Ah, yes. I do remember how I got banned from a certain computer audiophool forum for asking to see the source code of MQn (at the time the author claimed it was open source).
Sorry to sound harsh but my view is:
A lot of extra money for no real gains, look up simultaneous switching noise that the PC will create itself.
A lot of extra wiring with no regards to any possible EMC problems that this is going to cause and it will cause EMC problems. I don't even see any ferrites on the cabling or any filtering inside the unit. As to the PC end plugging that lot of cables into the supplies is bypassing the PC case (an major EMC barrier) and adding effectively antennas to the PC.
Power supply should be in the same case really. From past experience with these sort of things I would be surprised if the set up with an external linear supply would pass EMC testing, and EMC problems lead to signal integrity problems.
I guess a really cheap underpowered computer is what i've been lacking to appreciate how great Linux is. Because it is not.
The point i was trying to make is that you don't need an operating system to emulate the client PC in the 2PC Jplay setup in hardware. All file operations are done entirely on the server and a predigested stream is sent to the client ip address. Apart from some basic parsing i cannot think of much else a client is going to do. Of course there is the issue of getting the output to a dac. In case of USB things will get ugly. I2S, otoh should be real easy.
The rpi is definitely not underpowered for simply playing music. Extra power won't buy you anything (except noise and heat).
And the point I am making is that with linux it is very easy to remove all the parts of the OS that you don't need.
Indeed. pretty much what any DAC does.
You might not like the looks of USB, but it works. What extra benefit do you think you get with I2S?
Is it really that simple?
1. The data flow is controlled by the clock located at the DAC. You have to establish a feedback loop controlling the data inflow speed.
2. Operating the DAC/soundcard controls:
* switching sample rate - controlling the clock and reconfiguring the dac chips accordingly
* configuring the card outputs
* controlling whatever form of HW volume the sound device offers
* setting preamble of SPDIF output stream if this is used - e.g. dolby digital output vs. PCM output
etc.
Basically you have to implement a communication protocol for these control elements, along with the feedback-controlled data stream.
Of course an FPGA is perfectly capable of this. Just the development costs...
And then you hit the question of support for various sound devices people already own and will want to use. The result - drivers. An embedded linux board supporting almost any soundcard on the market out of the box is WAY easier solution.
Quite a few DAC chips require external setup (pins or I2C regs) when clock changes. That is what drivers do.
Anything can be done in DIY. I just do not see any advantage compared to embedded linux with full alsa drivers stack.
Sorry to sound harsh but my view is:
A lot of extra money for no real gains, look up simultaneous switching noise that the PC will create itself.
A lot of extra wiring with no regards to any possible EMC problems that this is going to cause and it will cause EMC problems. I don't even see any ferrites on the cabling or any filtering inside the unit. As to the PC end plugging that lot of cables into the supplies is bypassing the PC case (an major EMC barrier) and adding effectively antennas to the PC.
Power supply should be in the same case really. From past experience with these sort of things I would be surprised if the set up with an external linear supply would pass EMC testing, and EMC problems lead to signal integrity problems.
Hi Marce,
thanks for being the first to actually reply regarding my questions. There seems to be an interesting tangent going on there that I'm going to clip soon if it doesn't get back on track. I can see the underlying issue, whether it's worth it to go with a full 2 PCs or some other combination, however I think we're a bit futher off base now and arguing something that belongs in another thread. We'll see.
Firstly, I've read a lot of positive reviews from the "a motherboard inherently has switching supplies on it, so linear supplies can never fully take care of that", but then someone tries some sort of multi-rail Linear PSU and it sounds good. In fact, if you read back just a few posts you'd find someone commenting on just that. Most of the time I hear someone saying "it can't sound better, it's science", but they've never tried and can't tell from experience. You, on the other hand are saying that you have, but with less than positive results. Interesting, mind filling me in and what you tried? Help me see further whether I can see the parallels in what I'm talking about now.
And then about the cabling, I think you may have misunderstood some part of the design. There would be one umbilical connecting the PSU to the computer chassis, then inside the protection of the computer chassis all of the looms would go to their respective slots. Once inside, this is no different than a normal computer power supply. As for the umbilical that ties the two together, this is common with any low or high end component with an external PSU. Nothing wrong or out of place there that I can see.
Now on to the EMC(or EMI compatibility/control) issue, that's a potentially different story. I've never seen computer wire looms with ferrites on them, but I think about $5 could solve that problem if I wanted to give it a shot. Which now that you've mentioned it, I think I might. And given that we aren't talking about Delta Electronics here, I bet the EMC testing was indeed on the slim to non-existent side.
->Because I've also been going down this path for a while, I have EMI absorbing/reducing material, copper shielding tape, and copper shielding braids. So I'd say once I get it all into position, I'll start shielding all the looms anyway. I can report back on any difference if I hear one.
As for internal PSU filtering and EMI mitigation, that is true and I don't really see anything like that. There is a smaller transformer inside the blue, plastic housing, so potentially that's it? But yes, no chokes, and not raelly any by-pass caps, that's for sure. Any ideas on how to increase anything like that, from a basic point of view. I know you've ruled out this PSU, but if you were to think about improving some part of the internal EMI control, without overhauling it all, what would you do?
The problem with PCs is not noise getting in....its noise getting out. As you may have gathered, its pretty noisy in a PC and for all the digital stuff the PC doesn't care and it doesn't matter and they can be built at a reasonable cost, so all the effort goes into keeping the noise in the PC then passing CE etc is easier. Adding that umbilical may allow this noise to escape, a decent ferrite would help here.
My own view of these things and from working with other stuff where analogue and digital meet (its more common than people think) is for the best analogue fidelity isolate the last stage with the bare minimum digital (the DAC or ADC) components in that last section.
As I said I was possibly being unduly harsh, but for practical reasons. it is something to watch out for.
I have seen linear supplies feeding systems and have seen problems, but then again I have seen SMPS main supplies cause problems, but from a power density point of view SMPSs have the edge, also the noise created can be easier to filter as it is generally higher frequency in nature.
One of my first jobs when I was a student was servicing linear supplies for computers, a full rack of paralleled 2n3055s, plenty of heat, but they worked.
Both have their advantages , good quality well thought out designs of either type will do the job at the end of the day.
Have fun
My own view of these things and from working with other stuff where analogue and digital meet (its more common than people think) is for the best analogue fidelity isolate the last stage with the bare minimum digital (the DAC or ADC) components in that last section.
As I said I was possibly being unduly harsh, but for practical reasons. it is something to watch out for.
I have seen linear supplies feeding systems and have seen problems, but then again I have seen SMPS main supplies cause problems, but from a power density point of view SMPSs have the edge, also the noise created can be easier to filter as it is generally higher frequency in nature.
One of my first jobs when I was a student was servicing linear supplies for computers, a full rack of paralleled 2n3055s, plenty of heat, but they worked.
Both have their advantages , good quality well thought out designs of either type will do the job at the end of the day.
Have fun
Here's a pic of a small circuit board I've been working on to manage voltage monitoring and ATX power on sequencing. It monitors 12V, 5V and 3.3V, provides power to the motherboard at the right time and manages the PWR_OK functions. At its core is the TPS3510. C1, C2 and C3 are placeholders for now - they provide voltage slew rate control to protect the pass transistors.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.