Hi, some of you are over estimating how much power a PC uses. A fully modern, high end gaming PC will draw ~250W from the wall typical when under load. But you can get silly and push this up to 400W or more quite easily with multiple graphics cards. About 85% of this 250W is drawn from the 12v power rails I think.
The PC I'm using right now should be drawing around 15W (not including monitor of course). So I expect it will be drawing around 20W or so from the wall (I'll be surprised if its more than 25W)
The PC I'm using right now should be drawing around 15W (not including monitor of course). So I expect it will be drawing around 20W or so from the wall (I'll be surprised if its more than 25W)
cowanrg,
I'm also using my PC for all my audio too and am very happy. I had also (briefly) thought of a linear supply, but the thought of making something with bang-on 3.3v, 5v and 12v is too daunting.
Perhaps, however, they may be a way to mod an existing decent quality SMPS to make it cleaner/quieter? -- upgrade/add caps, shielding etc?
I'm currently using a Tagan 380w PSU, which is very nicely made and very quiet. I've compared to an older Antec PSU i had, and the only difference I could notice in audio was the fan noise....
I'm also using my PC for all my audio too and am very happy. I had also (briefly) thought of a linear supply, but the thought of making something with bang-on 3.3v, 5v and 12v is too daunting.
Perhaps, however, they may be a way to mod an existing decent quality SMPS to make it cleaner/quieter? -- upgrade/add caps, shielding etc?
I'm currently using a Tagan 380w PSU, which is very nicely made and very quiet. I've compared to an older Antec PSU i had, and the only difference I could notice in audio was the fan noise....
Actualy a linear power supply could prove to have a sense if used with boards like this:
Intel Mini-Itx
VIA mini/nano/pico-ITX
All of these are enough for audio playback. Not for complex DSP like FIR filters and so on but enough for common audio aplications.
Here you may seen the poer consumption of a modern desktop CPU on 45nm technology. Now imagine the consumption for the mobile versions.
I think that used in conjuntion with small boards and with small power demandings, a linear PSU could proove worthing. Of course remains the problem of the the SMPS on the board, although I suspect that boards in the nano-itx format are using just the regulation provided by the powersupply unit.
Intel Mini-Itx
VIA mini/nano/pico-ITX
All of these are enough for audio playback. Not for complex DSP like FIR filters and so on but enough for common audio aplications.
Here you may seen the poer consumption of a modern desktop CPU on 45nm technology. Now imagine the consumption for the mobile versions.
I think that used in conjuntion with small boards and with small power demandings, a linear PSU could proove worthing. Of course remains the problem of the the SMPS on the board, although I suspect that boards in the nano-itx format are using just the regulation provided by the powersupply unit.
You can get above average quality computer power supplies that are ultra quiet (acoustic noise). Check out http://www.zalman.co.kr/ENG/Main.asp
Also with a conventional transformer you loose the benefits of active PFC.
Also with a conventional transformer you loose the benefits of active PFC.
Maybe that i misunderstand the datasheet i been reading .
(afterall there is a little difference between mcu or cpu)
But still the discussion intrested me and i looked up my own processor,s datasheet .
http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/30430.pdf
At page 91 it states Vid & Idd is 1,30 volt @ 80 Amperes .
That makes a maximum thermal design power of 110 watt's .
The processor i have on my motherboard is a low power model but still consumes 65 Watt @ 1,30 Volt .
But i must say the idea is nice ,on one of my first pc's i placed a lineair powersupply .
(i blew up the smps by accidently switching it to 110 Volt)
But then again that was a 4,77 Mhz XT with an amazing 5 megabyte hard disc
(afterall there is a little difference between mcu or cpu)
But still the discussion intrested me and i looked up my own processor,s datasheet .
http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/30430.pdf
At page 91 it states Vid & Idd is 1,30 volt @ 80 Amperes .
That makes a maximum thermal design power of 110 watt's .
The processor i have on my motherboard is a low power model but still consumes 65 Watt @ 1,30 Volt .
But i must say the idea is nice ,on one of my first pc's i placed a lineair powersupply .
(i blew up the smps by accidently switching it to 110 Volt)
But then again that was a 4,77 Mhz XT with an amazing 5 megabyte hard disc
Most laptops nowadays use 65w or 90w power supplies. The mobile workstations can indeed use over 100w, however.Eva said:
An embedded system would be a better choice for a linear power supply. But there's no point. Switching supplies have advanced enough that they are even used in very sensitive instrumentation equipment.
I actually built a simple linear regulator for a PC,running from a 12V SLA battery once,kind of for fun.
It was just a simple AT-style supply,running an older Pentium 133mhz system. It worked okay,and ran from the 17AH battery for a couple hours.Certainly not efficient,but it was kind of fun and interesting.
Linky
It was just a simple AT-style supply,running an older Pentium 133mhz system. It worked okay,and ran from the 17AH battery for a couple hours.Certainly not efficient,but it was kind of fun and interesting.
Linky
A side order of
Hey there cowanrg, I'm interested in
Are you using Linux, or a Windows OS?
What about tweaking the regulators on the motherboard, and Fan isolation?
Fun with oil? http://www.markusleonhardt.de/en/oelrechner.html
Use Mineral Oil, vegoil goes off.
Mikee55
Hey there cowanrg, I'm interested in
Are you using Linux, or a Windows OS?
What about tweaking the regulators on the motherboard, and Fan isolation?
Fun with oil? http://www.markusleonhardt.de/en/oelrechner.html
Use Mineral Oil, vegoil goes off.
Mikee55
I'd have thought a small linear power supply or two would do the trick nicely. +/- 12V to power the op-amps and +5V, 3.3V etc. for the DAC chip. This might bring good gains, but you'd obviously have to lift some resistors or cut tracks to insert the power feeds. Not inefficient but good gains available for sure.
I upgraded some psu caps and shorted some DC blocking caps on my Soundblaster X-Fi and it got better. I've also heard one with an aftermarket clock and LM4562 SMD op-amps installed and it sounded very good indeed, almost as good as some half-decent modern CD players (with a bright, upfront, "modern" sound).
I've not done anything else with mine because I started to use my Yamaha DSP-E800 for a DAC, using Dolby Digital Live for a permanent DD 5.1 connection. The Yamaha sounds a little better than the X-fi, and isn't even modified yet. It then feeds my pre-amp at line-level.
Simon
I upgraded some psu caps and shorted some DC blocking caps on my Soundblaster X-Fi and it got better. I've also heard one with an aftermarket clock and LM4562 SMD op-amps installed and it sounded very good indeed, almost as good as some half-decent modern CD players (with a bright, upfront, "modern" sound).
I've not done anything else with mine because I started to use my Yamaha DSP-E800 for a DAC, using Dolby Digital Live for a permanent DD 5.1 connection. The Yamaha sounds a little better than the X-fi, and isn't even modified yet. It then feeds my pre-amp at line-level.
Simon
A motherboard will have at least 4 switching supplies on board, so I don't know what a linear or battery supply will achieve - also fan noise is a non-trivial issue here. Regulation on the 3.3/5V lines need to be 1% with very high slew rates. Batteries are feasible, but any reduction in charge will need to be swiftly dealt with. Sagging supplies can corrupt data and damage hardware.
In pro audio, galvanically isolated digital and balanced (digital or analog) connections are used to reduce the problem as far as possible, the best sound interfaces in the pro world use proper breakouts, firewire/USB and external supplies, to delink them from the PC's.
I have a few PC sources, but the quietest and best sound comes from pushing the signal through a pulse transformer (almost all pro cards use one for their coaxial output) to an external DAC. I am hopeful that the performance will improve substantially when connected using I2S to a Firewire interface with its dedicated power supply.
For now, the Corsair VX450 is good enough for powering my rig, though I still think fan noise is too high. It has about 30mV ripple on the low voltage rails, and twice that on the 12V rails. Frankly, quite pathetic compared to a linear supply, but not bad when compared across PC supplies.
Finally, I don't know if hiss in the speakers is the right way to measure the effectiveness (or not) of any audio source. I've almost always experienced a huge improvement when tweaking only the soundcard supply, so if you still use a PCI card for D/A (I don't), try lifting the supply lines for the DAC and opamps off the board, or cutting a few tracks and using tiny low-noise supplies for them. That is more realistic, though it does involve cutting tracks.
The E-mu1212m is a darling to mod like this, as it connects the analog board using a short cable, which can be hacked to separate the supply pins, and wire in a new supply. A regular 40-pin floppy cable is used, so one does not even need to touch the stock cable. Unfortunately it also transmits I2S, but some modding can help shield this line. AFAIK the Juli@ can also be fruitfully modded as the analog and digital are two separate boards. With some ingenuity, the two parts can be separated and new supplies hacked in. There may be other suitable candidates.
As for specifications, they don't matter. A onboard Realtek chip claims -100dB THD+N, which is true of you only add up 2nd and 3rd. The chip spectral analysis shows -100dB products all the way to 15th harmonic, which brings the distortion up to -70dB, at best, with lots of high order harmonics. Needless to say, it sounds quite nasty. Even the much-acclaimed Xonar DX from Asus falls in this trap, though it still offers -10dB or so improvement over the Realtek. In my limited RMAA testing of the card, it showed very good spec, but the spectrum was similar to the Realtek. Given that it uses the CS4398, supposedly a very good DAC, the results are not encouraging for PC audio in general. In general, I seem to think it is best to get the signal out of the PC at the earliest, while it is still data and not converted to any form of audio.
Not really sure Simon. I unfortunately have very limited test equipment at my disposal - primarily due to my highly mobile work situation, I move cities every few years so it's enough trouble to move guitars and speakers and stuff.
However better supplies can never hurt, they will help. As an example, the well-regarded m-Audio 1010 (the rack unit, not the LT version) uses a voltage doubler with two diodes for the negative supply and IIRC the modder realised significant gains by changing it out for a proper supply. The DACs were all outboard in this case, on the rack unit itself.
But for sure somebody with a scope can quickly examine the effect of power supply noise on jitter of a given SPDIF or I2S stream, or even the final analog signal. That should give us something to work with. At least, it does have an effect on the clock itself, unless the interface itself operates asynchronously of the audio signal - such as Firewire or USB 2.0 with custom devices. This can separate the DAC clock from the PC's supply influences on clock accuracy.
I'm not sure how that works though. A CPU operates at over 1GHz or so, so clock jitter usually will have some performance issues on the CPU and core process itself, far more so than on a 192KHz (at best) audio stream. But whatever I read and try and experience, it seems that a modern motherboard is a EMI disaster, and to convert a data stream to audio as far outside the PC as possible is the best way (so even the 'controller' to be externally located and galvanically isolated). Even my poor little Alien DAC is a demonstration of this, it is easily able to come close to the performance of the much more expensive Delta 66 in its stock form. so yes, I believe a power supply may have an effect. How much, and whether it will be audible I don't know.
I don't believe in the shielding myself, never really worked for me except to make the boards nice and toasty. But I'm willing to look at some proof that it works, as I would like to believe it does.
However better supplies can never hurt, they will help. As an example, the well-regarded m-Audio 1010 (the rack unit, not the LT version) uses a voltage doubler with two diodes for the negative supply and IIRC the modder realised significant gains by changing it out for a proper supply. The DACs were all outboard in this case, on the rack unit itself.
But for sure somebody with a scope can quickly examine the effect of power supply noise on jitter of a given SPDIF or I2S stream, or even the final analog signal. That should give us something to work with. At least, it does have an effect on the clock itself, unless the interface itself operates asynchronously of the audio signal - such as Firewire or USB 2.0 with custom devices. This can separate the DAC clock from the PC's supply influences on clock accuracy.
I'm not sure how that works though. A CPU operates at over 1GHz or so, so clock jitter usually will have some performance issues on the CPU and core process itself, far more so than on a 192KHz (at best) audio stream. But whatever I read and try and experience, it seems that a modern motherboard is a EMI disaster, and to convert a data stream to audio as far outside the PC as possible is the best way (so even the 'controller' to be externally located and galvanically isolated). Even my poor little Alien DAC is a demonstration of this, it is easily able to come close to the performance of the much more expensive Delta 66 in its stock form. so yes, I believe a power supply may have an effect. How much, and whether it will be audible I don't know.
I don't believe in the shielding myself, never really worked for me except to make the boards nice and toasty. But I'm willing to look at some proof that it works, as I would like to believe it does.
Thank you for your considered reply. I'm surprised to hear about the shielding having little or no effect in your experience. I did read someone's mods to an X-fi where the modder found big gains by using some kind of shielding material over the whole card. He said the sound became a lot more relaxed and laid back. Sounds good to me!
But he was using the analogue outputs.
Simon
But he was using the analogue outputs.
Simon
A friend of mine found something similar. He first clipped the component leads flush with the back of the card, then covered it with a layer of plastic tape and then a layer of metal tape connected to the mounting bracket. Of course, he used the analog outputs as he used it with his hybrid amplifier. He says the SNR improved a little bit. As expected, there is no difference for digital.SimontY said:
Even better is to use fiber. But fiber is more fragile and more expensive than good coax.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.