Best source for Directstream DAC - latest thinking?

I agree. Having the Oppo work with I2S would be ideal. I’m waiting for the upcoming PSA streamer, while pricey, will probably advance the state of the art and be optimal for use with the DS DAC.

I don’t have a microRendu, but I do have a pretty good Mac Mini sytem. Up until recently I had a solid USB chain including LANRover and nice Nordost USB cables. Then I got the UpTone LPS-1 which actually did tighten things up a little but. Not huge, but a bit. Things were a little more clear and crisp, maybe with a lower noise floor. Then I finally went for the Bridge II and hooked it up to my switch in a standard config. It didn’t sound significantly better, to my ears, than my USB chain. My buddy says he thought it was a little better, but I couldn’t hear it. Then I added optical isolation in between the switch and Mac and the switch and the DAC (powered by LPS-1). Now I thought the B2 sounded a little better than the USB, but not by much. Again, a little more punch and clarity with cymbals more realistic. Then I figured out how to direct connect my Mac Mini to the DS DAC via a “bridged” ethernet connection and upgraded to the Tera Grand Cat7 cables from the Blue Jeans cables Cat6a cables. Now I definitely noticed a significant difference in the B2 sounding better than the USB chain. I think this is mainly due to the direct connection since I didn’t notice a big step backward in putting the Cat6a cable back. For grins I also put in the optical isolation between the Mac and DS DAC, but didn’t notice any difference. From all this I have come to a few conclusions:

  1. USB with LANRover is very, very good. As good, I think, as through a B2 connection through a decent quality switch.

  2. Optical isolation is useful in isolating the DAC and PC/Mac from the switch. I got no significant benefit in isolating between the Mac and the DS DAC using direct ethernet connection, however.

  3. Cat7 cables can be a significant benefit, especially if EMI/RFI are a concern behind your rack, like mine. Behind my rack is scary!

  4. Direct, bridged ethernet connection is the ideal connection to the B2 from a Mac/PC. There are pages and pages of testimonials on this on computeraudiophile.com, mainly with micruRendu and SoTM sMS-200. It’s not the easiest in the world, but it can be done.

So what is my point? I think that if you optimize your B2 connection with good cables, optical isolation (+LPS), and direct connection, you will be 95% of the way to the best SQ short of using the I2S. I believe my setup would be competitive with an optimized microRendu with direct connection, FWIW. For that last 5%, I’m going to try the PSA server, Octave, with I2S. That’s been my experience.

Adam

Paul McGowan said

It can’t work in the DS because Oppo does not support I2S format and we do not support HDMI format.


Is it even possible for the Oppo to output DSD through the HDMI like they say?And why would the Bryston BDA-3 accept DSD over HDMI? That doesn’t make a whole lot of sense to me. But what do I know? sorry_gif

There are multiple levels of HDMI support. DSD was added a little ways in and there were many HDMI interface chips built before it was introduced it so there are plenty of HDMI devices that don’t support it (and plenty that do.) The HDMI DSD protocol includes the required security for the SACD licenses. Even getting an HDMI chip that supports DSD for a prototype development board involves licensing and a private key…

amgradmd said

So what is my point? I think that if you optimize your B2 connection with good cables, optical isolation (+LPS), and direct connection, you will be 95% of the way to the best SQ short of using the I2S. I believe my setup would be competitive with an optimized microRendu with direct connection, FWIW. For that last 5%, I’m going to try the PSA server, Octave, with I2S. That’s been my experience.

Adam


Hi Adam,

Thanks for the very detailed response.

I’m following that “bridged” thread on CA quite closely and am intrigued to say the least.

My QNAP HS-251+ NAS has two ethernet ports and as cited by other respondents on that thread seems ideally suited to the bridging concept.

It’s one of the reasons the Microrendu - LPS-1 approach appeals (so I can test for myself).

Can I clarify. When you say “direct connection” above in the context of the Bridge II are you talking about the bridging tweak applied directly to the Bridge? (as opposed to Microrendu).

If so do you think it’d be possible to implement the bridged tweak directly between a my NAS and the Bridge II?

Many Thanks,

Alan

BigAlMc said

I have the Oppo 105D too (as a Blu Ray & Netflix) source primarily but also for my few SACDs - so that would have been the dream solution. Had it been truefrown

Meantime my quest for the ‘optimum’ path into the DS continues. I read the Ted Brodie review of the DS with the Sonore Signature Series Rendu and thought that perhaps this was the answer. However, further investigations showed that Sonore had discontinued this product and Adrian at Sonore (who was really helpful with very detailed emails - great customer service! clapping_gif) said he felt the MicroRendu (with the right power supply) was better than the SSR.

I was trying to avoid USB and eager to go the I2S route so this put a hungry cat in with the worried pigeons!

Also raises the question of whether the MicroRendu with say the LPS-1 would beat the Bridge II for SQ. Any thoughts here? And apologies if that’s controversial. I love my DS and my Bridge II - just want to make sure I get the most out of it.

So in terms of that elusive I2S source does that mean it’s down to the Pink Faun, waiting for the PS Audio server or seeing if anything else comes to market with (DS compatible) I2S as it’s main output?

Also feedback/thoughts on the Bryston BD-2 or BD-3?

Many Thanks,

Alan

I'm about to add in a linear power supply to feed my Pink Faun I2S card. I finally receiced everything I need to implement the linear power supply. If you do order a card be sure to tell them you have a PS Audio DAC. I was told that PS Audio does I2S differently in that they switch the left and right channels. Apparently this is different from all other HiFi brands who utilize I2S. They will make sure your card is setup properly when they manufacture your card. Love this card to death. It sounds wonderful.

Hi Alan. I was also, obviously, interested in the CA thread, although it has long outlived it’s usefulness. It goes on and on far off base from the original purpose, IMO. It’s exhausting!

Yes, I am talking about the bridging tweak applied directly to the B2. It works great! I don’t see why you wouldn’t be able to make the QNAP work with the B2. I went into detail how I made it work with Mac on post 23 on this link:

http://www.psaudio.com/forum/directstream-all-about-it/bridge-ii-to-mac-mini-via-ethernet/

I realize you have a different setup with your NAS, but the DS DAC setup I used might help. I honestly think that if you could get the connection to B2 to work, the microRendu might be redundant! You could always get the uR and try it and if you don’t notice a difference, sell it. Those get great resale right now, so it wouldn’t be a huge risk. I’d honestly be more interested in the SoTM, if I were you. They allegedly have a slight advantage is SQ. Slight. UpTone LPS-1 will power both. Good luck!

Hi amgradmd,

Thanks for the response and the link on bridging. Food for thought, as I’ll defo give the bridging tweak a go at some point. After all it’s a freebie in terms of feeding the obsession! (well free if you don’t count the time, stress and swearing that are no doubt involved wink)

Unfortunately, I’d already ordered the MR before you recommended the SoTM. That said both Sonore and SoTM have new models coming out in the next couple months that will presumably improve things further.

As per my comments on another post I’m leaning towards testing out the MR (with an LPS-1) and then potentially adding the Singxer SU-1 to convert to I2S. That should keep me busy/happy the remainder of the month. I guess once I’ve decided on the optimal setup in amongst that lot (and who knows - it might be the BII after all that!) then I’ll have a think about where the implement the Bridging tweak.

I’ll also keep an eye on what the reviews and feedback are like on the new MR & SoTM models (compared obviously to what the additional cost is).

Cheers,

Alan

Alan - Keep us apprised on how things go with the MR and the Singxer SU-1. I’m very curious how the MR will sound versus the Bridge II, hopefully with the direct connection tweak to both the MR and Bridge! I’ve come so close to getting both the MR and the sMS-200, but held back since I already had gone down the other path away from USB. That Singxer is an interesting unit, though. I must say, I’m rooting for the Bridge in this fight!!

So the first round results are in - well sort of as it’s pretty much neck and neck.

Disclaimer: I think I lack a cultured palate. Put it this way. I can tell the difference between a £10 bottle of wine and a £5 bottle. But the difference between a £20 bottle and a £10 bottle. Not so much… Kinda the same with audio. I go to these audio shows and most of the rooms sound pretty much the same to me. But then there are a couple rooms where you are like whoaa! But you don’t know if it’s the ten grand speakers, the ten grand amp, or the 10 grand source that’s ultimately pleasing you.

Same thing with my own kit. I tinker with different cables, different PSU’s, etc, etc and most of it sounds just the same. Sorry but it does to me at least. That said occasionally I add something (the DS Dac 77_gif) and hear immediately it’s way improved.

Most the kit that doesn’t make any difference I can actually hear, I tend to keep anyway. Audioquest ethernet cables etc. Because it’s the peace of mind of knowing you at least tried to improve that element. And perhaps after burn in yada yada it will improve. Though don’t get me started on bloody burn-in! A high-tech piece of equipment I understand. But a cable with no moving parts?? Hmm…

Anyway. Set up the MR and the LPS-1 last Thursday. Then came the Easter break which was handy as were away so just put Roon on radio mode and left it playing until today. So perhaps 150+ hours of burn in.

To A-B test it I took my FMC out (so as to not give one route an unfair advantage) and set it up as follows:

NAS to switch with Cat7 ethernet 1m sheilded. Switch powered by floating battery power supply.

Switch via AQ cinnamon ethernet 0.75m to MR. MR powered by LPS-1. AQ Carbon 1m USB to DS DAC.

Switch via AQ cinnamon ethernet 1m to DS DAC (Bridge II).

Set Roon to play the same CD (ripped to FLAC) to both the MR endpoint and the Bridge II endpoint.

Turned the volume up a bit mmm_gif and flipped between the USB and BII inputs on the DS every 30 seconds or so. After 30 mins I think it’s too close to call (for my uncultured ears anyway) as both sounded great.

Kudos to both Sonore and PSA for each endpoints great SQ!

On pure SQ I think (but am not certain) that the BII edged it very, very slightly. But that may be expectation bias as I kinda wanted the BII to win.

Factor in cover art on the BII route. And the fact it’s a less expensive, more direct route. And I think the BII wins this round.

Have re-implemented the FMCs for fibre optic isolation in the Switch to BII path and am listening to that for now. Left the MR running tho so it keeps burning in.

I ordered the Singxer last weekend and it’s due any day. So the remaining question is whether the Switch>MR>SU-1>I2S path can better the BII path. Will report back in once I’ve had a chance to test (and done the bloody burn-in!! dull_gif)

And sorry - this post is a lot longer than I intended…

Cheers,

Alan

Bit of an update.

So the Singxer SU-1 arrived on Saturday and I set it up between the MR and the DS DAC. Immediately out of the box with no burn-in whatsoever I was really liking the sound. Left it playing overnight and on Sunday afternoon I tried comparing it to BII. I had planned on letting it burn-in but was too impatient as to whether I really was hearing an improvement or not.

Set up comparison similar to the way I compared the MR to BII:

NAS to switch with Cat7 ethernet 1m sheilded. Switch powered by floating battery power supply.

Switch via AQ cinnamon ethernet 0.75m to MR. MR powered by LPS-1. AQ Carbon 1m USB to SU-1. SU-1 via AQ cinnamon 1m HDMI to I2S input on the DS DAC.

Switch via AQ cinnamon ethernet 1m to DS DAC (Bridge II).

The difference wasn’t night and day. [But then it never is! sad-029_gif]. But listening to the same music sent to the MR and BII endpoints I felt there was something more in the SU-1 path. More what is hard to express and I’m nervous about getting over descriptive or carried away as it was pretty subtle. But the SU-1 to I2S path seemed to contain a little more breath in the vocals and the instruments seemed slightly ‘warmer’ I guess.

Like I said. It was pretty subtle and as per my disclaimer in the previous post I’m not sure I have the most expert ears. But I heard enough to add the FMC’s back in but between the Switch to the MR and am listening to this path rather than the BII path.

I think in fairness to the BII I’m going to try and A-B it again after the SU-1 has 150+ hours or so on it. Just to be certain before I decide whether to stick to the (let’s face it) pretty elaborate path to get I2S into the DS. I’m still a little wary of all those conversions as seems counter-intuitive.

Also if this NAS>MR>SU-1>DS DAC I2S path really is better then that has me wondering a couple of things:

  1. What would the SU-1 be capable of with a better PSU? (There are some discussions on CA with people reporting favourably on modding it to replace the inbuilt linear PSU).
  2. What would a server/solution that has I2S at source (Pink Faun, W4S, the PSA server when it materializes etc) be capable of?
The nervosa clearly isn't satisfied quite yet despite the recent adventures.....shaking-head-no-smiley-emoticon_gif

Cheers,

Alan

Thanks for the great writeup, Alan! I must say I’m curious about the Singxer unit myself. I might have tried it before I got the Bridge and have optimized it with the direct “bridged” connection, but I’m pretty happy with the sound from that. It’s definitely more musical than my USB chain with LANRover, which was very very good, in itself. (BTW, I’ll bet you can do a similar setup from your NAS to the MR or B2. You may want to try it!) If your chain with the Singxer unit is better than the bridged direct connection, I doubt it’s by much. I’ve resigned myself to only upgrading in the future by using a direct source I2S unit like the new PSA server, when it comes out. I’m very much looking forward to that one. And, BTW, I applaud your efforts to try something new and contribute to the knowledge base for the community!

Discussion about different file servers, whether computers dedicated to the task, computers not dedicated to the task or networked storage (NAS) leaves me cold.

Some facts (Professional, technical people correct me if I’m wrong.)

  1. Networking as we know it in our homes and across the Internet uses protocols with effective error correction. Application files have to be 'bit perfect': if a single bit is wrong then there's a substantial chance that the code won't run correctly. The fact that most downloads of code work properly first time is proof that for the vast majority of the time the error correction works as intended.
  2. Files streamed from your music server to your player use the same protocols so they almost always arrive bit perfect. So if you hear a difference between one server and another it won't be because of changes in the data (music files) sent. That's not to say that it's impossible for a 1 to come over as a 0 or vice versa, just that it will be very rare.
  3. Electrical noise passes along conductors from one piece of equipment to another. It may originate in the mains supply, the power supply (PSU) of a component (especially if it's switched mode (SMPS) rather than linear), the component itself will (especially if it's digital) or be picked up by audio interconnects (except for optical TosLink) or Ethernet cables.
Don't fret about the bits coming as files over your Ethernet being correct: they will be (see 1 above). Your DirectStream will do a great job of extracting the right bits from USB or SPDIF, though I find more solace in Ethernet than either of those because with Ethernet the music file is turned into digital audio in an audiophile component whilst with USB or SPDIF the music file is turned into digital audio in a computer (or perhaps non-audiophile player).

Now when I read that such-and-such a NAS or computer server sounds different I cannot draw any conclusions that help me improve my system without knowing about the whole chain of devices that got the file from its origin (on a HD or SSD) to the player. What network switch are you using (that includes your router as it will contain a switch) and what kind of power supply does it have, the original, electrically noisy SMPS or a linear PSU? Is your Ethernet cable well screened and does it take a path in parallel with domestic mains cables or past other electrical equipment? As always there are very many variables. So what can I do to get the best out of my system?

I think the answer is to optimise galvanic isolation, to minimise electrical interference coming into my player. I shall start a new thread to explore this topic.

Peter

Definitely there aren’t any bit errors in a nominally functioning system - bit errors aren’t a problem for any reasonable music system. If there’s any worry at all, just run a bit perfect test.

Unfortunately there are other sources of noise: e.g. groundloops (which have an induced current proportional to their area and the electromagnetic flux thru them.) Balanced cables can help with common mode noise but they aren’t always sufficient.

Also, in a less significant way, noise can become jitter and jitter can become noise relatively easily, however jitter only matters at the final single spot in the DAC where the digital signal combined with the final clock become audio and from then on only noise is the enemy of an audio signal.

A bigger practical problem is finding where the unavoidable noise/jitter in a system really causes audible artifacts: it can be anywhere where there’s an signal (digital or analog) and if you’ve ever chased a hum in your system you’ll know that, at times, there are many surprises awaiting.

I think that many of us overgeneralize our personal experience to other’s systems. I have no doubt that I can take essentially any NAS/computer and make up for it’s shortcomings noise/jitter wise in a given system and conversely I could take an essentially perfect NAS/computer and make most seemingly insignificant changes in it perfectly audible in most given systems.

Following anyone else’s specific recipe to build a great system isn’t reliable: instead you have to have a lot of experience or be prepared to gain experience or hire experience to deal with whatever issues come up in your setting with your budget, room, listening preferences etc.

I really get annoyed when people say things that amount to “your system can’t possibly sound good if you use xxx” or “your system can’t sound good if you don’t use xxx”. There are many valid reasons for particular choices for components of a system, we just have to be prepared to change our minds with experience or to do some “patching” if it’s practical.

[Edit - now that your post has been edited I see that we were headed the same direction :slight_smile: ]

Hi Ted @tedsmith

I understand the following key differences between isochronous and asychronous:

Isochronous/synchronous protocols guarantee bandwidth availability at the cost of reliable transmission

Asynchronous systems will guarantee uncorrupted data

But if I have the Asynchronous USB input of the DirectStream being fed over an Isochronous network (between the USB Dac and the host computer), will the Async USB Dac guarantee uncorrupted data, even with isochronous protocols are being used in the chain?

The chain to give some context:

PC is the music server > PS Audio LANRover over the network (Isochronous) > PS Audio DirectStream’s Asynchronous USB input

Obviously the Asynchronous DirectStream connected directly to the PC would ensure uncorrupted data but does the PS Audio LANRover (which is isochronous but is not asynchronous) compromise the guarantee of uncorrupted data?

Cheers!

You are correct that in theory there’s no such thing as a free lunch: it’s a fact of the universe that you can get guaranteed correct data, but it may take an unknown amount of time or you can get data in a guaranteed time, but it might not be correct.

Still USB and wired networking have to be pretty darned reliable otherwise the number of retries necessary to get correct data would slow the whole system down enough that some other better designed interface(s) would displace it/them. In practice these errors don’t occur often enough to change the character of the sound: run a bit perfect test and see how often it fails. With DSD or DoP you’d never ever hear a single bad bit, but a whole missing packet would be very obvious. How often have you heard them?

Thanks Ted, I get ya. The Bridge II (like the LANRover) is isochronous and not asynchronous, right?

Run the bitperfect test before you get all worked up about any of this:

There are differing jargons for differing protocols and for that matter for USB vs USB audio and things are never as simple as a few words:

For USB proper the main data carrying connections are isochronous and bulk. timely/but not necessarily reliable vs. reliable/but not necessarily timely.

For USB Audio there’s synchronous, asynchronous and adaptive protocols: but note that the audio data is always sent isochronously (i.e. it may arrive late and cause an error.) The terms synchronous, asynchronous and adaptive have to do with whose in control of the USB packet timing and who controls the overall data rate: see http://www.pearlaudiovideo.com/blog/expertise/computer-audio/usb-asychronous-vs-usb-adaptive/

In Ethernet land there are different things like UDP, RDP, etc that deal with the reliability of the data and only recently was the ability to deal with timeliness of data added (AVB, etc.) but since most network switches (until the last few years) don’t support real time data guarantees, in essence when using Ethernet for audio the data can in principle be too late in most audiophile’s setups and, for that matter, the bridge itself doesn’t support AVB, etc. PS Audio looked into it (at some cost) but, at the time, it was too early to rely on it end to end so there was no reason to restrict which kind of switches, etc. that PS Audio customers could use.

All inputs to the FPGA proper have their own clocks and no error correction, the DS’s output clock (and hence DAC function) is plesiochronous: it runs at the same average rate as the active FPGA input and keeps that rate close enough to not exceed the buffer space available to absorb small differences (your standard elastic buffer.)

The I2S, S/PDIF, AES/EBU, TOSLink inputs to the DS are essentially no error correction and have a very small time window for delivery.

The DS’s USB Audio interface runs in asynchronous mode, the DAC’s clock controls the delivery timing of the USB data and the FPGA’s clock is centered (since that’s lower jitter) and the USB processor controls the data rate with the host. The data can, in principle, be unreliable. The USB Audio more doesn’t really matter since the DS will track the incoming clock well anyway and it will dejitter the data, but rather than explain any other USB Audio mode choice we just use what people expect.

The Bridge talks to the FPGA more like an I2S input, but it has it’s own clocks and uses varying high level protocols UPnP, DLNA over Ethernet to get the data - When the bridge is reading the data from a file on the network that process is error free and in principle therefor the data may come late and cause a hiccup…

The LANRover uses network packets to support USB isochronous transfers at the low level and hence is supports USB Audio fully, but once again with the possibility of late data at times.

BUT I repeat: run the bitperfect test before you get all worked up about any of this.

amgradmd said

I must say I’m curious about the Singxer unit myself. I might have tried it before I got the Bridge and have optimized it with the direct “bridged” connection, but I’m pretty happy with the sound from that. It’s definitely more musical than my USB chain with LANRover, which was very very good, in itself. (BTW, I’ll bet you can do a similar setup from your NAS to the MR or B2. You may want to try it!) If your chain with the Singxer unit is better than the bridged direct connection, I doubt it’s by much. I’ve resigned myself to only upgrading in the future by using a direct source I2S unit like the new PSA server, when it comes out.


Hi @amgradmd ,

Thanks for the kind words.

I’m almost certainly going to try the direct “bridged” tweak at some point because it’s:

A. Intriguing

B. Free cool

I’m waiting until I have decided upon the optimal path into the DS (BII, MR, or MR>SU-1). Once I’ve decided that then I’ll need to figure out the best way to Bridge. Presumably to the BII or MR are the options as can’t bridge to the SU-1 via USB. I think anyway.

Famous last words but small-ish tweaks aside I think I’m going to try and quell the nervosa and settle on whatever paths and tweaks I find work the best. In other words I think I’ve ruled out buying anything more high end like the W4S, Nimitra, Melco etc at this stage. Next significant upgrade is on amplification so I’m following the M700 threads closely.

The SQ I’m getting at the moment is pretty great so if the bridging tweak can take it a bit further still then I think I’m good until something comes to market that either has I2S completely solved or is a game changer on another SQ front.

Cheers,

Alan

Peter said

Discussion about different file servers, whether computers dedicated to the task, computers not dedicated to the task or networked storage (NAS) leaves me cold.

Hi Peter,

I’m not that technical and need to admit I barely follow some of what you wrote. But as background I was simply comparing the MR to the BII because of all the stuff I’d read about it and likewise with the SU-1 because I’d read in many places (including the esteemed Ted Smith on this forum) that I2S was the best signal.

My rationale was also that these devices are doing things to try and clean the signal and there might be benefits from that rather than just the whole 1’s and 0’s.

My comment about what a source system that had I2S out of the box might bring to the table was just because I’m concerned about all the conversions taking place from ethernet to USB to I2S etc.

Peter said I think the answer is to optimise galvanic isolation, to minimise electrical interference coming into my player. I shall start a new thread to explore this topic.
</blockquote>

I’ll look forward to this thread and do my best to understand the technical stuff that often goes over my head confused

Thanks,

Alan

Ted Smith said

Run the bitperfect test before you get all worked up about any of this:

Thanks for the explanation Ted. Not sure why or how I come across as ‘being worked up’ with these technical queries - I assure you life is too short to worry or stress about these things :slight_smile: But like you, I am just interested ! Hence my questions.

Bit perfect test runs fine - it’s not a failure that made me interested in asking about this. I’m just a bit geeky like that. Sad I know lol

So back to my original question - since the DirectStream’s USB audio interface runs in asynchronous mode, then even with an isochronous non-asynchronous source like the LANRover in the chain, then the data will still be 100% de-jittered and uncorrupted data is 100% guaranteed to reach the DirectStream?

The potential for late data you mention with the LANRover doesn’t exist when the LANRover is used with the DirectStream Dac’s Asynchronous USB input, right?

Thanks again in advance