Member
Joined 2009
Don't motherboards already have very intricate circuits to keep the voltages steady? A fluctuating ground will be pretty bad in a computer, right?
Can somebody post a link to a comprehensive study on this topic? I'm interested on seeing measurements of how much a switching power supply can affect the sound.
Can somebody post a link to a comprehensive study on this topic? I'm interested on seeing measurements of how much a switching power supply can affect the sound.
It's a very easy experiment to do on your own...hook up the analog output of a laptop to your sound system and listen to music while running on battery power. Then plug it into the SMPS power brick. You will usually hear an added overlay of squealy hash from the SMPS supply that is not at all subtle. Go back to battery power and the noise will be greatly reduced.
On computers with cheap internal sound cards, I've often heard very audible sounds coming out of the headphone as soon as I moved the mouse. Depending on the volume setting, my laptop with audio chips on the motherboard frequently produces soft high-pitched sounds when it is supposed to be silent. When you try recording with internal sound cards using a programme with a good level meter, you rarely see it dropping below
-50 dB quasi-peak while the CODEC chips are specified for 85 dB(A) or higher. There is always a 12 dB or so difference between quasi-peak and A-weighted RMS, but 35 dB difference clearly indicates that there is some interference problem.
So yes, the internal interference can definitely mess up the audio performance.
About spurious signals in the megahertz range: these become problematic when they produce mixing products in the audible range. For example, if your DAC is a sigma-delta modulator, and it almost certainly is, interference on its reference voltage at just about any frequency will mix back out-of-band quantisation noise into the audio band.
-50 dB quasi-peak while the CODEC chips are specified for 85 dB(A) or higher. There is always a 12 dB or so difference between quasi-peak and A-weighted RMS, but 35 dB difference clearly indicates that there is some interference problem.
So yes, the internal interference can definitely mess up the audio performance.
About spurious signals in the megahertz range: these become problematic when they produce mixing products in the audible range. For example, if your DAC is a sigma-delta modulator, and it almost certainly is, interference on its reference voltage at just about any frequency will mix back out-of-band quantisation noise into the audio band.
Member
Joined 2009
Ah well...I see what you guys are saying about the bad enivornment. I guess it makes sense. I think I'll wait on it for a while...
But...if I do build one somewhere down the line...
I like the USB interface because although I have an optical SPDIF port, I'll then need some sort of transciever amongst other things. USB is just more versatile anyway.
I know there are a ton of USB DAC chips. Surface mount is obviously fine. I'll either just make a 100% SMD board, try to make a dual sided board, or use a SMD adapter to get through hole.
Again, I'm not going to build it now, but just to quench my curiosity. Is something like a PCM1704 any good? Or are there better. Simplicity would really be a plus.
Thanks.
But...if I do build one somewhere down the line...
I like the USB interface because although I have an optical SPDIF port, I'll then need some sort of transciever amongst other things. USB is just more versatile anyway.
I know there are a ton of USB DAC chips. Surface mount is obviously fine. I'll either just make a 100% SMD board, try to make a dual sided board, or use a SMD adapter to get through hole.
Again, I'm not going to build it now, but just to quench my curiosity. Is something like a PCM1704 any good? Or are there better. Simplicity would really be a plus.
Thanks.
Maybe someday they'll come up with a sound card that has a discrete output stage, but I don't think that's going to happen any time soon.
Of course now that I've said that, Audio Research is going to come out with a PCI express version of their DAC8 that I can pine for. Oh, if only it were true.
Of course now that I've said that, Audio Research is going to come out with a PCI express version of their DAC8 that I can pine for. Oh, if only it were true.
For your information, inside laptops are lots of SMPS that convert the battery voltage (12-19V) to the multitude of voltages necessary (0.8-1.2V variable for CPU, 1.8V for RAM, 3.3V for PCI bus, 5.0V and 12.0V for USB and HDD). So no, your "experiment" shows nothing.
Last edited:
Shows that laptops running off wall-power SMPS are a lot noiser than when not...hardly nothing!
I have a plan! Go with me on this one....
Remove the sound card from the PC. Power it up externally from a linear supply. See if one can hear any difference? Or try shrouding the card with foil? It'd still be connected to the possible noise sources inside the computer however since it'd still have a bunch of pins attached to the motherboard.
I'm guessing most of you are worried about noise in terms of it causing corrupted and delayed packages? Has anyone done some measuring of this and seen if the switching is interfering?
There must be something in the EMI regulations that states the switching noise shouldn't be allowed to expand that far out of the supply, so it doesn't interfere with the buses and so on?
Is it really just the noise people are trying to get away from or are the external DAC noticeably clearer sounding?
My soundcard is an on board one, in the cheapest possible packard bell I could find (being a student at the time).
The cable on my HD650's just went, so that needs swapping. The amplifier I was using is gone, so I'm now looking at making something with the LME49xxx's. In for a penny in for a pound I guess... the DAC
Remove the sound card from the PC. Power it up externally from a linear supply. See if one can hear any difference? Or try shrouding the card with foil? It'd still be connected to the possible noise sources inside the computer however since it'd still have a bunch of pins attached to the motherboard.
I'm guessing most of you are worried about noise in terms of it causing corrupted and delayed packages? Has anyone done some measuring of this and seen if the switching is interfering?
There must be something in the EMI regulations that states the switching noise shouldn't be allowed to expand that far out of the supply, so it doesn't interfere with the buses and so on?
Is it really just the noise people are trying to get away from or are the external DAC noticeably clearer sounding?
My soundcard is an on board one, in the cheapest possible packard bell I could find (being a student at the time).
The cable on my HD650's just went, so that needs swapping. The amplifier I was using is gone, so I'm now looking at making something with the LME49xxx's. In for a penny in for a pound I guess... the DAC
I'm not bothered by it at all. I'm trained as a digital designer, its my bread and butter. However the noise gets into audio kit and that is bothered by it.
OK so then I'll say I sometimes notice I'm inclined to turn the system off.
The issue here is not that the laptop becomes noisier, it doesn't. Rather with the mains power connected there's now a loop in the system for the noise (from both laptop and SMPSU) to be conducted into the amp. That I think is where the squealy hash arises from.
So the next experiment is an isolation transfomer, then the external smps. Still a notebook cant sound that good, Odds are its monopole power supply opamp running at 5v with some tantalum 10uf cap coupled output.
True that the notebook has an internal smps, if it didnt, how would the 19v magically get converted to 3.3v,5v and all the little 1.5v for the ram and chip. At least they are filtered out the wazo with coils, just like t-amps.
True that the notebook has an internal smps, if it didnt, how would the 19v magically get converted to 3.3v,5v and all the little 1.5v for the ram and chip. At least they are filtered out the wazo with coils, just like t-amps.
I remember powering a TA2020 based amp from my PC-SMPS while the PC was on, it sounded nasty, so much noise. When powering the same amp from another non PC (Meanwell) PSU, it sounded a million times better...
As for laptop sound quality, it's a shame more of them don't feature TOSLINK outputs, the only ones that do I've found are Macbooks, they had TOSLINK outputs in 2006, yet I could go out and buy a brand new Dell or Samsung laptop now and it would still lack this feature...
As for laptop sound quality, it's a shame more of them don't feature TOSLINK outputs, the only ones that do I've found are Macbooks, they had TOSLINK outputs in 2006, yet I could go out and buy a brand new Dell or Samsung laptop now and it would still lack this feature...
As for laptop sound quality, it's a shame more of them don't feature TOSLINK outputs, the only ones that do I've found are Macbooks, they had TOSLINK outputs in 2006, yet I could go out and buy a brand new Dell or Samsung laptop now and it would still lack this feature...
It's probably going to be less and less popular. The newer formats for audio won't even pass through toslink iirc, you have to use HDMI to get DTS-HD and such.
Its not just the SMPS that create the nolise within a PC, all the clocksn operating the various sections (DDR memory, Ethernet, USB etc) create noise as does the numerous lines switching simultaneously (simultaneous switching noise), keeping everything working inside is signal integrity ans stopping it going outside (and avoising outsidwe influence) is EMC. Adding tin foil is not going to help, the noise is present on the supplies and ground. We have observed the audio out of some digital set ups (its not just PCs, any high speed digital based system with have some noise) wher you get the analogue signal, but if you zoom in on the scope you can see the superinposed digital noise, up in the high MHz. I would hope most of is filtered by the next stage of bandwidth limited analogue circuitry to some extent, but it is present.
I tried a sound card (ASUS Xonar Essence STX) with even better specs.
ASUSTeK Computer Inc. - Multimedia- ASUS Xonar Essence STX Looked fabulous on paper. Excellent component parts. Sent it back. It could not touch my external DacMagic.
100dB SNR for a 24 bit DAC hardly represents SOTA performance, a good 24 bit DAC chip is capable of at least 120dB SNR, and I am not sure how much I trust the mobo designer to even achieve something close to the performance of the chip set in the hostile environment of a motherboard. The theoretical noise floor for 24 bit quantization is slightly better than -140dBfs which is rarely if ever achieved in practice...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.