Hello,
Ok, this is a weird one. How can the quality of passive components affect a digital stream?
I'm not talking about the components around a DAC or in the analog output stage. That I can accept and understand (or at least think I understand). But how about recievers, upsamplers, or even better, how can discrete components have a role in computers, motherboards, DSP units and so on?
I'm asking this because I see some people using mostly OsCon caps for example, in digital devices, saying these are better in the digital signal path...
Ok, this is a weird one. How can the quality of passive components affect a digital stream?
I'm not talking about the components around a DAC or in the analog output stage. That I can accept and understand (or at least think I understand). But how about recievers, upsamplers, or even better, how can discrete components have a role in computers, motherboards, DSP units and so on?
I'm asking this because I see some people using mostly OsCon caps for example, in digital devices, saying these are better in the digital signal path...
Hi,
Off the top of my head, you missed out ADCs - sample hold filters tend to be particularly sensitive to a property called dielectric absorption. But technically, that's still an effect on an analog signal.
As for digital circuits, there are other issues:
Parasitic capacitances and inductances along signal traces. If the signal gets too degraded it may simply be too weak to be picked up.
Power supply filtering. DSPs often require supply accuracy to within 0.15V or even less, as well as a very fast response to changes in the current. For that they usually require ceramic capacitors positioned as near as possible on the supply rails. So I don't think it's the "quality" of the components that's usually in question, but the type (eg: X7R ceramic versus electrolytic) and the board layout.
As for certain brands being somehow superior to others, I think that's mostly nonsense. There might be quality differences between manufacturers, but when someone makes 100s of millions of capacitors and they all have to be identical or at least fit into some kind of curve, it's just not the same as a small-time amplifier producer claiming that they "take the utmost care and dedication, and only source the very best components".
Off the top of my head, you missed out ADCs - sample hold filters tend to be particularly sensitive to a property called dielectric absorption. But technically, that's still an effect on an analog signal.
As for digital circuits, there are other issues:
Parasitic capacitances and inductances along signal traces. If the signal gets too degraded it may simply be too weak to be picked up.
Power supply filtering. DSPs often require supply accuracy to within 0.15V or even less, as well as a very fast response to changes in the current. For that they usually require ceramic capacitors positioned as near as possible on the supply rails. So I don't think it's the "quality" of the components that's usually in question, but the type (eg: X7R ceramic versus electrolytic) and the board layout.
As for certain brands being somehow superior to others, I think that's mostly nonsense. There might be quality differences between manufacturers, but when someone makes 100s of millions of capacitors and they all have to be identical or at least fit into some kind of curve, it's just not the same as a small-time amplifier producer claiming that they "take the utmost care and dedication, and only source the very best components".
Especially on digital circuits, its important to have a local reservoir of power supply (i.e. a decoupling capacitor) so that sufficient current can be supplied during peak demands. A typical example would be a logic chip switching its output.
Also, if you have local decoupling, it puts less load on the main PSU reservoir, so its voltage doesn't dip due to capacitor discharge. IOW, other circuits fed by the same PSU will be less exposed to the effects of current demands from the logic chip mentioned above. They are decoupled (to some extent) from the rest of the circuit.
So, its important for digital circuits to be able to switch properly, and for the analogue circuits to not have polluted PSU rails, to have at least capacitive decouping.
That's my theory, may not be 100% accurate.
Cheers,
Phil
Also, if you have local decoupling, it puts less load on the main PSU reservoir, so its voltage doesn't dip due to capacitor discharge. IOW, other circuits fed by the same PSU will be less exposed to the effects of current demands from the logic chip mentioned above. They are decoupled (to some extent) from the rest of the circuit.
So, its important for digital circuits to be able to switch properly, and for the analogue circuits to not have polluted PSU rails, to have at least capacitive decouping.
That's my theory, may not be 100% accurate.
Cheers,
Phil
One place where coupling caps are very audible is at the input of the spdif receiver. Of course, one may argue that the affected signal is not truly digital.
Hi SunRa,
"I see some people using mostly OsCon caps for example, in digital devices, saying these are better in the digital signal path..."
I think it's not on the signal path, but supply decoupling near the digital IC. It works because OsCon have low impedance at high frequency, lower than normal electrolytics. Even some P4 motherboard uses them.
Hartono
"I see some people using mostly OsCon caps for example, in digital devices, saying these are better in the digital signal path..."
I think it's not on the signal path, but supply decoupling near the digital IC. It works because OsCon have low impedance at high frequency, lower than normal electrolytics. Even some P4 motherboard uses them.
Hartono
Thanks all for the reply,
I too believe that psu is the main problem in digital devices. In the same way, proper use of decoupling caps is a must (although I tend to believe that different capacitors have different noise levels, not to mention the importance of low impendance).
I am courious though if other characteristics of say capacitors can have a influence in the digital domain. I tend to think that bits are just bits and important is (as philpoole said) for the digital circuit to switch properly. But if you have other opinions... 😀
The reason I've opened this thread is that I want to figure out how to take full advantage of a PC. For me PC's have countless advantages: high processor power, DSP, room corection, crossovers, and all this while beeing very flexible. Even more, some internet web sites are offering losless 24/96Khz/192Khz content. Playback from HDD or system memory or flash avoids mechanical jitter of optical devices. More about all these on soundcheck's thread "linux audio the way to go".
The problem of extracting a clean signal I think it can be adressed in the best way by using an ethernet/firewire audio card or diy solution. For me ethernet seems to be the best as it completly isolates the DAC from the PC. So noise interfering with the dac isn't any more a problem.
What I am concerned though, an this brings us back to the topic, is that the digital signal in a PC may allready be corupted. One of the way I am thinking of resolving a problem like this is to take a laptop motherboard as it is battery powered. Of course there is still the problem of onboard switching regulators. This problem I don;t know how to adress; change the capacitors, change the regulators with better ones... these are just opinions.
Regarding capacitance along PCB traces... I think motherboard designers are allready considering the problem. I think that high-end motherboards, built specialy for overclocking are designed very well to cope with the high frequency and maintain stability.
I have some other ideas too, some of them in the oposite way. (small motherboards with build in low power processors like VIA, linear/batery suply and so on) but I am waiting on your opinions.
I too believe that psu is the main problem in digital devices. In the same way, proper use of decoupling caps is a must (although I tend to believe that different capacitors have different noise levels, not to mention the importance of low impendance).
I am courious though if other characteristics of say capacitors can have a influence in the digital domain. I tend to think that bits are just bits and important is (as philpoole said) for the digital circuit to switch properly. But if you have other opinions... 😀
The reason I've opened this thread is that I want to figure out how to take full advantage of a PC. For me PC's have countless advantages: high processor power, DSP, room corection, crossovers, and all this while beeing very flexible. Even more, some internet web sites are offering losless 24/96Khz/192Khz content. Playback from HDD or system memory or flash avoids mechanical jitter of optical devices. More about all these on soundcheck's thread "linux audio the way to go".
The problem of extracting a clean signal I think it can be adressed in the best way by using an ethernet/firewire audio card or diy solution. For me ethernet seems to be the best as it completly isolates the DAC from the PC. So noise interfering with the dac isn't any more a problem.
What I am concerned though, an this brings us back to the topic, is that the digital signal in a PC may allready be corupted. One of the way I am thinking of resolving a problem like this is to take a laptop motherboard as it is battery powered. Of course there is still the problem of onboard switching regulators. This problem I don;t know how to adress; change the capacitors, change the regulators with better ones... these are just opinions.
Regarding capacitance along PCB traces... I think motherboard designers are allready considering the problem. I think that high-end motherboards, built specialy for overclocking are designed very well to cope with the high frequency and maintain stability.
I have some other ideas too, some of them in the oposite way. (small motherboards with build in low power processors like VIA, linear/batery suply and so on) but I am waiting on your opinions.
"Playback from HDD or system memory or flash avoids mechanical jitter of optical devices."
mmm.... as far as I know HDD playback have more jitter than CD playback.
Hartono
mmm.... as far as I know HDD playback have more jitter than CD playback.
Hartono
Hello Hartono,
Well, I've obviously heard the other way around 🙂, altough my sources may not be as good as yours. The thing is that ripping a cd with EAC allows you to extract every single bit available on that CD even if this requires more spins for the cd-rom unit. Ok, even if HDD may have more mechanical jitter (although I don't see why), playing from the memory resolves this. Again, soundcheck in the thread I've mentioned above obtained some great results by properly configuring his laptop.
You've mentioned some P4 boards using oscon capacitors. That's quite intersting, could you name some brands?
Well, I've obviously heard the other way around 🙂, altough my sources may not be as good as yours. The thing is that ripping a cd with EAC allows you to extract every single bit available on that CD even if this requires more spins for the cd-rom unit. Ok, even if HDD may have more mechanical jitter (although I don't see why), playing from the memory resolves this. Again, soundcheck in the thread I've mentioned above obtained some great results by properly configuring his laptop.
You've mentioned some P4 boards using oscon capacitors. That's quite intersting, could you name some brands?
"You've mentioned some P4 boards using oscon capacitors. That's quite intersting, could you name some brands?"
It was a few years ago when I started using OsCon, then P4 was relatively new, over a couple years (I think around 2001-2003) I noticed some board IIRC made by Intel do use OsCon. I even planned to buy some broken board to take out the OsCon, but never (never had the time to find)get around finding any broken ones. The oscon in the board is low voltage 5V some even lower , 3.3V if I'm not mistaken, but the capacitance is pretty high.
Hartono
It was a few years ago when I started using OsCon, then P4 was relatively new, over a couple years (I think around 2001-2003) I noticed some board IIRC made by Intel do use OsCon. I even planned to buy some broken board to take out the OsCon, but never (never had the time to find)get around finding any broken ones. The oscon in the board is low voltage 5V some even lower , 3.3V if I'm not mistaken, but the capacitance is pretty high.
Hartono
I've only briefly skirted around the subject of PC audio, but a PC is a terribly noisy environment and could easily create jitter on an outgoing digital signal.
I think the merits of PC audio become worthwhile using an external DAC, with its own clock, a buffer and a reclocker of some description, to synchronise the data in an attempt to reduce jitter - away from the noisy environment for the PC.
PC motherboards are incredibly well designed things for what they are, considering PC bus speeds these days, they often have plenty of decoupling, sufficient to allow the silicon to drive buses and USB ports and the like. I'd just use them as is, and concentrate on tidying up the I2S, once you've got hold of it.
I think the merits of PC audio become worthwhile using an external DAC, with its own clock, a buffer and a reclocker of some description, to synchronise the data in an attempt to reduce jitter - away from the noisy environment for the PC.
PC motherboards are incredibly well designed things for what they are, considering PC bus speeds these days, they often have plenty of decoupling, sufficient to allow the silicon to drive buses and USB ports and the like. I'd just use them as is, and concentrate on tidying up the I2S, once you've got hold of it.
I just googled and found this example:
"If you take a closer look at the Albatron , all the capacitors are using high quality server grade OS-Con capacitors. It also features dual power support (notice the two ATX power connectors on the lower left?), a digital clock generator for the best overclocking stability and DirectFET power transistors."
http://www.hardwarezone.com.ph/articles/print.php?cid=22&id=1846
scroll down to albatron section motherboard type PX975X
Hartono
"If you take a closer look at the Albatron , all the capacitors are using high quality server grade OS-Con capacitors. It also features dual power support (notice the two ATX power connectors on the lower left?), a digital clock generator for the best overclocking stability and DirectFET power transistors."
http://www.hardwarezone.com.ph/articles/print.php?cid=22&id=1846
scroll down to albatron section motherboard type PX975X
Hartono
Just found this, sort of puts more accurately what I was trying to say...
http://en.wikipedia.org/wiki/Decoupling_capacitor
http://en.wikipedia.org/wiki/Decoupling_capacitor
If you look carefully enough, you'll probably find at least half a dozen gapped or air-core inductors on the motherboard, not to mention the radiation from the high power multi-phase converter next to the CPU.
Who cares if you just inhaled a Carbon-14 atom - you should be more worried about the nuclear reactor that you're standing on!!
Who cares if you just inhaled a Carbon-14 atom - you should be more worried about the nuclear reactor that you're standing on!!
I'm afraid you are so right... I was trying to get some more ideas/opinions (maybe solutions about this). I was aware about these problems on desktop motherboards but I have no experience with laptops... it seems the situation is the same though.
Another solution could be a diy FPGA/MCU board with enough horse-power for drc and brutefir and of course sata/ide conectivity. This would allow custom low noise regulators and so on.... But this is getting to science fiction even for me 😀
- Status
- Not open for further replies.