the biggest criticism for the DSD, since day 1 and also in this review with Yale, seems to be a relative lack of impact and ultimate resolution esp at the frequency extremes. some call it soft or polite sounding but many, including myself, agree that the DSD is otherwise reference caliber stuff.
personally, i am not really that hung up on this softness issue but finally found the two words that fully describe both my criticism of the DSD and expectations for the next release: bass clarinet.
this instrument has an awesome tonal & dynamic range and can attack like crazy. for some reason that only Ted can know, the DSD rendering of this instrument is a full embodiment off all negative opinions that the DSD garnered thus far.
there is no other instrument, to my knowledge, that the DSD struggles more to replicate. play, for instance, john surman - portrait of a romantic or ginger baker - jesus, i just want to go to sleep and listen to them again from the Rossini.
Solve this Ted and we all go home and never bother you again.
A good while ago I dated a young woman who played bass clarinet. I am thus extremely familiar with the sound. It is a unique animal, but I never thought to use it as a reference.
The sound of a clarinet is tricky to reproduce. As a closed cylindrical tube it produces square waves (only odd-numbered harmonics, each with an amplitude of 1/n where n is the number of the harmonic). As we all know, a square wave is difficult to reproduce, even with a limited number of harmonics present. (The wave becomes more square as more odd numbered harmonics are added. With an infinite number, the wave is perfectly square.)
Related trivia: A flute is a cylindrical tube open at both ends and produces a wave which contains all harmonics.
After listening a while to the final Torreys i feel i do miss the power and push in the sound.
it’s all to polite not live.
(I did compare the sound to my beloved Audio Research cdl)
it’s gettin better after reloading Torreys B.
Am i the only one in this?
I also preferred Torreys B but ended up using torreys final.
I am using bridge 2. After adding optical isolation (powered by lps) to network and closest switch powered by hdplex lps, it made Torreys final sound more live. After that I upgraded ds dac fuses to Audio Horizons platinum reference. For my surprise sound even got better for my ears. Polite (thin?) sound of Torreys final even more shifted towards Torreys b!? Fuses seems to matter. Then last but not least hdplex lps powers now Uptone’s LPS-1 ultracap lps which then powers switch closest to ds dac. This last change made very noticeable positive change to overall SQ. Much better transients, timing, soundstage, bass texture, punch - everything. They claim lps-1 eliminates/blocks completely EMI and RFI.
So as a conclusion IMHO ds dac needs “clean” bits and less there is noise, EMI and RFI, better is SQ. So good was the improvement of adding lps-1, I have now ordered P5 regenerator to get better juice to ds dac and Hegel h-20 power amp. It is interesting hear the impact of p5.
It should be bridge’s task but IMHO unless the bridge connection is not optical, it’s not 100% isolated. With tweaks I did, it can be minimized the losses of purity of digital signal. Please Note that bits arrive as 100% correct in all cases. Information (bits, ip data) is transferred as ip traffic on OSI layer three. We can think that EMI/RFI/noise is transferred on OSI layer one (physical layer)
so if connection is optical there is only left noise generated by bridge when it processes received ip packets. Someone has mentioned it as a “ip packet noise”. This generated “packet noise” can make e.g. UPnP and Roon audio sound little bit different, because they has been implemented differently (different protocol etc.) by different code.
So as a summary IMHO same firmware can (and will) sound different on different setups. The degree of difference of course varies.
I have two main switches. First one is hp procurve 24-port switch. All house pcs, wlan stations, nas(es), firewall, internet 4G modem are connected to that. All those equipments (execpt pcs and won stations) are powered by double conversion ups. On hp switch there is hp’s optical connection module (standard SFP module component) which is used to connect hp switch to tp-link 24-port switch which has only optically/fiber connected equipments: two tp-link mc200cm media converters (copper2fibre) and hp procurve, my music server is intel i5 nuc with centos 64-bit OS and Roon server running. Music files are in one NAS and music folder of NAS is mounted to intel nuc. Intel nuc connects to one tp-link media converter. The other tp-link media converter connects via house cat6 network living room’s switch. Before the switch there one galvanic Ethernet isolator (giso isolator). Switch connects still to one more isolator (optolinedriver OLD6000). Optolinedriver connects to bridge 2.
tp-link media converters are powered by LPS. One hdplex and another brand I don’t remember (next to intel nuc)
optolinedriver and closest switch were powered by hdplex LPS. Now there is still installed UpTone’s LPS-1 which powers optolinedriver driver via hdplex LPS and hdplex also powers switch. So dramatic was the improvement after installing LPS-1 that I am planning to get another LPS-1 which then powers switch via hdplex.
I use multi mode fiber connections. I use AQ vodka Ethernet cable to connect optolinedriver to bridge 2.
Described setup may sound an overkill but all changes improved the SQ.
And like said in-revious post I have ordered one p5 to power DS dac and power amp. Control point used is iPAd pro (Roon)
Although I use a galvanic isolation device before network entry to bridge and heard very slight improvements, according to what Paul posted some time ago, the network connection from NAS to bridge alone totally isolates all that happens before that stage. So there should be no influence and no difference if an optical connection is used inbetween or not or whatever is modified before network entry into bridge (assumed all power supplies before bridge are connected to a different power line than the audio components).
There’s no such thing as “totally” isolates. In particular a transformer can do a great job of getting rid of common mode noise and (hence) can provide great isolation of DC offsets, etc. But it still has some parasitic capacitance that lets very high frequencies thru. The thing is that the sharper the corners on the electric signals (e.g. 1Gig nets or even 100Meg nets) the less jitter, but the more very high frequency noise… Optical does a better job of electrical isolation, but the optical transceivers may introduce jitter and / or the power supplies for them may introduce noise in many ways. Which of these noises may affect your particular system more is hard to say. You just have to try different things in your system to really know which works better.
As I have no components before bridge plugged into the same power circuit as the audio stuff, I guess changing power supplies of those non audio components is too much effort compared to the expected outcome for me.
Trying the optical conversion until back to Ethernet again short before bridge would be interesting. Never heard about an optical/Ethernet converter before…maybe too expensive for such a trial, too, but let’s see…
Compared to other improvement potential and measures already taken, it might be just a very small one.
Ted Smith said
There's no such thing as "totally" isolates. In particular a transformer can do a great job of getting rid of common mode noise and (hence) can provide great isolation of DC offsets, etc. But it still has some parasitic capacitance that lets very high frequencies thru. The thing is that the sharper the corners on the electric signals (e.g. 1Gig nets or even 100Meg nets) the less jitter, but the more very high frequency noise... Optical does a better job of electrical isolation, but the optical transceivers may introduce jitter and / or the power supplies for them may introduce noise in many ways. Which of these noises may affect your particular system more is hard to say. You just have to try different things in your system to really know which works better.
Thanks Ted for your comments. My learning by doing expirements with different LPSs has thought me that they make a difference. And DC provided by LPS-1 to optoisolator seems to be very clean ( so positive impact on SQ). They use a bank switching of ultra caps, controlled by fpga. That operational model makes sense to me. While another bank is charged, the other one provides power. Clever idea I must say. By this way there is 100% isolation of ground loops etc, they state.
If optical connection creates more jitter (what exact amount, I don’t know), it could be handled like bridge 2 does - buffering audio and reclocking it (digital lens). IMHO jitter created can’t be that big that it would have an impact on SQ if that’s taken care properly.
if the noise profile (average freq range) generated by e.g. 1gps interface is known, could the noise be filtered out how easily to avoid “pollution” of other system parts inside then dac ?
As I have no components before bridge plugged into the same power circuit as the audio stuff, I guess changing power supplies of those non audio components is too much effort compared to the expected outcome for me.
Trying the optical conversion until back to Ethernet again short before bridge would be interesting. Never heard about an optical/Ethernet converter before…maybe too expensive for such a trial, too, but let’s see…
Compared to other improvement potential and measures already taken, it might be just a very small one.
E.g. Tp-link mc200cm multimode fiber media converters cost about 50€ each. Of course then LPSs cost more (recommended). Optolinedriver OLD6000 I also use cost about 200€ (100mbps Ethernet copper in, copper out. Inside there is optical isolation). Fiber cables are cheap. No audiophile versions availlable…
then if in switch there are free sfp slots and switch needs optical link, optical Ethernet sfp module cost about 50€-100€
You invested considerably in sophisticated power supplies for your network equipment…are they connected to the same as your audio power circuit? This is where I’d expect bigger improvements… but if this is also the main house circuit, there’d be the same need to improve any other power supply used in the house?
If you have separated circuits, it’s really interesting you’re able to improve sound so much by improving network power supplies in a different mains circuit.
cabin said
Thanks Ted for your comments. My learning by doing expirements with different LPSs has thought me that they make a difference. And DC provided by LPS-1 to optoisolator seems to be very clean ( so positive impact on SQ). They use a bank switching of ultra caps, controlled by fpga. That operational model makes sense to me. While another bank is charged, the other one provides power. Clever idea I must say. By this way there is 100% isolation of ground loops etc, they state.
Like I said there's no such think as 100% isolation - if they used relays there's still a parasitic capacitance and if they use solid state switches there are even more parasitics. I'm not saying that switching back and forth is a bad idea I'm just pointing out that it doesn't provide 100% isolation. The devil is in the details and some ways of isolating work better than others.
cabin said
If optical connection creates more jitter (what exact amount, I don't know), it could be handled like bridge 2 does - buffering audio and reclocking it (digital lens). IMHO jitter created can't be that big that it would have an impact on SQ if that's taken care properly.
Optical transceivers shouldn't be too bad, but, for example, typically ECL transceivers will have less jitter than LVDS which is less than CMOS, ... A digital lens also has jitter/noise leakage - the FPGA in the DS has a plesiosynchronous buffer and there's another on the bridge, they both help, but neither is perfect.
cabin said
if the noise profile (average freq range) generated by e.g. 1gps interface is known, could the noise be filtered out how easily to avoid "pollution" of other system parts inside then dac ?
High frequency pollution often affects preamps, amps, and speakers more than DACs - DACs expect a gob of digital noise and should take some pains.
All of this noise and jitter stuff is in principle possible to calculate given enough information about one’s system but in practice one is better just trying various “tried and true” methods of taking incremental steps to clean up one’s system. Every thing matters and no rehabilitation is perfect.
I’m really admiring all this knowledge of removing all those little and bigger stones on the way to good sound, that were taken off the road more or less already!
Seems it’s no less complicated than getting a turntable to life finally
Listened to a great Volodos/Schubert DSD last 20 minutes which showed how good the results are already
