe drops 
1 Like
As I modify parts of the code I’m always keeping noise and jitter in mind. At times putting more things in parallel is good (potentially lower clocking rates and noise.) Other times using a pipeline instead gives faster results and evens out resource utilization better.
For quadrate I2S I’m redoing that code. Also over time the I2S code has gotten cleaner. Much of the I2S code is doing checking for DoP, inferring that native DSD is being used, etc. Some of the non-obvious things involve letting the resources used span a bigger part of the FPGA to allow better routing of other features.
This is still on the back burner, each release I’ve found (or been guided to) other things that had the potential to make a bigger difference.
Each of the last three releases have involved balancing memory blocks/multiply block against general purpose fabric.
I don’t remember writing like in particular but there have been two recent releases where I’ve cleaned up the SDM code. By changing the clock rate a little it allowed a more straight forward implementation.
That doesn’t ring a bell without more context, but in early releases much of the sequence of processing
alternated left and right samples, the newer code usually does the left and right in parallel. The biggest result was that the older code sometimes swapped channels, now it can’t.
It’s easy to say, but harder to do well. You need to look further into the future to see which stream is better than I’d like. The each additional sample further into the future can double the choices with a naïve implementation. The smarter implementations need to make decisions about which streams are better with less information when using, say, one more sample might lead to a very different decision. I’m still working on it every once in a while.
This has some of the problems from the last item. Picking the best stream is more complicated that it might seem (or perhaps I should say, I haven’t found efficient, reliable code to pick “the closest stream”. When I can do that well I’ll see which of these two items helps the most.
? There’s a balance of raising low pass filter cutoffs to allow more highs from the original signal and how much more ultrasonic noise is let thru. I’m already higher than most SACD players, but there are certainly some amps, speakers that do less well with more ultrasonic signal (or noise.)
PCM has been upsampled to 352.8k (or 384k) for a few releases now.
That’s PS Audio code, they’ve done some of it in the last two releases.
Using overlapping FFTs to implement the low pass filtering for the upsamplers can allow implementation of much larger filters, but in the current hardware the memory to store the filter coefficients is the bottle neck.
Overall for each release I start from a blank slate and work on the worst current problem and/or a new feature. As I work on those I sometimes choose some unfinished ideas from the past, simple changes in the current code and sometimes writing a bunch of new code. The disadvantage of a lot of new code is potentially a lot of new bugs.
At this point implementing quad rate DSD input over I2S is the primary focus, after that I’ll probably do some more work to improve the sound.
19 Likes
After reading Ted’s post. Maybe I’ll hold onto her for a tad bit longer. LOL
4 Likes
You are a fickle fan, Mr. D.
[PS, I don’t recall from previous posts, do you have any experience you could share regarding Aesthetix pre-amps. I have been intrigued ever since Paul “blamed” one for his “on the Road to Damascus” conversion from the “there is not preamp like no preamp” camp.]
I own a Janus Signature. I sold the BHK preamp a few months ago. I liked the BHK alot, but the Janus Signature is on another level, plus rather good to very good built in phono stage. I am thinking next year to send it in for the Eclipse mod. The Janus has more features as well, dual mono design, full tube output. More air, depth, width, and air. Less noise too. I am very happy with the Janus.
2 Likes