And note that there are variations in the “endpoint” - choices of both hardware and software.
So it’s a good bet that at least some of those variations will sound better than the Bridge.
Almost any endpoint will beat the Bridge II for sound quality. If you have a Bridge II your best bet it to take it out of your DirectStream and sell it. Then buy a good endpoint like an ultraRendu and get much better sound quality with support for higher PCM and DSD rates. And that does not even factor in being able to use HQPlayer…
The nucleus connects directly to a day via usb as we]l as Ethernet!
Roon servers optimally connect over network to an endpoint. This minimizes noise injected into the dac. I employ an opticalrendu for streaming from my i7 sonictransport. Get the nucleus out of your music room and stream- you will notice the diff. Using HQPlayer only ups the result.
I have my nucleus in another room and connect by Ethernet. My point is that I have compared it to usb and it is better sonically. I have compared this to my vinyl setup with LP’s of the same songs and it is very, very close. Hard to imagine that improving dramatically with there other devices.
I had Roon running Ethernet to the Bridge 2 (which is an endpoint) and thought it sounded great, until this week, when I added the Matrix and a Mac Mini endpoint external to the DSD. I’m currently using Roon Bridge, but will give HQP a try this week.
I was wrong. My system sounds Mucho greater now, in many ways. The Matrix device is indeed a dramatic improvement - well worth the investment. And HQP still to come…
Not buying this dramatic improvement. I would need in home demo before a purchase for the reasons I already mentioned. I am not swayed unless I hear it with my setup which sounds great as I’d! Glad it solved a sound quality for you.
Be sure and download the latest version of the HQP trial that has the new EC modulators.
Your loss.
BTW, with the Matrix X-SPDIF2 device I’m not hearing much difference between Roon Bridge and HQP (without filters selected in HQP), with one exception - RB plays louder than HQP by several dB.
I’m basing this on setting both programs to max - HQP to 0dB and RB at 100%. I don’t know why this should be, but there it is. I have noticed that setting HQP higher than 0dB results in loud blasts of white noise clipping when the music gets loud.
It’s not recommended to set HQP’s max volume higher than -3dB… the technical reasons are similar to this explanation (clipping):
Intersample Overs in CD Recordings - Benchmark Media Systems
Set HQP volume to -3dB. Or -6dB to be even safer.
Curiously, HQP seemed to default to 0dB after installiation.
Also, cutting it back to -6 like a recording studio would make the lower gain vs. RB even more apparent, and RB doesn’t audibly clip at max gain.
A well mastered Redbook CD won’t clip, but it’s easy to generate 16/44.1k PCM that clips, e.g. many people use a function to normalize gain so that the maximum sample is at or near 0dBFS. Look up “intersample overs” on the web for details.
You don’t need the -6dB in HQP if you only play valid Redbook, but music that isn’t properly bandlimited (e.g. some synthesized music, music that’s carelessly normalized, etc.) will require a little headroom. We discovered by listening that a lot of Dennon SACD/CD players would clip on intersample overs, but that the Sony players we had didn’t. In the DS I have an extra 6dB above 0dBFS to handle these things.
2 Likes
I’m mystified - how does Roon Bridge play louder than HQP without audible clipping?
I have used 0dB in HQPlayer and rarely saw any clipping…in fact, I don’t know that I ever saw any clipping. I have it set to -3dB just in case.
I only saw clipping when I raised the control level to +6dB while experimenting.
As I recall, the maximum possible inter-sample value is +3.01dB in excess of 0 dBFS.
I don’t use HQP so I’m happy to be corrected about it. I’m just talking here about what’s possible and what the DS does, also I don’t know for sure that “volume at max” mean unity volume in all pieces of your signal chain, if any part of the chain allows gain that can cause all the problems we are talking about.
If you are sending 16/44.1 (or any PCM) to the DS the DS will gracefully handle any “illegal” PCM. HQP doesn’t have this option: it can’t go above 0dBFS since its output has to stay in PCM and 0dBFS is the limit for any PCM. DSD allows up to about +4dBFS by spec so having HQP output DSD (I assume that’s an option) would let it handle “illegal” PCM without a volume level change (unless it does something else with DSD volume levels… Not that I’m recommending that. I feel for HQP, sometime you get blamed when you do the “right thing” to handle someone else’s screwup.
It can go higher if you raise the low-end of the control above -60dB.
No, if you input is long enough and your filters are long enough you can represent as high of a peak as you wish. e.g. have a single point at the level you want, bandlimit it to make a valid PCM signal, sample the signal at twice that bandlimit. Now if you lower the sample rate (or otherwise resample) but end up with no samples at that maximum (say it lands exactly between two samples) you’ll have a sample stream that represents a signal whose maximum is arbitrarily higher than any actual sample.
[Edit: well not arbitrarily, but the negative maximum of a normalized sinc signal is much lower than the 1/2 that 3.01dB would represent and you can still have all points but the original maximum be non zero to give more range… I’m not expressing myself well here. But a look at the sinc function while imagining sampling that doesn’t happen to hit the peak should clarify things: Sinc function - Wikipedia]
Thanks, Ted.
This sounds like a lot of work. 
But I understand your point.
My 3.01dB number comes from years ago when studying digital recording. This is the maximum inter-sample signal level I recall one can see if you normalize a hot 44/16 signal to 0dBFS. Is this incorrect?