@Steven_Klemow OK this is pretty cool, if difficult (almost impossible) to discover. So it looks like they are running a mini web server and still keeping the processing to a minimum.
So currently I am playing one of my few 24/192 albums through SPDIF with no noticeable breakup at all. I think I might leave it like this for a while.
44100 x 512 x 2 = 45.1584 Mbps
768000 x 32 x 2 = 49.152 Mbps
Assuming you want stereo…
There may be clipping involved. You can turn on headroom management and set the threshold to zero. This will keep the bit perfect stream, but the signal icon will flash red when it clips…
Already have that on at -3 or something.
But it worked great and no micro breakups like I normally hear.
Switched back to USB and DSD later.
It’s a pity nobody knew about this during beta testing
It defaults at -3, but that introduces digital processing into the signal path and degrades the bit perfect stream to some extent, like all volume leveling does. If it’s necessary due to hardware limitations or to prevent hearing damage, then that’s what it’s there for. But if you set it to zero, it doesn’t disrupt the bit perfect signal—it merely alerts you when clipping occurs. It’s an elegant feature on Roon’s part.
You probably know this already, but when there are no DSPs introduced into the signal path, Roon will indicate “Lossless” at the top of the signal path. Otherwise it will say “enhanced” when DSPs are active or “low quality” if the source is an MP3. Optimally maintaining the lowest number of DSPs is best to get the least degradation or errors—the default headroom management volume leveling is one that is in most cases an unnecessary one. Just my take on things!
So happen i own 2 Farad and ZS, and saw the argument here, so i did a ABA test by myself. Same AC DC cable, fuse, placment location, just swapping the LPS unit.
My finding is:
12v is standard good SQ
15v sounds better to me, relatively beauify everything, emotionally dig deeper. In contrast, 12v is relatively very slightly flattened.
I finally made the leap and updated to v2.31.7 this morning. Rather than repeating myself, you can read my conclusions here.
Let me know if you have any questions. I retract most of the mean things I said about this update a few posts earlier. Apart from the DSD512 thing, missing information in the signal path, and lack of public documentation, this update seems to be perfect.
Could you set Volume to Software and check whether this setting is still persisted after switching DAC off and on again?
That was the showstopper for me - but I tested it in AIO mode.
I keep my DAC on all the time so this wouldn’t affect me, but I have encountered issues in the past with some of the ZS’s settings reverting back to defaults when I disconnected my DAC—so I wouldn’t be surprised.
I’ve noticed that the DSD Direct setting reverts to DOP when I disconnect the DAC, and frustratingly the ZS GUI won’t let me save the setting at native unless I reboot the ZS altogether. I’ve made iFi aware of this.
However, now that it’s a full-on Roon Ready endpoint, it is Roon that detects the capabilities of the ZS and automatically populates them, whereas before when it was a bridge, Roon would detect the capabilities of the attached DAC. I’ll go back now and see if the issue still persists.
Thanks David, I saw this on Head-Fi a bit ago and appreciate it.
One thing that still persists is the vestiges of the Roon Bridge in the Roon interface, as indicated in the Audio Settings attached.
There are other bits of the bridge remaining in the signal path. The two other photos show the signal path if I identify the device as the ZS when prompted in the audio setup, vs when I leave the device unidentified and type in my Topping DAC. In the former case the ZS shows up in the signal path twice—I presume once as a bridge and again as an endpoint. With it unidentified the ZS only shows up once.
The bridge in the ZS is hard encoded, and I think iFi did this as a place holder until they could get certified as Roon Ready. Other Roon Ready endpoints don’t ask you to identify the device attached—it’s automatic. So I wonder whether it’s the bridge architecture that causes the USB Output variation you mentioned in your review? (Note the photos are in backwards order, sorry…)
Yes I know but I listen to a lot of black metal and symphonic metal and I find clipping without the added headed.
Sure, makes sense. I compensate for that by lowering the SDM gain adjustment in the sample rate conversion section, since I listen to those genres too and the only time I get clipping is with upsampling—when on lossless there’s never clipping and there shouldn’t be. The downside is that my upsampled tracks are occasionally a little quieter than the others, but the sound is good.
But I have no data to support that one strategy is any more deleterious than the other, other than the SDM adjustment being one DSP vs two. But from what I understand Roon’s 64 bit float volume leveling is minimally-destructive.
Frustrating thing is that I can’t get the clipping to stop for my DSF files when they’re upsampled. I created them from SACD ISO files and I’m worried I set the gain too high in JRiver. And we’re talking 250+ albums—I shudder at the thought of going back through that ordeal again Maybe I’ll try your volume leveling solution and see if that fixes it, thanks!
Good thinking I will give it a try.
Currently it works well for me
When you’re applying upsampling or pretty much any kind of DSP to a signal, you need to add some headroom to avoid hard clipping. In the PCM world, this is to avoid intersample peaks in tracks that have been normalized by the mastering engineer to 0 dB. For DSD, it has to do with the Delta-Sigma converter and related processing.
Take heart though…once you’ve decided to upsample, you’ve already crossed over into the non-bit-perfect world. At that point, there’s no need to be concerned about maintaining partial bit-perfectness. Just do what you must to achieve the results you’re after. Roon’s 64-bit DSP engine is among the best in the business, so you don’t have anything to worry about.
Also, remember that an “Enhanced” signal path can be better than “Lossless”. That’s why it’s called Enhanced!
Thanks David, I’ll try it—I only apply it to my DSF 64 tracks to try to reduce quantization noise anyway. Most of my DSF tracks are sampled from ISO at 256 so noise shouldn’t be much of an issue there.
Did you happen to get a chance to check out the reply I posted to you above about the output path?
Since the ZEN Stream is really more of a Bridge device than an Audio device, I think it’s appropriate to leave the Audio device set to “Unidentified” for now. Perhaps at some point, Topping will submit some of their products for “Roon Tested” certification. At that point, both the Bridge and Audio devices can be identified (probably how this works already for many iFi DACs).
I wanted to take a moment to announce you all are bad influences.
I picked one up to take the place of a NUC. It will arrive soon.
Haha. Congrats! Which DAC(s) will you pair with your shiny new ZS?
The SMSL SU9 for now. It may end up driving my headphones and that is a Topping D50s. The STAX amp having an RCA passthrough lets me daisy chain two amps from one DAC. it’s kinda neat IMO. My current thinking though is the ZS will replace an I7 NUC I’m using solely as a HQPlayer NAA endpoint.
I’ve not tested the HQPlayer NAA functionality. In this case, you won’t benefit from the Roon Ready firmware, but I’ve not heard any problems with NAA, so you should be fine.
I would suggest giving Roon RAAT a try as well to see which you prefer. Roon’s DSP has seen several improvements over the last year.