Thank you for your donation!


DAC discussions
#1
(08-27-2020, 01:44 PM)Tim Curtis Wrote: moOde is a server using MPD as the music server engine. MPD supports the tags listed in the following document https://www.musicpd.org/doc/html/protocol.html#tags. The "Extra metadata" option in Appearance just displays some of the tags under the Cover art.

The Library in moOde uses a subset of the MPD supported tags as follows;
Code:
file
track
title
disc
artist
albumartist
composer
comment
date (or MusicBrainZ OriginalDate or OriginalReleaseDate)
time
album
genre
Last-Modified

moOde works well for most collections but due to the way the Library is generated it will not work well for really big collections. You would be better off using another music server and client system for example one of the UPnP based control points like Kazoo and a UPnP AV media server.

Thanks Tim. 
Off topic. Why do I persist with Moode? Well, first of all I like the things you have accomplished with Moode; it's left-field compared to other existing streaming solutions. Being an audiophile (not of the flat earth variety), I hate how much some (most) manufacturers charge for streaming devices. 
So, I took the left field approach. I spent a lot of time with daphile, unfortunately only available on x86 architecture but running even on 20+ years, quite low-powered old PC's or laptops. However, one ends up with an large form factor compared to RPI. 
My next stop was the Raspberry. I tried systems like Rune, piCorePlayer, Volumio, Raspian, RoPieee and others. Of all these systems I regard Volumio as your biggest "rival" though its creator, Michelangelo, has assembled a development team around him and is now focusing on the commercial side of things. But I do also have an RPI3B+ as well as an RPI4, running Volumio for the exact same content. Volumio has a real different approach to indexing, namely using mpd directly in conjunction with node.js (node Red?). Indexing takes a real long time with CPU loads frequently over >100%. With Volumio you really do need to edit mpd.conf to suit a large library; otherwise it will crash (guaranteed). 

By letting Moode and others do their own indexing, extra functionality becomes possible.

Then there are commercial products (music server systems) based on the RPI. Two (at least) stand out for me: Bryston's music server, and Pro-ject's Stream Box S2 Ultra designed by the famous John Westlake. Pro-ject's system uses an industrial version of the RPI. The cost of these commercial products is many times higher than the components that make up the cost of a Moode based system.

I do own expensive audio hardware (though nothing over the top) and wanted to prove that, when it comes to streaming, it does NOT have to cost the world AND that such an RPI system can hold its own in an otherwise expensive audio hardware environment. Cut to the chase: both system CAN hold their own in such an environment! 

Crucial in an RPI based system is the DAC that is being used either as a HAT or a USB DAC. Forget the usual HATS though (Allo, Justboom, HifiBerry etc) and generally any DAC equipped with TI's 51xx DAC chip. The problem with a really good DAC, is the limited Linux support for DAC's. Some Linux drivers (such as for Marantz Ruby / SA-12, and drivers for Pioneer, otherwise very good streamers) are totally CRAP. This is a big problem; it's widely held that Linux has excellent support for UAC 2.0 DACs, FORGET it! Unfortunately, for reasons beyond my understanding, particularly the 'good' drivers don't make it to the Linux kernel while the crappy ones do. BTW: it's relatively unknown that Windows 10, from a particular version onward, supports UAC 2.0 as well. 

For a HAT, I can strongly recommend (UGeek) Aoide's DAC II HAT though one would need customized versions of Moode or Volumio (no kernel support). You will need to get these custom Moode and Volumio versions from Github or Sourceforge. With USB DAC's you don't have that problem but you'll probably need the RPI4 (because finally USB is independent from Ethernet so no more stuttering and other digital artifacts). The problem with USB DAC's is the developer's willingness to implement a matching driver if the DAC is not kernel supported or when it is, it may be crap. Those kernel driver also often omit DSD, so that DoP is the only option. However, this creates substantial stream overhead and is limited to DSD128, if I'm not mistaken.
Aoide's DAC I mentioned handles resolutions up to and including DxD (PCM 384KHz) and DSD256 (11,2 MHz). Now that seems pretty high, but for computers that is really peanuts.
With Moode I use a Topping D10 USB DAC (same format support), an excellent little DAC for very little money. I will be upgrading to Topping's E30 (PCM 768KHz, DSD 22.6 MHz, preamp etc. (http://www.tpdz.net/productinfo/434825.html)) shortly, though, hoping that it will be supported by Moode.

In conclusion, if server systems can handle such "large" libraries on comparable hardware (or even less powerful CPU's), Moode and Volumio should be able to do so too, maybe by letting Moode etc. leave the indexing and presentation to those server systems. I can envision an extension in Minimserver or a special (customized) version of it to do so, with the extra functionality mentioned earlier. All in the spirit of open source. But I have no idea if such a solution would be viable architecturally, or workload intensive for the developers.

Cheers,
marco
Reply
#2
Given the wide-ranging and idiosyncratic nature of this discussion, I think it belongs on the PC-based subforum of diyaudio.com where it can be read by the proponents of the various systems and components you have mentioned. I'd suggest titling the thread something like "The quest for the perfect low-cost solution" to attract the attention of more than just moOde users.

Regards,
Kent
Reply
#3
lol, OT for sure but...

The Pi Audiophile space has moved well beyond the noisy stock Pi's, cheap power supplies and HAT's. The Allo USBridge SIG, an OEM grade platform, + Allo Shanti LPS is a super low noise, clean power system for driving a well designed, Native DSD capable DAC.

Linux and it's single, universal USB audio driver and quirks config supports many native DSD capable DACs.

Here's some screenies of native DSD out to an Allo Rev DAC. The files are DSF64 and 128 format but the DAC also supports 512 . I personally don't do much with DSD other that to test that DoP and native work with moOde.

         
Reply
#4
(08-29-2020, 01:37 PM)Tim Curtis Wrote: lol, OT for sure but...

The Pi Audiophile space has moved well beyond the noisy stock Pi's, cheap power supplies and HAT's. The Allo USBridge SIG, an OEM grade platform, + Allo Shanti LPS is a super low noise, clean power system for driving a well designed, Native DSD capable DAC.

Linux and it's single, universal USB audio driver and quirks config supports many native DSD capable DACs.

Here's some screenies of native DSD out to an Allo Rev DAC. The files are DSF64 and 128 format but the DAC also supports 512 . I personally don't do much with DSD other that to test that DoP and native work with moOde.

 

Now, even more off topic!
Nice stuff for sure, the USBridge SIG alone is $ 240. The Shanti another $ 160. Add the RPI, case etc. Pricewise you could probably purchase a streamer such as the Onkyo NS-6170, now a few years old, but with functionality far exceeding anything available on the (user attainable) RPI platform, such as a digital radio receiver (in Europe DAB+), a dual differential DAC implementation (AK4490 no less) etc. I've heard it's not available in the US, but surely comparable ones are available. 

Look here http://archimago.blogspot.com/2018/12/me...-b-as.html for noise/distortion etc. measurements of a relative simple RPI3 system, including a simple switching power supply and a battery (one of those power banks). Recently, the author carried out an upgrade to that system. 

It only proves that the RPI3B was already suited to audiophile application, and it took Allo quite some time (relative to the speed of developments in the digital domain) to come out with equipment as sophisticated as yours. The Aoide Hat retails for around $ 50. 

As to "native DSD", many DAC's that supposedly support "native DSD" do not take the "DSD Direct" route. So the native DSD signals pass the PCM route, with all kinds of processing 'goodies' instead of bypassing it. It's a design choice made by the DAC designer. The most blatant is the conversion to PCM, following its path, and finally, reconverted to DSD. By its very nature, there's very little one can do with a DSD stream in the digital domain; for instance simple digital volume control or plain editing to change something in the song. The moral: "native DSD" is only "native" if it's also "Digital direct" (so straight to the analog conversion stage).

It's actually very clear when one studies the functional block diagrams of the DAC chips that every manufacturer publishes, with the exception of ESS; their interpretation of functional block diagrams are laughable. The implementation of the DAC chip decides whether or not the Digital Direct mode is available. 

And don't even get me started on "power supply cleaners", "USB cleaners" or "jitter killers". Objective measurements reveal that they are ineffective (a nice way to avoid the word 'bogus'). In several cases they make matters worse, not better. Now the RPI's are targeted too. 

Alright, enough already with the off-topic stuff!

The original problem still exists, but I will continue searching for a solution.

Cheers,
marco
Reply
#5
(08-29-2020, 10:57 AM)RPI4 Rambler Wrote:
(08-27-2020, 01:44 PM)Tim Curtis Wrote: ............snip ................

Thanks Tim. 
Off topic. .................................. snip ............................... Forget the usual HATS though (Allo, Justboom, HifiBerry etc) and generally any DAC equipped with TI's 51xx DAC chip. .................................. snip .................................. For a HAT, I can strongly recommend (UGeek) Aoide's DAC II HAT though one would need customized versions of Moode or Volumio (no kernel support). You will need to get these custom Moode and Volumio versions from Github or Sourceforge ......... snip ............

Yesterday, 03:55 PM -->
Now, even more off topic!

Nice stuff for sure, the USBridge SIG alone is $ 240. The Shanti another $ 160..... snip ...... The Aoide Hat retails for around $ 50.


Cheers,
marco

Aoide DAC HAT... not a hit with AudioscienceReview folk: https://www.audiosciencereview.com/forum...-hat.8662/

Of course they have not reached the sofistication of the Power Bank testing employed by Archimango (no expenses spared) but they are striving...

All the above is 'off topic', please accept my apologies.


Reply
#6
(08-31-2020, 06:52 AM)CallMeMike Wrote:
(08-29-2020, 10:57 AM)RPI4 Rambler Wrote:
(08-27-2020, 01:44 PM)Tim Curtis Wrote: ............snip ................

Thanks Tim. 
Off topic. .................................. snip ............................... Forget the usual HATS though (Allo, Justboom, HifiBerry etc) and generally any DAC equipped with TI's 51xx DAC chip. .................................. snip .................................. For a HAT, I can strongly recommend (UGeek) Aoide's DAC II HAT though one would need customized versions of Moode or Volumio (no kernel support). You will need to get these custom Moode and Volumio versions from Github or Sourceforge ......... snip ............

Yesterday, 03:55 PM -->
Now, even more off topic!

Nice stuff for sure, the USBridge SIG alone is $ 240. The Shanti another $ 160..... snip ...... The Aoide Hat retails for around $ 50.


Cheers,
marco

Aoide DAC HAT... not a hit with AudioscienceReview folk: https://www.audiosciencereview.com/forum...-hat.8662/

Of course they have not reached the sofistication of the Power Bank testing employed by Archimango (no expenses spared) but they are striving...

All the above is 'off topic', please accept my apologies.



Ouch, so it would seem. Ok, the OT continues. Time for me to start looking for my own measuring gear. I wonder which drivers Amir used. The HAT is specifically designed for (a modified version of Volumio) but it should work with generic drivers, which it doesn't (I seem to recall a driver named Audiophonics ES9018K2M or some such). On Github there's an installation routine to make it work on other systems, but even then it will not even run on a stock Volumio (or Moode) system with manually installed drivers. BTW: on Amir's site you can also find reviews of "USB cleaners", "Power Cleaners", "Jitter Killers". Those reviews reveal the truth as does stuff on Archimago's blog. 

It will not fit Ropieee and the combination with the Aoide; it really is designed for Volumio, so there's more going on than just the drivers.  The measurements are so far removed from the DAC chip's capabilities, with THD+Noise of 18% (!) it must be audible. Which it isn't in the right setup. So while my "defense" may sound lame, I really think the cause lies in the system + driver setup, or he had a broken sample. FWIW: these are the official specifications:
Line out:

Frequency Response: 20Hz ~ 20KHz +/- 0.1dB

S / N: -119dB +/- 1dB

THD + N: -120dB

DNR: 127dB

 

Headphone out:

Frequency Response: 20Hz ~ 20KHz +/- 0.1dB

THD + N: 0.0021% (16ohm load)

Output level: 138mW (+ 4dB gain)

SNR: 93dB

PSSRR:> 100dB
CMRR: 69dB

Which is in line with the DAC chip's specs. A designer must work very hard to yield the measurements Amir got. With these results you'll hear the noise through speakers or headphone when idling. I can assure you it doesn't. Moreover, with Amir's measurements, it is easily outperformed by the RPI's embedded DAC.


Cheers,
marco
Reply
#7
I'm not trying to create a confrontation about who is right or wrong, but I have my doubts about the Amir and Archimago tests. For me, the measurements they make are just information about the behavior of the audio signal in that situation but it does not determine whether the sound is good or bad, who determines whether the sound is good or bad are their ears.
I know that it is a subjective assessment and that it needs some experience to interpret the gains or losses or even any changes. I have as a great example the "USB cleaner" Uptone Regen, when he is in the system the stage is another, much more defined. Of course, this situation will not occur in all cases as for example on an audiophile PC that has a dedicated USB card the result will be minimal or null, but on a raspberry that has an unstable USB output the result will be more significant.
As I mentioned before, in my opinion the measurements are information and there is no way to hear them, the best way to determine if a DAC, or a USB cable or filter will be good or bad in the system is to hear it, your ear is who will make the judgment.

greetings
Rinaldi
Reply
#8
(08-29-2020, 10:57 AM)RPI4 Rambler Wrote: With Moode I use a Topping D10 USB DAC (same format support), an excellent little DAC for very little money. I will be upgrading to Topping's E30 (PCM 768KHz, DSD 22.6 MHz, preamp etc. (http://www.tpdz.net/productinfo/434825.html)) shortly, though, hoping that it will be supported by Moode.

It is supported (Moode 7). Appears as E30 in the MPD settings.
Reply
#9
(12-24-2020, 10:47 AM)Rinaldi Wrote: I'm not trying to create a confrontation about who is right or wrong, but I have my doubts about the Amir and Archimago tests. For me, the measurements they make are just information about the behavior of the audio signal in that situation but it does not determine whether the sound is good or bad, who determines whether the sound is good or bad are their ears.
I know that it is a subjective assessment and that it needs some experience to interpret the gains or losses or even any changes. I have as a great example the "USB cleaner" Uptone Regen, when he is in the system the stage is another, much more defined. Of course, this situation will not occur in all cases as for example on an audiophile PC that has a dedicated USB card the result will be minimal or null, but on a raspberry that has an unstable USB output the result will be more significant.
As I mentioned before, in my opinion the measurements are information and there is no way to hear them, the best way to determine if a DAC, or a USB cable or filter will be good or bad in the system is to hear it, your ear is who will make the judgment.

greetings
Rinaldi

I have to agree with you, sometimes different DAC's just sound different, not necessarily better.  I have the Topping E30, great DAC for the money, I use coax out on my Hiby R6 Pro just to a get a different sound to the Sabre DAC in my MoOde speaker setup or the dual DAC's in the Hiby itself.
Reply


Forum Jump: