T,
Incredible work man! Do you have a room at Axpona? I will be there.
Hi - You may have posted this earlier, but I didnāt see it. What is your goal with all of these mods? Is this something you are going to offer as an upgrade or a kit? Or are you simply geeking out and having fun?
2 Likes
Hello, no room there. Iāll just be browsing all the stuff 
My friend had to pull out though, so am on the fence about the trip now. Leaning towards go, but weāll see.
The goal is simply to get the sound of this thing as good as we can We have a very high bar to try and reach. After all these years weāve become tired of more of the same basically. Was one of the reasons why we took a chance on the DS DAC to begin with. Different, and therefor maybe we can break out of more of the same as well. We have, thankfully.
Ultimately I think it would be very interesting to let an independent party try it out vs theirs and see what they think. Actually you were one person I had in mind, if interested (and located in the US)
What we did recently has made a huge improvement, to my ears at least. My friend can also hear the ~ shocking performance over cell phone clips so far. Not knowing what would be the result going in- would it ruin it? But clearly hearing the results now. What actually did most of it, hard to say exactly at this point. To be precise I should have done 1 change at a time, but did not.
The tantalum R at the input and base of the feedback node? The dramatic OA circuit changes (drop from MFB, and a perfect step response)? Disconnecting the other 3 closed loop circuit OA from sharing the common output net. Ie do these closed loop circuits maybe āfight and churnā each other some, raising the noise floor? One might consider what 4 power amps might do if connected in parallel for example (donāt try it )ā¦ However yes, the OA do only connect to each other through 150R so probably fineā¦ but still, sorta wondering.
There is a final āphilosophicā approach change Iād like to try before we āfinalizeā the config here and nail down 2 OA/channel, how exactly to drive the transformer, the output wiring, etc. Since weāre down to 1 OA/channel currently for these tests it is relatively straightforward to implement, thankfully. And most of the pcb land is already cleared for it 
Bare with me here please as we ponder out loud a bit on the philosophical side. In this case weāre thinkingā¦ If the OA does the D/A via LPF, it is fundamentally a very high gain closed loop circuit doing the analog conversion output. Therefor it is susceptible to closed loop IMD, which (likely?) raises the noise floor. This is what we still have now; the OA is still doing most of the LPF. Though from what my ears say, it is doing a much better job of it now. Hmm 
But IMD- for example, the large improvement in everything we hear when moving to passive bi-amping; the improvement is generally immediately audible. More clarity, more detail emerges, it all just sounds better top to bottom. This improvement is due to a large reduction in IMD essentially, in the cabling and in the amps. Which we only know is there generally when we finally hear it reduced. Based on my experience yes, we can clearly hear the cabling effect. But larger still seems to be the amps. My system is bi-amped; so is my friendās.
Sooā¦ a way to reduce the IMD factor in this case would be to do the D/A in an open loop circuit. Perhaps even better, a low impedance fully passive simple low pass filter with perfect step response. Which you may recall also circles back to my original posts in this forum 
So what we want to try here is: use the OA to simply buffer the bit stream basically. Then the re-CLKer does not drive the analog section. The OA will muscle the bitstream still digital bits into the filter. By bits, we mean still full 1/0ās with some filtering to round them off slightly and help reomve ultrasonic content, but they still reach full 1 and 0 and level off at each state. Thus the signal is still fully digital as it comes out the OA and feeds into the LPF. Then out of the LPF comes the beautiful analog, provided by an open loop circuit.
Hope that makes some sense. Weāll try it next. Does it improve further on what we have now? Maybe, maybe not. Weāll seeā¦
Hearing what the other DAC does in terms of incredibly low noise floor is what is pushing this deep dive. By the way it also uses an open loop conversion circuit. But being b*tch slapped sonically, then pushing hard to reduce the gap of what is possible is a good thing ultimately, having already brought a large improvement in performance of the DS (to my ears).
I did a quick cell clip playing from Tidal just now. Will link it. You try and decide
T
I am in the US. Pacific Northwest.
I tried to play the clip but it never started. I saw the image and once I hit play, it just spun with no audio.
My question back to you is whatās your intent? Are you going to open up shop and do these mods on a per customer basis? Are you going to publish the mods with parts lists and some level of āhow toā? Or are you simply having fun and this is an exercise in whatās possible?
BTW - I really dig how open you are about these mods, and especially your online ponderings. Itās a good bit how I think and write too.
Pac NW- I love it. Well I think basically weāll pursue where it leads. If others want it enough, we can figure something out. But that isnāt the intent, to try and make a living off it.
I assume most who hear it will want it. But thatās just me, not everyone is after the same things I guess.
The clip should play. Itās just serving from Google Drive. Try again?
T
1 Like
When you get to the point where you feel like youāve gotten to the point that youāve run out of improvements, it will be interesting to learn your thoughts as to where all of this work leads. I can tell you are having fun and that may be all that matters.
Sadly, it doesnāt want to play:
Those who have been following this topic for a long time remember that I changed the output transformer a long time ago, in fact, the first of all those present, and organized a separate power supply for the analog board. And all this time I was very pleased with the result. But itās time to move on)
I decided to follow the example of Turbotk and add additional capacitance in parallel with the resistor R_6, for better noise filtering in the power supply circuit of each of the eight U1s.
More than two weeks of warming up have already passed and I can safely draw conclusions on this modification without fear that the first impression could be wrong.
The result exceeded all my wildest fantasies - installing these capacitors is equivalent to replacing the output transformers with better ones!
There is no point in describing my subjective impressions a lot here, everyone who has replaced transformers can understand what is at stake. My friend also followed my recommendation to add eight capacitors to U1ās power supply and got the same effect. Turbotk, thanks for the great idea.
At the next stage, I plan to replace the AD8132 with the AD8139 and organize a separate power supply for them, since the regular one cannot provide the voracious AD8139 with the necessary current. Thanks to Ted for the recommendations for this mod.
Iāll post the results later.
7 Likes
I clearly understood that your article is about DS. I canāt say that the topic title didnāt help either 
It would be nice if a friend of ours could get his hands on the next software job while these things are ready.
Is this model suitable? I couldnāt find the one you used?
Very nice work Yuri! And thanks for confirming my impressions as well. It was a big improvement.
Just got back from Axpona audio show in Chicago. It was fun!
Thanks, T
1 Like
Iād recommend you use only the BP (Bi-polar) ES series for this duty. Again BP do sound better in general than polarized.
Or maybe even better, the Kasia BPā¦ Kaisai are the best sounding electrolytic that I know of, generally, but the Muse are close.
But we know (confirmed) the Muse BP on Vocm makes a dramatic improvement here. Highly recommended.
Note that I also had provided a PPS film bypass across each Vocm R as well. Not sure that Yuri did. The Vocm node does already have 1 or 2 uF of high speed ceramic bypass nearby I think, so the additional PPS may not be required.
47uF 16v BP ES works: UES1C470MPM: 47uF 16V Nichicon ES type | HIFICollective
Example Kaisai: CAP-100-R-50U-16V-NP: 50uF 16V Audio Note Kaisei NON-POLAR Electrolytic Capacitor | HIFICollective
I have ordered many thousands of $$ in parts from HFC over the years (Nick would confirm that I have bothered him many times lol , even though they are in the UK and Iām in the US.
Iād think you could get a great result with a film as well, but getting ~ 50uF to fit is a challenge.
T
2 Likes
There are also those that I used:
2 Likes
16v are smaller, which can make it less difficult to install. And probably less prone to being knocked loose later (less leverage on the solder joints).
It might even be better to use the 6.3v in this case, due to the above.
Plus an additional comment. Just my own experience. Confirmed as well at least by my high end friend who also has a DS DAC, his DAC currently has my XS4400 in it. He also has more $ than me (lol) (not hard to do sigh) and has had Chord Dave (and Qtest and Hugo), MSB Analog, Naim streamer, etc, etcā¦
IME the M6 steel transformers sort of limit what you can hear, imposing a bit of āsamenessā effect to the sound, veiling it somewhat. IMO we must break out of that shell to achieve the satisfaction we seek.
The mu metal or amorphous are more transparent in comparison (otherwise, why bother?), allowing you to hear more of what the Vocm mod actually brings for example.
We should also thank said friend Dave in a way, as heās the one who dropped the life changing DAC into my lap a few weeks ago (which I have also bought btw). It was this sonic b-tch slap wake up call basically that we still had far more work to do, which resulted in the deep dive push that resulted in the OA work that resulted in the DS now a life changer as well, already Even in just 1x test mode so far. Iām still buzzing from what its doing now. This is truly a breakthrough moment IMO.
And was mentally comparing to what I was hearing from the 100 or so(?) high end systems I got to hear in an exhausting but great weekend of Axpona. Whew. That all really helps provide a basis too.
Many of those systems were limited by their sourcesā¦ Stridency in the vocals for example is one of my pet peeves. I internally call it āwince momentsā, where you almost need to turn or duck your head momentarily. And human vocals, female especially is absolutely where it must be minimized IMO. Ackā¦ It was present in the majority of systems to varying degrees, but not quite all (bravo). And not always strictly due to the source either, of course.
I should keep my mouth shut here but I will use this as a particular interesting example. Linkwitz was a true genius IMO. In my open baffle phase years ago I was very close to pulling the trigger on a set of Orion. But I could not get past the fact that he also rejected that components, materials, cabling, etc have effects on sound. Thomas Dolby was prescient, blinded by Science! I guess sighā¦ Happens alotā¦ Go look at Science audio reviews site (or whatever its called) for exampleā¦ OMG, no, do NOT go there GAGGGG(!) Diabolical narcissism 100% on display there. And you are strictly required to drink the kool aid or else be scorned.
BUT that is why I did not buy those Orion, knowing that he was wrong. Glad I did not.
So fast fwd to this weekend. There in a room were the current gen Linkwitz speakers. I know they are extremely capable fundamentally. But there they are, knee capped by the data sheet type electronics modules, driven by $30 cabling, etc. And what did I do during the ~ 1 song I managed to get through- wince, wince, ouch, wince, sighhhā¦
And the young gentleman became visibly angry, turning red when asked about the cheap cables for example lol. Yes, a true disciple of SLā¦ Sigh. See ya later pal. Come back when your mind awakens.
Iāll add that it was a real pleasure to chat with many of the fine reps/people who were here from across the pond. I know it is their job to be friendly warm and all, but was also refreshing for me.
T
2 Likes
You know the famous review. Dave and DS were unable to gain the upper hand.
Of course, this review was before the DS mods and software update.
The tables are turned now, at least with mine It would kick a Dave *** all up and down the streetā¦ Thereās no more softened transients, etc, theyāre sounding ~ perfect to me in fact. That wasnāt the case before at all.
Listen to the percussion blasts in the vid clip I linkedā¦ Howās it going to be rendered much better than that, I ask(?) If you know, please let me know as well.
T
Going forward from here, weāre going to swap to the pure Cu amorphous transformers and see if we can hear anything there vs the regular Cu amorphous.
Then weāre going to try the OA as bitstream buffer into passive LPF approach and see how that sounds compared to what we have currently.
Then weāll try and determine if a separate transformer primary coil drive per OA sounds any better than parallel primary drive. Weāll have to cut the OA output loose from the common output net to do so. This will also allow us to use wire of choice from OA to transformer.
Also ultimately need to get the analog section shielding back in place as well, and the local PSU stuff. After we nail down the config.
T