Jump to content


  • Content Count

  • Joined

  • Last visited

About nbpf

  • Rank

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. The min. amd max. frequencies can be set for all governors. Avoiding max. frequencies is one of the reasons why one should not use the ondemand governor. The problem with ondemand frequency scaling is that, under varying load, the CPU frequency is constantly changing between min. and max. All basic setups for dedicated audio applications suggest to avoid this. On certain CPUs, frequency scaling can even be heard as acoustic noise, perhaps induced by thermal effects. I am travelling at the moment and I have little time to look for references. But if you remind me in a few days I should be able to provide pointers to web pages with the basics of OS settings for audio applications. The simplest way to avoid frequency scaling on RPi devices is via config.txt, see
  2. My understanding is that the ondemand governor should never be used on dedicated music servers or transports. Just select performance and limit the max. CPU frequency via config.txt.
  3. On the RPi 3B+ that hosts the DigiOne Signature, I run upmpdcli (https://www.lesbonscomptes.com/upmpdcli/) and MPD in a minimal Raspbian Lite installation: no X, no web servers, just ssh, rsync and mandatory packages. On that RPi I also turn off HDMI, bluetooth, the internal sound card, wired Ethernet and the USB bus. This brings the power consumption of the RPi + DigiOne Signature combination down to about 300mA. I have tried RT kernels as an alternative to the standard kernel (4.19.66-v7+) with no obvious effects on the sound quality. I have also tried Volumio and DietPi. Sound quality wise they were neither better nor worse than my setup but I did not like their web interfaces. The RPi that hosts the DigiOne Signature is connected via wireless to the LAN and receives music data from another RPI that is wired to the router. This second server-RPi runs MinimServer. also in a minimal Raspbian installation. Best, nbpf
  4. In my system, a Sonore ultraRendu (ppowered by UpTone Audio JS-2) into Mutec MC-3+ Smart Clock USB into Naim DAC is neither better nor worse than an Allo DigiOne Signature (powered by JS-2 [RPi side] and LPS-1.2 [clear side]) straight into the Naim DAC. Just perhaps a little bit different.
  5. I haven't explored the idea systematically: as a default, I do not have a wired connection and I stream either form the local T5 or over the wireless. I typically lay down a wired connection only temporarily for testing purposes. But I was surprised to see how much power can be saved by turning off the USB bus and I like to do so to keep the JS-2 just warm. I think the new RPi 4 has separated USB bus and wired ethernet interface. Thus, it should be possible to switch off the USB bus while still keeping a wired ethernet connection alive.This is also something that I have not checked so far.
  6. I have a 1TB Samsung T5 attached to the RPi 3B+ that hosts my DigiOne Signature. An instance of MinimServer also runs on that RPi. After boot, I can replay from the T5 via the MinimServer running on the RPi. Alternatively, I can turn off MinimServer and the USB bus (and, thus, wired ethernet) and replay from a server (also running MinimServer) via wireless. Or unmount the T5, turn off MinimServer and replay from the same server via wired ethernet. I freely cannot detect significant differences in the sound quality in my system and through my ears. Perhaps the rendition has a little bit more weight (or noise?) when replaying from the attached T5. As a default, I typically turn off MinimServer and the USB bus on the RPi that hosts the DigiOne Signature. This reduces the overall consumption of the RPi + DigiOne Signature to about 300mA which is a very low load for the JS-2 that feeds the RPi side of the DigiOne Signature.
  7. If you do not plan to run a server, upsample or transcode on the RPi that hosts the DigiOne Signature, then an RPi 3B+ is more than what you need, as @suzywong has pointed out. Upsampling and transcoding is anyway better done on a separated machine.
  8. I have a number of RPi 3B+ and one RPi 4B but I have not compared them as hosts for the DigiOne Signature. In my view the most important question is the software, not the hardware. Once you have settled on a software, you can run it on a RPi 2, 3 or 4 and check whether the hardware matters or not. I do not think it does but I might be wrong, of course.
  9. I use a DigiOne Signature with an RPi 3B+ running upmpdcli (and thus MPD) in a minimal Raspbian Lite installation. upmpdcli (https://www.lesbonscomptes.com/upmpdcli/) gives you Tidal (and Qobuz, Google music, etc.) integration. No need to use Volumio, DietPi or other so-called audiophile distributions. I typically fix the CPU frequency to 1GHz and switch off HDMI, bluetooth, the internal sound card, wired Ethernet and the USB bus on the RPi that hosts the DigiOne Signature. This reduces the power consumption of the RPi (3B+) + DigiOne Signature combo to about 300 mA. It sounds very good to me. Best, nbpf
  10. This works fine but you have to setup the RPi that runs LMS to act as an access point for the USBridge. Just search for "RPi access point" and you will find a lot of documentation. I have used this one https://www.raspberrypi.org/forums/viewtopic.php?f=36&t=191453 to set up an RPi as a wireless access point. You will have to do that the other way round, provide wired access and reroute the packages that come from the wired USBridge to the wireless interface. It is pretty much straightforward.
  11. I guess you mean which responses are suitable because what is allowed and what is not allowed is stated in the forum's rules. The answer to the question of what are suitable responses to posts that "cannot possibly be correct" very much depends on what does it mean not to be correct in the specific case at stake. A statement or post can be logically incorrect or it can be in contradiction with facts. In the second case, providing suitable responses is often more difficult and open to controversy, I would say. But it is also the case that logical arguments are often takes as arguments about facts and, of course, the other way round! Also, we sometimes face posts that are neither correct not incorrect but plainly meaningless. Finding a correct answer to such posts is often difficult. Finally I would like to remark that not all posts need to be answered. As long as they do not do any harm, posts that cannot possibly be correct can simply be ignored.
  12. upmpdcli (https://www.lesbonscomptes.com/upmpdcli/) gives you access to Tidal, Qobuz, internet radio and more. Control via Bubble UPnP app, Linn Kazoo, Lumin app, etc. Installing and setting up upmpdcli is very straightforward. No need to faff around with awkward web interfaces. A dedicated headless RPi 3B+ (with HDMI, bluetooth, USB bus, internal audio card and Ethernet switched off) running MPD and upmpdcli and hosting a DigiOne Signature will typically consume less than 300 mA.
  13. I see the point. In this case, I would suggest that you run your preferred UPnP/DLNA server in the OS you feel most comfortable with. I do not expect the OS to have any significant impact on the sound quality as long as the wireless data transfer to the renderer is managed well and the CPU frequency scaling is off in all systems. I might be very wrong, of course and, if you play around with different operating systems, I would be interested in your findings.
  14. I run Debian/Raspbian on all my systems. For serving and transporting audio streams I meanwhile use almost exclusively Raspberry Pi devices. On these, I run MinimServer (https://minimserver.com/) and/or MPD and upmpdcli (https://www.lesbonscomptes.com/upmpdcli/) in a minimal (about 500 packages), headless Raspbian Lite distribution. On RPis that host transports such as the Allo DigiOne Signature, I typically turn off HDMI. Bluetooth, USB, LAN and the internal sound card. This reduces the power requirement of a RPi 3B+ to about 200 mA when the CPU frequency is fixed at 1 GHz and has turned out to be a very stable and reliable setup. I do not have time to play around with OS tweaking: I typically install a system by cloning an image saved on my laptop and run that system without upgrading for years. Audio Linux, Volumio and DietPi seem obvious candidates for filling in the space in your drive. I tried them a bit but mostly disliked their user interfaces and found them too complicated.
  15. It is true that the language and, up to a certain extent, the results shown on ASR (not only for the EtherREGEN) do not inspire much trust. But this does not imply that they are flawed. The tests also seem quite straightforward and not very different from those that can be found on other sites. Assuming honesty and accuracy, I do not see obvious flaws in the methodology. If there are obvious ones, I expect UpTone Audio to point out such flaws and to provide more rigorous or more relevant measurements. So far, this has not happened.
  • Create New...