 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
24.18.240.176
In Reply to: RE: RAM / PSU posted by cics on January 01, 2008 at 08:17:30
At what point would it make sense to use linear, non-switching, power supplies instead? I realize they're big and expensive, but if you've figured out how to divide up the PC's power needs, and if we're in any case using relatively low amounts of power, would it make sense to employ an inherently quieter design for the critical parts?
Thanks for posting all this stuff, cics! Even for those of us operating at lower levels of perfection, your work has been a source of lots of good hardware and software ideas.
Follow Ups:
Artcheng also suggests this option. For me both the 12V1 (mobo 24 pin) and 12V2 (CPU 4 pin) are critical. Yes you both right but I'm thinking why stop at linear PSUs and why not go for the purest source, i.e. battery?
The idea is to have a form factor like a normal PSU (maybe longer). It houses 2 laptop batteries: each providing shielded harnesses for 12V1 & 12V2 respectively. With this low power consumption setup, batteries can last for a good few hours playing music. PSU has built-in charger with IEC power inlet and switch (Off, Charge, Auto). In Auto mode, charger kicks in when system is idle. If there's something like this I'd buy it immediately!
I hope this makes sense.
Form factor aside, does anyone know of a source for either linear regulated PSU or battery pack with shielded harnesses, that would replace switching PC PSU?
You can always use ordinary SLA batteries with a trickle charger -- you just need a switch that switches the battery or batteries between the application (on) and the charger (off). It's a lot bulkier than a normal PSU and there's an extra switch to turn, but it's technically simple -- I've done it for DACS and T-Amps.The question seems to be total power needs. Since you've moved the motors to another supply, the battery would be running the CPU, RAM, mobo, soundcard, right? Just poking at the numbers, for something built around a VIA board and chip, sipping a few watts, this seems pretty do-able. But for a fast CPU to run Secret Rabbit Code, and someone wanting say eight hours of continuous play, the amp-hours climb. I also don't know about battery longevity. There must be a battery-using subculture on one of these boards that knows this stuff better.
The question seems to be total power needs. Since you've moved the motors to another supply, the battery would be running the CPU, RAM, mobo, soundcard, right?
Correct. Total power would be ~30watts (underclocked E2140 CPU @17, RAM @5.4, mobo+s/card @8). RAM power consumption based on Kingston HyperX range with 5.4 watts reported in SiSoft's Sandra Memory Bandwidth Benchmark test. I haven't actually measured total consumption which could be less (as most devices like onboard sound, LAN etc. are disabled thus consuming less power). If someone has the XG Magnum 500 PSU, the power reading on its display would be the correct consumption.
But for a fast CPU to run Secret Rabbit Code...
No problem here. I run SRC at 96k and CPU remains underclocked. Vcc remains at 1.16V (giving 1200MHz) so no additional power load here.
cics - you say Sandra's Memory Bandwidth Benchmark reports the power consumption. I must be really tired from partying all last week, but I can't seem to find it. What version are you using? And can you be specific as to where I can find the power consumption numbers.
Thanks!
I'm using Sandra Lite: 2008.1.12.34
Select Benchmark Tab > double-click Memory Bandwidth > press Refresh icon to start test.
Look at the output report, a line entry reports power. Here's my Dell pc which has dual channel access for 2x1GB RAM (speed is 200MHz, cheap quality):
The entry I'm referring to is 'Chipset(s)/Memory Power' which shows 17.4W - with my Kingston HyperX it shows 5.4W.
Ahh...I was using an older version. I downloaded this new version and tried it with a couple of different PCs/configs. Do you think it's very accurate? I mean is it actually taking a measurement, or just looking up average values from a database depending on what brand you have? Also, that value includes the chipset AND memory, so I don't think it's fair to say your Kingston is consuming 5.4W. And I don't think there's RAM that consumes 17.4W. Most of that value is attributed to the chipset. My Kingston RAM shows it only consumes 2.1W according to the datasheet. And I tried the same 1GB stick of Corsair RAM on two different PCs and Sandra reported 15.5W on one PC and 5.4 on the other. Also, I compared that same stick either configuring it to DDR2800 w/2.1 voltage or DDR2400 w/1.8 voltage and the Sandra value did not change.
Such highs of 17watts is likely - the rule of thumb was 3 watts per 256MB giving 12watts for 1GB.
Sandra calculates whenever it recognizes the mobo chipset. From my understanding, this is the only way to determine RAM power consumption as it depends on RAM commands performed (which is what the software does and how RAM is implemented - the chipset). On datasheet specifications (Kingston KHX6400D2UL) you'll get 'operational' power consumption of 1.922watts or in your case 2.1watts - this doesn't include activity or RAM commands. See here on detailed power calculation. Vdd in your case is 2.1V.
Power consumed is based on manufacturer specifications (by command like IDD7) and other items. In your case with Vdd at 2.1V, power reported would not be correct. As a rough guide add the proportional increase in voltage to watts reported by Sandra. Sandra also adds the chipset power component - thats why you get different readings from different chipsets. The large difference is a surprise - have you tried contacting Sandra? My Dell uses Intel chipset giving 17.4watts total. What chipset does your PC that shows 15.5watts use?
OK - here's what I got:
With 1 stick of 1GB Corsair XMS2 (CM2X1024-6400C4)
on ASUS P5W DH Deluxe motherboard (Intel 975X chipset) = 15.5W
and on Biostar P4M900-M7 SE (Via chipset) = 5.4W
Incidently, I'm currently using 1 stick of 2GB Kingston (KVR400D2N3/2G) and, on the Biostar, Sandra reports 6.6W
I've swapped it out with the Corsair, only a couple of times, comparing both with the same settings (PC2-3200 3-3-3-8), but so far I like the sound of the Kingston better.
That makes sense. Intel chipsets use more power hence more heat.
Kingston HyperX quality is very good, my settings are here (PC2-4200 with RAM specified for PC2-6400 - still underclocked).
.
I don't have the technical chops to do it myself--at least without some help--but I am interested in powering a computer with SLAs.
I am using Red Wine Audio S-70.2 SLA powered amps, and this seems to me the way to go. Red Wine is coming with a battery-powered DAC/tube preamp this spring, which I will certainly try. That will leave only my computer running on AC.
Check out the Red Wine forum
http://www.audiocircle.com/circles/index.php?board=78
The advantages are that battery power is free of ac-line grunge and the inherent 60 cycle hum; it is free of the noise that is produced in rectifying and filtering AC; and SLAs allow very powerful current surges that get choked out by AC power supplies. This seems what the doctor order for an audio PC.
The S-70.2s each have two 12 volt SLAs. The amps put out 30 watts, and they will run 20 hours or more on a charge. The forthcoming preamp/dac is supposed to run 6-8 hours on a charge, keeping the filaments of two vacuum tubes hot.
I am satisfied customer of Red Wine Audio, but have no other connection.
db
| FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
||||||||||||||
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
