Ted, I was wondering : you said that you suspect it is a configuration parameter problem.
Would it help if someone living in Boulder (or in the close vicinity) with a “bad” Windom load were to bring his DSD Sr. to the PS Audio plant for analysis of the memory and compare it with a DSD Sr. with a good load with the hope of isolating which area of the memory gets corrupted?
Offering a good rebate on some PS audio equipment could be a good incentive 
.
I can’t speak for PS Audio engineering (and I don’t work in Boulder) but that’s a lot of flash memory to comb thru looking for a difference that matters and then one would have to find the code that stomps it (which may not be a part of the update process…)
Yes! When you told us you had reduced noise and jitter, I fully expected the soundstage to improve, detail and clarity to increase, and a more full sound. Why? Because noise and jitter take away from all of those things. This is why I find it interesting that some people still prefer Snowmass. It means they like the what the noise and jitter is doing the signal.
3 Likes
First of all, I would like to thank Ted, Paul and the whole PS Audio family for such a service. We are never stuck with what we already bought and actually have choices on how our DAC sounds. On top of it all, they are free!
After saying that, and going from Windom to Snowmass many times for the last couple of days, I also decided on Snowmass for now. I have a horn loaded + open baffle speaker and a high powered integrated amp (Pass Int-250) and with Snowmass there is a perfect balance in musicality that I won’t get into detail (not to bore you). With Windom the balance is tipped a little. Symptoms of harshness begins to show that I was used to when I had dCS Rossini + Clock. By the way I UPGRADED from Rossini to DS DAC. That’s another story.
So, in the future when I buy a better matching hybrid amp (or a 211 vacuum amp) for my Diesis speakers, Windom might balance the system out.
But, even without having the perfectly matching amp, the sound is sublime with Snowmass in my listening room.
Thank you Ted, Paul and the team.
4 Likes
I was thinking that with this approach, we could at least confirm that the FPGA load is correct: since it has a CRC, I would expect the FPGA firmware memory area to be absolutely identical on both units - that should be somewhat quick to validate.
Then, may be using good educated guess, you could identify some potential problematic memory area of the code to be investigated (i.e. narrow the field). As a coder myself, I do that all the time when I have to debug software: of course the more experience you gain in the code, the better you become at making good educated guess.
You could also give that task to a promising student (less expensive) and lend a DSD Sr. with a good load to the customer who brought you the DAC with a “bad” load.
In this way, you would a lot of time to investigate with a student that would be supervised by an experienced employee.
I know I may sound cray, but I am just trying to suggest a resolution strategy to the bad loads that are being reported.
I don’t presume to tell the person who gets assigned this task how to do it. But as you mention, when the time comes there are probably many volunteers with update problems if the problem remains unreproducible at PS Audio.
I just did an experiment that worked as I had hoped.
I took my BHK preamp out of the loop; listened to the DAC directly and the improvements in Windom became immediately obvious. Very easy to hear that Windom is better.
The absence of the preamp is still a net loss but it seemed to mask the Windom improvements—at least from my ears. Yet another puzzlement to me.
1 Like
Well, that’s unexpected… Thanks.
It could well be that I keep hearing what I expect to hear.
@RonP, that’s is why I don’t make rash judgements at first listen even if I can hear what I think is an improvement. I need to listen for a decent amount of to know if the difference I hear is better, worse, or just different.
Thanks Ted, good arguments I didn’t think about yet in this context. Probably a reason why a lot of better gear tends to sound richer and more natural.
This is all so peculiar! For me, Windom increased detail over Snowmass, but decreased richness. Windom’s bass is more interesting, I’m hearing different layers that I hadn’t before, but for me Snowmass’ midrange is more engaging. I went through a few faithful recordings that always prove my system sounds great, things like Lyle Lovett’s Pontiac, a variety of super high quality 1950’s sax and trumpet based jazz, and Cecilia Bartoli’s Vivaldi album. On those Windom left me wondering what was wrong, it was thin.
In my particular case, I wonder if HQPlayer is playing a significant role. I have Roon/ HQP all on my MacBook Pro which is not the preferred high brow way of setting them up. HQP tips the scales already in the detail direction, so perhaps Windom, on my system, tips it all too? Amarra’s rich sound is probably more compatible with Windom, but I’ve been down that buggy road and remain all in on Roon/ HQP.
Do you have Qobuz or Tidal? If you do, find Bob Dylan’s “Oh Mercy” album and play the track “Man in the Long Black Coat” and tell us what you hear.
1 Like
I use Roon without HQPlayer. Windom does sound exceptional. I especially like the details that I’ve never heard before and for me Bass and Mid Bass is there. You can defeat HQ Player and try it that way. It seems counter intuitive to Up sample when the DAC up samples in DSD.
Thanks for the suggestion - I just don’t have time to listen to all of my stuff between releases 
3 Likes
It’s not. Ted and I have discussed this in another thread. I upsample all PCM to 352.8 and all DSD to DSD128.
No worries, that information wasn’t coming off of my hat, but from @Schroedster over the phone when he kindly provided me the SD card for 3.06, which gave a slight improvement in my system.
I received Windom today on SD card from PS Audio.
Since the cool kids are all doing it, I first went back to an older OS (Torreys was what came out of the hat for me) and then to Windom and then rebooted with the card out twice and let it spin with a Walter Davis Japanese Blue Note cd for a few hours.
Then I sat down to listen. I hear the treble brightness that mark-d does, for me it’s okay–I welcomed a bit more treble when I went from Snowmass 3 to 3.06, and I can adjust in my system via riding the gain between components and with my tube-powered EQ component. To my ears it stands out because there’s a diminution of a slight upper mid/lower treble congestion that I didn’t realize was there til it wasn’t. I hear different texture in the bass, it seems a bit leaner but is also giving me a sensation of LOWER bass. It seems a bit leaner in the midrange (no richness) but more open as well, giving me a slightly clearer look into the soundstage.
I like this, it’s not fatiguing. I also know that it will sound better/different tomorrow, and better/different Thursday and Friday. That’s how these OS changes work in my system. And when it settles in a bit more I can start tweaking round it. That’s become fun for me.
Thanks to Ted and the beta testing team!
6 Likes
Hi Ted, this is exactly as i hear it.
The strange thing is, 40 years ago listening to my lp’s on a much cheaper stereo it all sounded just as real as now (without the definition, the space, the detail etc.) but then there was plenty of background noise. How is that possible ? I suspect the jitter is the biggest culprit ?
At last Windom makes my cd’s with old recordings sound reminiscent of my vinyl from long ago.
This is unique, because i bought remaster after remaster and it never sounded as the original lp’s i once had. My very first cd player, a Yamaha natural sound player from 1985 (with R2R converter) was the most reminiscent of vinyl, until now the DSD with Windom.
Thank you very much for making this possible !
@tedsmith, +1 (866)
The problem must be in one or more parameter values, not the code itself because the update finished correctly.
So only the values of (all) the parameters from the 2 machines have to be examined / debugged.
I love the updates ! But as it is it’s very difficult to know if your update was 100% correct.
Maybe it’s possible that the update process for the next Mountain-top initializes all parameters with the desired values from scratch ?