Sgr, If you transfer files to DAC using Ethernet bridge then nothing that you mentioned makes any difference since they affect only timing while you transfer data (no timing). Some things might reduce overall noise but things like amount of RAM or processor speed make no difference at all. Solid State Drive might reduce ambient electrical noise but does not affect data sent. Playback program makes difference only if it does any processing. For direct transfer even Itunes will do. Again, it is for Ethernet bridge (or wirelsss) connection. |
The CAPS system has the trick USB card. I bet that makes a huge difference.
I'm a Mac guy and I have a Mac Mini, but if I were solely interested in sound quality, I'd go with the CAPS. |
Kijanki,
I hear you loud and clear. The computer should not make a difference via DLNA and Ethernet. But . . In my experience there are many things that can influence Ethernet sound.
Like Cat 5 vs Cat 7 cable or
Linear PSUs for Router, and Gigabyte switches or
PSUs for WHS server
Speed and amount of RAM in WHS
Speed, ram, and processor power of the main PC that runs jRiver or eLyric which transcodes Flac to WAV
The amount of Flac compression
Power cords for each computer and DAC
Using PS Audio Noise Harvesters, Ultimate Outlets, and
Quattros, and Regenerators like the P- 10 and PPP
The sound of player and transcoding software
Software jitter of the computer compiling the programs
Fidelizer 3.0
The list seems to go on and on, and yet each tweak I try just continues to improve the resolution of the digital playback system. I refuse to use the word analog any more to describe the system as it seems to have gone beyond any analog I've heard and I'm proud to own a digital system which is something I could not say without some reservations in past years.
So finding the right computer to run jRiver and can be especially important.
If you have a chance visit the PS Audio Community where many other digital pioneers have written up their findings and others have independently tried and discovered new methods of finding new notes where none were supposed to be.
If you are working with this type of a system try some of these ideas for yourself, you might be surprised.
Thanks for taking the time to read this.
Steve |
Some things like type of Ethernet cable might change overall ambient noise and the same goes for the computer. On the other hand faster computer can make it much worse generating more noise at higher frequencies. Amount of RAM makes no difference as long as computer can run. What is being send on Ethernet has no time base - it is data in packets. On the other side of the bridge timing is recreated. It doesn't matter if packet is early or late because it goes thru FIFO buffer with new clock. Yes people experienced improvements with faster computers, more RAM and better playback programs but not in Ethernet or Wireless connection.
As for my (wireless) system - no difference with playback program or amount of RAM (no surprise here). No difference between internal or external HD. No slightest difference when I use computer for other tasks during playback. |
I think if you want to go the MacMini route, you can consider the Mojo mods which replaces the switched mode internal PSU with a DC filtered board. There's an external linear PSU that's larger than 4 Minis :P
A custom built CAPS server may sound better if care is taken in the choice of components. Like others have highlighted, there are audiophile/esoteric PSUs (fanless), USB cards etc which can make a difference.
But if there's one thing I am not too keen on with Windows setups is how it progressively gets slower with each update from Microsoft. I'm now currently typing this on a freshly installed BootCamped MBP 13" dual core (circa 2009) and it's surprisingly zippy compared to a quad core that was last clean formatted about a year ago.
I don't find this to be the case with my Macs.
The other reason I use Macs is cos I use iTunes for my music archiving (also sync-ing with my iPod/iPhone/iPad etc) and I use the free Remote app on iOS to control the headless Mini (it works with PureMusic/AudirvanaPlus/Amarra etc). iTunes for PC is a bit more bloated and less integrated with the OS. |
i've had both - a caps type server and mac mini. i'm currently using a late 2009 mac mini (SSD, 8 gigs of memory... the empirical audio site has some great info on this) with audirvana and it's the best sounding computer playback i've yet attained. |
I just got the CAPS Zuma server about 3 weeks ago....and I have not switch back to the Mac Mini since I last compared between the 2.
The Zuma just has more details and better phase accuaracy that is truly amazing as far as Computer Music playabck is concerned. Prior to this, all playback from RBCD and SACD was my prefered source over the Mac Mini. Not anymore.....
(Audio) |
Audio signals in a computer are processed in real time. This is the case regardless of whether you use JPLAY and memory playback engines, or any method.
The first common misconception is that data on the computer is just ones and zeros. This is not the case. The audio data is a square wave, which is an analog representation of a digital signal. This square wave, a pulse width modulation, contains both amplitude and timing information.
Why does this matter when it comes to choosing between Mac and PC? Several considerations.
First, where does distortion come from in the computer?
Because audio is a real time process, the square wave in the computer is duplicated and modified based on software algorithms thousands of times. Each new "version" of the square wave is generated from voltage in the power supply.
Noise radiated from the computer and noise from the power supply, specifically high frequency noise, gets folded over into the square wave. This introduces harmonic content that shouldn't exist in that signal and results in the digital sound many hear in computer audio.
The longer it takes the computer to process the music, the longer the signal can pick up noise and the more amplitude distortion that occurs.
The next misconception is that the computer doesn't matter if you are streaming over the network or if you have a high-end DAC that fixes Jitter. This is not true.
While it's possible to fix timing information and jitter, it's impossible to fix amplitude distortion on the square wave. This amplitude distortion becomes the square wave and no amount of filtering will do anything besides create other forms of distortion and coloration.
Even if you are streaming to a network player or DAC, all you have done is removed the cable. This can, of course, reduce noise, but it will not eliminate your source as a vital component to your sound.
The signal created in the computer must still be processed before being sent down the network. Once the data hits the network output it is converted into packets, which are buffered. These packets are then sent off to the DAC or network player for further processing. The idea of a network player makes little sense to me as the signal is then being processed by two computers instead of one. Digital doesn't like handshaking.
Ok, so what makes the computer process data faster and what impacts the sound of the computer?
The CPU clock speed means almost nothing. Modern CPUs process data SO quickly that there really isn't much difference between 2ghz and 3.5ghz. Sort of, there are a lot of other differences that play into those clock speed.
First, and probably the most important factor in computer/audio performance is the amount of L3 cache in the CPU.
This cache interfaces with the memory controller and determines how quickly the CPU can PREFETCH data for processing.
Here's the problem, the CPU must load data into memory for processing. That means it will always be limited by the memory speed and bandwidth.
Memory speed is limited by its CAS latency, which is how many CPU clock cycles must pass before a new string of data can be loaded for processing. Lower, of course, is better.
When the CPU must wait for data to be loaded into RAM (and this is constantly), the CPU incurs a wait-state. The wait state is exactly what it sounds like, the CPU is waiting to process. The longer it waits, the longer the square wave floats and picks up noise and distortion.
Comparing the Mac Mini and Zuma...
The newest Mac Minis can have up to 6MB L3 cache, whereas the Zuma (being that it's just a computer) can have anywhere from 6MB on up depending on the CPUs used. For instance, a 2GHZ Xeon with 24mB L3 cache will sonically outperform a 3.5GHZ Core i7 with 8MB L3 Cache.
So the PC Route has better CPU potential, and I believe the CPU used in the Zuma has 8MB L3 cache.
Power Supply...
This is probably the most mission critical element of music server and computer audio performance.
Using a Linear power supply on a mac mini is not the same as using a linear power supply on a PC like the zuma.
The Mac Mini operates off a 12V power supply where as a computer operates off an ATX Specification power supply. The ATX Specification requires (simplified) 12V, 3.3V, 5V, 5Vsb, -12V and various logic circuits to function.
With a mac mini it's as simple as using an external linear power supply. With a PC, that same linear power supply faces a few more obstacles to the best performance.
With a PC, if you are using a 12V Linear Supply, you will be limited in several ways. First, that 12V supply will have to go through a DC-ATX converter, which is a switching power supply, to generate the additional voltages. This will limit your hardware selection due to current limitations and be exceptionally noisy even with the best DC-ATX converters and linear supplies.
If someone was to compare the same linear power supply (assuming it's perfect) on a modded mac mini vs a PC with a DC-ATX then the Mac Mini will win.
In my experience, the only way for a PC to outperform a decked out Mac Mini is with an ATX Specification linear power supply.
Comparing stock Zuma vs Stock Mac Mini...
This is a no-brainer... the Zuma is a more powerful machine running a more customizable operating system with more upgradeable hardware.
Comparing Upgraded Mac Mini to Stock Zuma...
The Mac Mini, with the right design, will significantly outperform the Zuma.
Comparing Zuma with ATX Linear Supply and fully modded mac mini...
The Zuma will significantly outperform the upgraded Mac Mini.
I have a lot more information about this on my website. Hopefully this will help you make a decision between the two units. Both have a lot of potential and are quite good.
Courage,
Ryan
Core Audio Technology |
Ryan, how does the Red Wine Audio battery powersupply that is an option for the Zuma factor in to this. Is this an instance of an "ATX Linear Supply". |
so does "electrical noisiness" of the motherboard components also enter into this? |
The battery power supply from RWA is a single rail supply. They basically output using a 12V LiFePO4 battery, this 12VDC goes through a DC-ATX converter (basically a mini switchmode power supply) and that DC-ATX converter generates the necessary voltages from the battery along with sequencing and logic info the motherboard needs. It is not anywhere close in performance to a linear power supply, let alone an ATX linear power supply. Now, it's certainly a good option over MOST linear power supplies and a stock switchmode supply, just not the best solution by any means.
The electrical noisiness of the motherboard is going to depend on several factors. I prefer full ATX motherboards because they generally have more space for filtration between sections of the board.
The CPUs are generally quite well filtered with a bank of 24-30 capacitors right under the socket and several organic polymer caps around the outside of that.
To combat EMI on the board itself I shield the motherboard.
With my ATX power supply and ATX filtered cables it allows for both very low noise on all 10 power supply rails(<3uV RMS) and galvanic isolation between devices (HD, CPU, motherboard, mechanical devices, etc).
I have a thorough article on my website about batteries vs linear power supplies and several other articles about computer audio in general. Lots of tips for folks who want to build their own server based on my research. You can pretty much duplicate what I do in my Kryptos servers if you're willing to do a little homework.
-Ryan |
Thanks for explaining Ryan.
As always, the golden rule of audio applies; No matter how much money you spend and what you buy, there is always a better mousetrap. |
"Always a better mousetrap" is depressing. And, of course, subjective.
Personally, I'm a big fan of Intel's SSDs, but not so of their motherboards. I'm a HUGE fan of FANLESS! Both the garbage created by the motor and sound it makes as it moves air are distractions, to me.
There is a lot of pseudo-science out there aimed to explain what sounds good, and what does not sound good, in the world of computer audio. The trouble is, we've hit a point where our theories of electricity (theories are just elaborate models) cease to describe the entire situation. To some degree, we are still pioneers in the field of digital audio and our theories are horribly unproven.
Many who sell digital audio gear are pushing theories in an effort to further their marketing, or justify their design decisions. (Reminds me of "tubes sound better/worse" thinking.)
Ultimately, you have to listen. Within any given budget, at any price point, there are many options. Until you listen to a unit in your own system, you won't know for sure. Perhaps you can trust a reviewer to listen for you. Perhaps you can take a suggestion from a friend. Perhaps you can believe the sales pitch from a manufacturer -- and perhaps you can't.
What you can do is trust your ears. I, for example, only buy items I can listen to for an extended period first... and I only buy items that sound more accurate to me than the item it is replacing.
So... I'd suggest you negotiate a loaner, or find a manufacturer with a money back guarantee. Best case would be to find one who encourages you to listen for yourself! |
Since he owns the company I'm sure they have plenty of time on them. This thread really does belong under the "Audio Club" section though. |
Disclaimer: I am proud owner of Musica pristina server and immensely happy with its sound quality (way better then my Esoteric K-01) and service as well
Still, I disagree with esteem member "Musicapristina" regarding his position that ONLY your ear is final judge and therefore there is no need to talk and look at hardware/software specs etc.
There are hundreds of amplifiers, for example, for sale. If I am after amplifieer - how will I choose which to listen so my ears will be the final judge????
Its not possible practically.
So, I rely on a few sources: a) review of the reviewer whose previous reviews I found to be credible, b) reviews on-line by my peers, amp owners - I can see the systems matching etc. Again, I try to listen to these members who already proved to me that they know what they are talking about - too many deaf people (including reviewers) talk about music here. Finally, specs! Yes, specs. From 1960's its is fashion to say something bad about specs and at that time it was true - there was no correlation between specs and sound iality. We simply did not understood which specs are important and what their importance is.
Today, good specs indicate on the POTENTIAL of the equipment. For example, if BW = 300 kHz then I know its fast and therefore if not screwed, music will be not sluggish (but it may, if screwed). So, potential is there but reality must be examined i.e. heard. If, on other hand, BW = 20 kHz then I know (know!!!!) that this equipment CANNOT play fast and therefore I will not even bother to check it.
Same with servers - the main reason why I went to Musica Pristina was their published spec of power supply: ultra low rip and no other server manufacturer publish this spec. For me its very important spec saying that power supplies POTENTIALLY (only potentially) are very stable, will not produce ripping artifact associated with ear-piercing odd order of harmonic distortions. There were a few more good specs indicating POTENTIALLY excellent server.
I took it for 30 days home trial and...fall in love with it. THE END!
Rafael |
Further to Core Audio's comments, it's hard to overstate how stupid an idea the CAPS Zuma is - although it would be wrong to take away from Ryan's post that batteries are a bad idea.
Assuming you solve the problem of low-noise AC-DC conversion, by far the next biggest issue is DC-DC switching noise. In fact, unless you deal rigorously with both, there's no point calling a computer transport 'designed for audio'. Get the power right, and you're about two-thirds of the way to the endpoint. Get it wrong, and you're miles off-track.
A motherboard with a single-rail input, or a Mac Mini, or a computer using a PicoPSU suffer equally from the same problem - lots of unnecessary DC-DC switching noise. Game over.
In our experience, using a Red Wine battery with a Zuma-like machine or a Mac Mini is roughly level-pegging - with the latest Macs ahead by a nose sonically. Given that they're also cheaper, more reliable and easier to use, I would suggest the answer to the OP is Mac all the way.
However, as we've demonstrated on this side of the pond, and Ryan's Core Audio products have blazed a similar trail over there, the only way to do this properly is to deliver at least three linear-generated or well regulated battery-powered rails into multi-pin ATX, Molex and PATA.
That's where a computer 'designed for audio' begins. Anything else is rather broken for this application. |
