”The USB output is also not connected directly to the pi, and instead is ‘reclocked’ (retransmitted) using the RED’s clock and a clean 5V supply.”
”The Holo RED is the lowest jitter and one of the lowest noise digital sources I’ve tested. Absolutely incredible performance that sets the current benchmark for a streamer or DDC. The fact that this level of performance is achieved at under $1000 is honestly incredible.”
And yes, you can use something like Holo Red with your setup. You should not connect the HQP server machine directly to your DAC if you want the best performance. Streaming to a separate low noise and low power endpoint (like Holo Red) is highly recommended. My server machine also runs Win11 and I oversample to DSD512x48 with HQP desktop. The same machine also runs my Roon server. Then I stream over LAN to Holo Red which is connected to the DAC with USB.
If you are connecting the NAA to dac by usb,
It is the dac internal clock that determines the timing/jitter. The red has built in usb isolation that regenerates/reclocks, no different than using a regular rpi4 connected to an intona that does the regen/reclocks in order to conduct the isolation.
Had the red, sold it because it sounded no different than a regular rpi4 with an intona. Same for my fitlet3 with spf connection, it just sounds the same as my regular rpi4 with an intona. I have tried various psu as well - ifi elite, allo shanti lps, farad lps, now settled for the cui medical grade psu recommended by Jussi. None makes any differences as long as an intona is in the chain
Yes, with USB connection, DAC’s internal clock is used. Everybody knows this.
But PSU matters, it always does. This is the reason I wouldn’t connect the server machine next to my hifi, especially when it would run heavy processing like DSD oversampling with HQP. I want to keep its noisy SMPS away from the rest of the system. Obviously you can build a server machine with linear PSU but to me it’s easier to just isolate the server completely, in another room. Also the streamer’s PSU matters. Before Holo Red, I had iFi Zen Stream with external PSU. Changing it to linear PSU had an impact on sound quality.
Now with Red, even different OS running on its Rpi4 makes a difference in sound quality. Running dedicated HQP NAA sounds better than Red OS for example. Now I’m running GentooPlayer on the Red. Even the USB cable running between Red and DAC200 matters (I use Intona Premium, highly recommended).
So yeah, I’m sure Intona isolator does its job isolating the noisy PC from the DAC, but it doesn’t magically make the whole system immune to other variables. If you have to or want to connect the PC directly to the DAC then I’d definitely buy Intona also. I feel no need for it between Red and DAC though.
To me Holo Red is a steal. Highly optimised Rpi4 based streamer with integrated high quality linear PSU and clean USB output and top notch build quality.
For context, thanks to some help from both my friend Pete at DoubleHelix Cables and Daniel at Intona I’ve been doing some measurements of the actual noise at the output of several USB sources whilst they are in use/playing to a DAC.
This shows just how much cleaner even quite an affordable isolator is than a PC, phone, or many streamers.
(NOTE: Ignore the levels of stuff up around 250khz-1Mhz for the most part, I need to do some further testing to see how the activity of the device and keeping cable twisting consistent etc affects this so these shouldn’t be considered indicative yet, but the overall level of everything else was very consistent)
JDS synapse doesn’t get quite as quiet as the Intona 7055-C or 7055-D, but they’re pretty close, and both DRASTICALLY quieter than typical sources. Quite frankly if you’re wanting a guaranteed clean USB source, get one of these two devices and be done. (Unless your DAC already has galvanic isolation in which case leaving both ends of the connection ungrounded may not be ideal.)
RED isn’t quite as quiet as the intona, but again drastically cleaner than a phone, PC, or the Eversolo DMP-A6. Plus then you do have the benefit of keeping your USB ground connected at the source if your DAC needs it.
In some cases, lifting the ground from source may be beneficial too, depending on the overall grounding layout and whether the DAC/amplifier is using two or three pin mains connection. IOW, to avoid ground currents from flowing between the source and something else through a floating amplifier.
I would again also like to emphasize, that if one is connecting TV or similar that has antenna/cable-tv connection to any part participating in audio system through HDMI (computer etc), it is worth using a ground breaker on the antenna/cable-tv connection, and/or some form of isolation between this/computer and audio system. Otherwise the RF ground currents from the antenna/cable-tv connection may leak to the audio system. It is not just one or twice when this has been source of problems…
Yeah I think the main situation where you wouldn’t want to use an isolator or other method of ground lift would be if you are using a DAC with built in galvanic isolation, as then you can often end up running into connection instability in my experience.
Yes my hqplayer desktop pc is in the server room, I have 4 systems in different rooms = 4 NAA and 4 intona all running hqplayer NAA 5.1.3 so yes I have tried a number of different sbc and different power supplies and different usb cables (Supra Excalibur, intona, inakustik reference, startech, monoprice, some china sterlin silver cables) and I swap them for experimentation - red, up gateway, rpi4, fitlet 3 (being the only one connected by sfp, the rest just use cat 6 cables as recommended by Jussi)
My experience is with an intona
In the chain, all the sbcs sound identical
to a point I can easily just use a rpi4 with a cui medical grade psu and I will never be able to tell a difference
Thanks GoldenSound. Is there any harm other than potential instability when connecting an intona to a dac with built in galvanic isolator? (which I have never experienced)
I had been using the intona for the holo cyan 2 and still using the intona for my ifi pro idsd, both have built in galvanic isolation.
You mostly just incur increased risk of packet loss and dropouts.
There are some situations where the DAC is galvanically isolated, but done via methods like a transformer which can still pass unwanted noise. In these cases I have found that there can be a possible improvement with an isolator as long as it doesn’t make the connection unstable.
But for the Holo stuff where ALL lines are completely optically isolated I’d suggest just running them directly as there’s no benefit to using a second isolator.
IDK if the isolation in the iDSD is solely on the 5v/gnd lines or everything so it may be beneficial to use one there but I’ve not tested one
Yeah, isolation won’t make much difference there. At the moment I’m running my Spring 3 from UP Gateway based NAA straight (CUI Inc medical grade 5V PSU) and it measures fine. And Spring 2 straight from my HP Xeon server and that measures fine that way as well…
I’ve tested both with Intona too, as it also works and measures fine that way.
On T+A DAC 200 I’m using Intona 7054 and it makes measurable improvement. And on T+A HA 200 I’m using 7055-C and it makes measurable improvement there too. In addition, with Accuphase DAC-60, I’m using Intona 7054 and it makes big measurable improvement there as well.
On Marantz SA-12SE, I’m using 7054 too, but there the difference is smaller but still worth.
My Holo Red is paired with TEAC UD-701N and it is pretty good for that, alternatively I would recommend USB isolator with that DAC.
Thanks @jussi_laako and @GoldenSound for the measurements. Very interesting and new info to me. I hadn’t seen the Intona measurements before. So there might be something to gain with Intona between the Red and DAC200 after all. There’s only one way to find out if it’s something audible
Which model of Intona isolator would you recommend between Red and DAC200? I remember reading somewhere that the USB3.0 models wouldn’t work properly with Red’s USB output since it doesn’t output enough power? IIRC, this was in Head-Fi’s Holo Red thread.
I guess 7055-B would work with Red, but it has worse measured noise compared to -C model. Looking at GoldenSound’s measurements, Red is already so low with output noise that I’m not sure if using 7055-B would be beneficial. I didn’t find 7054 measurements but I wonder if it might actually be higher in noise than Red’s own USB output? 7054 data sheet says output noise <60uv 20hz - 22khz.
I can’t hear the sound of my computer even if I put my ear to it. And the Cyan 2 is directly connected to the computer and there is no difference even though at first there was a RED in between. i9-14900KS, Apex Encore, Seasonic Prime TX-1300, RTX4090 and 48gb 8000Mt/s.
When I summerize, the various reactions on my comment that I did not like the Raspberry Pi USB output, I was given the answer that Holo Red, using a Pi, proove me wrong for the noise and in a link the Holo Red was given as an example as very low noise output.
I have to check with you so that I do not misunderstand what I read: When the USB Host output is used, the Pi is NOT involved, the separate and isolated USB2.0 Host output is only connected to the Holo circuitry? Have I understood this correctly? That is, the Pi is never a subject in the case of noisy USB, as the output is never from the Pi?
So essentially the Holo Red is built a little like I have built my NAA, electronics separated from USB2.0 Host output and with power fed from a linear supply? Except mine is fed from external LPSU, the Holo Red from LPSU inside the device?
Please read the discussion above. GoldenSound even provided measurements, showing how Red has a very low noise USB output, only outperformed by a dedicated USB isolator devices. I know only what I’ve read from GoldenSound’s review/measurements and from elsewhere.
I think the bottom line I get from this discussion is, if you are a hqplayer user, you should use usb connection.
A great approach would be to get a rpi4 and add an intona because the red - while versatile if you need all these different digital outputs, usb performance is not comparable to a dedicated isolator.
I have highlighted a quote from the post you published here stating the USB2.0 Host output is not connected to the Pi, but reclocked/transmitted to a isolated USB output and fed from Holo Red 5V power (LPSU). That is more like an isolation measurement giving the small values in the test. The Pi is not connected to the USB out, hence the superior measurements. And that make sense as they probably came to same conclusion, the Pi is not a good enought USB output, not even powered from a LPSU, as there are onboard interference issues with the Raspberry concept as is with shared busses, onboard dc/dc converters. I think @Deric_Chan summerize well, if Pi is used or any computer based NAA, there is a need to lower/cancel USB noise distribution.
