Noted and as soon as I get home and onto a computer I’ll do that - whatever “that” is!
[EDIT: done, I think]
Apologies. I have owned dCS kit for a few years and this is my first foray into earning anything at all out of something which has been a personal passion for decades. I’ll work out whether I just need to change my signature or something else.
Most of what I read and post here will have nothing to do with anything commercial but I absolutely get the principle of being clear about my interests.
I don’t disagree with your findings or most of what you’ve written in this post. And while I agree it’s difficult to test many streamers, I do disagree with the concluding generalisation based on measuring just that one Streamer. I would have expected some clear caveats in that conclusion.
Speaking of that one streamer, while I have no direct experience with the Metrum Amber, it doesn’t appear to be particularly well built. As I posted, it seems to have rather cheap RF45 Jacks with integrated magnetics, and what looks like an internal unshielded Ethernet cable - connecting the input board to a Raspberry Pi compute module - running right next to a transformer! Seriously??! (Your Wavecrest is a sledgehammer trying to kill the Metrum mosquito …Perhaps I shouldn’t be surprised at all of your measured results )
Kudos to you/AA’s spirit of investigation, clearly you’ve spent a lot of time, effort, and money!
However, it appears as if AA picked a rather poor product, measured it, and then extrapolated/generalised your findings as applicable to all products, not to mention the complete audio chain. While completely avoiding the primary question of objectively testing whether sound quality is affected - which can only be answered by measuring the analog output of an attached DAC and cross-referencing/null-testing against the various Ethernet Switches.
The Ambre is not a poor streamer. It uses Tent-clocks and Metrum used optocouplers in the critical parts of the streamer. So, for example - the Pi is not directly connected to the in-/output.
Is it the best streamer in the marked? Of course not. But it is an interesting price-class considering my audience. Not everyone can affort a dCS, MSB, Naim 555, et cetera. I also firmly believe that this Ambre is better designed than a lot of streamers in its price class.
By the way: what I tried to show is that there are actual differences in switches and that it can actually influence your sound. Well… I think I did show that. And again: I never said it’s the same for all devices. I think we agree on that.
About measuring at the output of a dac. I tried. It is not that simple, because you need an A/D-converter and ALL converters ‘drift’ a little bit. So, the digitized files will never be exactly the same. I used a Grimm UC1 (that’s a serious A/D-converter with VERY good clocks) and still my test recordings with the same set-up - to see what happens - couldn’t be ‘nullified’ (don’t know the English word for that). So, when you load two (identical) tracks and put track B 180 degrees out of phase to create absolute silence, after a couple of seconds, you hear that B drifts away from A. It doesn’t work. Even if you pick 10 second samples, it is useless, for it is unreliable.
There have been fairly consistent demands on the forum for anyone claiming there are sonic differences amongst network switches (or even that a network switch can make any sonic difference at all) to “prove” it. Jaap has tried his absolute best to set up a comparative blind test which is far closer than most of us will ever get to demonstrating that switches can make a difference and different switches can make a different difference…
… I guess we can keep asserting that until we measure stuff it doesn’t count, but it would be nice to also cut the guys some slack here. They spent hours and hours on the comparison, typically focused on the more affordable and generic switches rather than exotica, and used an affordable DAC. No experiment will be perfect.
I think the debate is healthy and I actually enjoy trying to understand the various points of view. However, the ultimate question for this forum isn’t do quality switches make a difference but rather…
do boutique switches provide a sonic improvement when used with dCS equipment (that already feature a well designed and isolated steaming board)
no brainer for me: the impact - even with dCS- is huge! My dealer tested yesterday Melco switch with SFP and new Melco top-model with SFP connected to Bartok (last line Ethernet cable). He said this was incredible good. They started with all Ethernet copper, then SFP…
add on:
I’ll try next week an Ansuz switch again (some folks here got it up&running)- this gave me the most 3D stage and details I ever had in my system
I’m trying to understand the relative magnitude of what you hear Ralf. Not disputing it; trying to understand it. IIRC, yours is a Bartok stack? Or am I misremembering? If I am, my apologies. Also, are you using a Melco streamer or something else to feed the DAC? I also seem to recall you’ve got both UPnP and Roon stuff going.
But way of comparison, if someone were to ask me what difference my Clock makes in my Vivaldi stack, I would characterize it as surprising in how discernible it is, but still subtle. I was more surprised the first time I heard the Rossini Clock on the Bartok over three years ago; the revelation to me was that it could make such an audible difference. Is it better? To me it is, more than enough to justify the purchase price. But compared to the difference between Rossini DAC and Vivaldi DAC, it is of significantly lower magnitude. And I am much less surprised by it now, than I was then, when I was actually startled. So, for you, is the choice of switch and PSU of greater or lesser impact than the choice of server/streamer or say between your previous DAC and Bartok? When you think back to other revelations, which ones stand out similarly?
So, when you say “huge,” can you put that in some context so that folks like me have some sense of actual magnitude? Thanks!
This dealer gave me first a Bartok for test, later on he installed my dCS Rossini and Clock. It’s difficult to say what’s a huge step for me is vs a big step for one other…I would describe the improvement of SQ like the difference btw Redbook vs highres with a good switch.
No experiment will be perfect? Maybe not. But the very nature of experimentation means that it is subject to replication, questions, etc. That’s science! Why should anyone be cut slack here? I appreciate Jaap’s work and have found it educational, and I also appreciate his civil participation here. But without asking questions, how does a sensate human know whether to accept the work at face value or to look deeper? Sorry, Jaap is doing (apparently) what he does for a living. That doesn’t get someone a break. And here, we all offer up opinions and experience, some of it gained/gleaned from hundreds of hours of testing and listening. Why should any of us be “cut some slack”? If we offer assessments, they should be willingly subject to civil scrutiny.
I wasn’t suggesting we end the discussion there: perish the thought!
I have a dCS DAC though to your point my streamer (heavily pimped though it is) may not be of dCS quality. Even a budget switch makes a clearly audible improvement here. So then we’re into some vs most vs all and working out which factors have most influence.
The slack I referred to was a plea to build on what Jaap has done, acknowledge that like all experiments it has its limitations, rather than picking holes in it by suggesting there are aspects of it which might invalidate his findings. That’s all.
Yes, I understand the intent, but the fact is you didn’t show that sound quality was affected, only that the choice of Ethernet Switch affects the jitter/phase-noise of the SPDIF output of the Metrum streamer. That’s all. Whether that then has an actual audible impact on sound quality out of a DAC was not done objectively. The reader has to take your word for it.
Perhaps you’re right that at that price point that’s the best they can do. My only experience with a pure digital streamer is the dCS Network Bridge (which I still own).
Yes, and I know for a fact that it’s not easy because I have a TASCAM DA3000 that’s permanently connected to my Vivaldi stack (Clocked to the Vivaldi, and taking the DAC’s RCA output for ADC). And I use a combination of DeltaWave/DiffMaker/AdobeAudition to align captured PCM samples.
However, there’s a simpler solution to measuring the DAC’s output in your test case;
If indeed an Ethernet Switch is causing enough SPDIF jitter/phase-noise to be audibly different out of a DAC, then you will see that jitter/phase-noise manifest as increased distortion measurement (corresponding to the relevant Ethernet Switch) of a single (or multi-tone) frequency response test out of the DAC. No PCM sample alignment problems to deal with!
Just out of curiosity, which Ethernet switch(es) are you using for audio streaming to your Vivaldi and Bartok stacks? Can you share your Ethernet infrastructure, copper, fiber etc?
Well… I haven’t had the chance to use a dCS for this experiment. I do get the Bartok Apex and a normal one soon. After I measured them and listened to them, I will ask the importer if I may use it for further experiments. I will need to open them up and probe them… so I actually don’t think they will like that.
Ow… I think the discussion is very nice and respectful. Not like ASR (Audio Science Review) where they immediately mock you for even trying. I tried to reply there, but eventually I just stopped, because they just don’t want to even keep an open mind.
…Perhaps I shouldn’t be surprised at all of your measured results 
)
the impact - even with dCS- is huge! My dealer tested yesterday Melco switch with SFP and new Melco top-model with SFP connected to Bartok (last line Ethernet cable). He said this was incredible good. They started with all Ethernet copper, then SFP…