|
Most PSU's deserve to be grey: like second-rate politicians they drone on drably while you're trying to get some work done; likewise
they're not that great at handling power.
On the plus side, that grey box that comes with your store-bought system or replacement case is dirt-cheap & does the dullest of jobs.
High-end systems need a bit more juice though, & there are quite a few users who like or need to be able to hear themselves think as they work: between them these two markets make a fair-sized niche which
the Q-Systems PSU's aim to fill.
What's special & how it's done: This ATX PSU is "AMD approved" up to & including the 1GHz CPU's & is also
designed to be quiet: "Athlon" PSU's don't produce any more juice than an ordinary one rated at the same Watts; what they do is provide a higher proportion of the rated Wattage from the +3.3/+5v rail - which is what your mobo, CPU[s], AGP-accelerator, PCI-cards, & SDRAM all feed
off. This PSU is rated at a generous combined 165 W from the +3.3/+5v rail.
Your drives [fixed & removable] & fans are fed off the +12v rail, which is here rated at 120 W.
These rails are just well-regulated transformers, converting domestic current to the right voltages for your computer; a by-product of this conversion is waste heat, & inside every PSU are large alloy
heatsinks, one to each rail & more-or-less well-cooled by air drawn from inside your case & exhausted out the back by an 80mm case-fan - which, as a bonus, scavenges waste heat from your case.
The greater the quantity of current transformed & drawn, the greater the amount of heat needs to be transferred from the heatsinks to the cooling airflow. This PSU uses a variable-speed low-noise double
ball-bearing fan, which at all the loads we were able to put the PSU under runs real slow & very quiet - you just can't hear it at any load over the [quiet] 26db of a Golden Orb. Like a well-sorted 'clockers' case, the interior of the PSU offers a relatively unobstructed path to the cooling
flow & channels this flow carefully past & over the heatsinks:
- as you see, the whole internal-facing panel of this PSU is pierced for air-intake, & all major heat-producing components are clustered in front of & near the fan [the small PCB over & beyond it
is the speed-controller]. Design detailing goes as far as the fan grille & case-opening being optimally specified & made for minimum obstruction & noise, & the PSU having 5 Molex-plug output lines on fairly long leads, including 2 x floppy-plugs.
|
Test System Loads:
+3.3v/+5v rail: 2 x 366@550 Celerons [2.05/2.10 vcores] - around 70 Watts total - we chose .25 process Celerons for this test due
to their higher current-draw than FC-PGA's; MSI 694D-ProA mobo with 256Mb 150CAS2 HSDRAM; Matrox Millennium G200 SGRAM accelerator; 2 x PCI SCSI-hosts - total draw approx 125 Watts maximum.
+12v rail: 2 x 7.2k rpm IBM GXP 30Gb HD's [RAID0-array]; 10k rpm 9.1Gb U2W SCSI HD; 5.4k rpm UW SCSI HD; 2 x floppies; 2-3 x
[varies] SCSI CD-devices - again around 120 Watts notional maximum draw - plus various fans & external devices.
All in all this is about as much as you can cram into a GlobalWin 802ATX midi-tower - not a small case. Many if not most experienced users would recommend a 400W+ PSU for this much power-hungry kit &
most users make about half these demands at most on their PSU.
Test: [duration 3 months] The major stress on a PSU is at start-up - that's when a component or fuse will pop if you've overdone
it, or when a big sag in the measured voltage will occur if the regulation is poor or if the thing's near its limits: as with most inexpensive IDE RAID, our Promise controller doesn't allow a staggered spin-up, & we tried to set both SCSI-hosts to spin up their drives as near in time as
possible.
Once the system is up & running, moments of high voltage stress would be when, say, transferring large files from a RAID-array to a 10k rpm U2W HD at the same time as running two instances of Prime95
[one to each CPU]. So, wearing our torturer's kit & cackling insanely, that's what we did.
We took repeated +5v & +12v voltage readings with two digital multimeters direct from PSU lines [not the relatively meaningless readings you can take inside an OS from a monitoring utility like MotherBoardMonitor
- these are moderated by the [in-] accuracy of the motherboard's monitoring IC & can tell you nothing about the all-important start-up performance of a PSU]. The readouts from both multimeters closely agreed throughout.
|
|
Startup 12v range
|
12.10-12.28v
|
|
W2K 12v [no load]
|
12.12v
|
|
W2K 12v [Prime95 x 2]
|
12.28v
|
|
W2K 12v [Prime95 x 2 & 540Mb image-file transfer]
|
12.28v
|
|
Startup 5v range
|
5.03-5.07v
|
|
W2k 5v [HLT idle]
|
5.07v
|
|
W2K 5v [1 x Prime95]
|
5.03v
|
|
W2K 5v [2 x Prime95]
|
5.01v
|
|
W2K 5v [Prime95 x 2 & 540Mb image-file transfer]
|
5.01v
|
|
BTW - voltages as reported from the monitoring IC to MBM5 were all approx 0.1v wrong [low] - no relection on MBM - just GIGO
|
Analysis:
The 12v rail effortlessly coped with the loads we put it under.
The +3.3/+5v rail was loaded to about 75% of its rated combined limit: the drop in +5v when one then two instances of P95 are run is closely paralleled by MBM5
readings of vcore drops - each CPU's vcore independently drops 0.04v under 100% load [then holds solid]. This drop is sufficient to affect overclocking stability on this very tricky motherboard & means each CPU's default vcore must be set 0.05v higher than strictly necessary to give
complete stability.
Having said this; the system thus set will run stable 24/7 multi-tasking under continuous 100% load.
Summary: This PSU performs as specified: it is very quiet & has an amply high-enough output from the +3.3/+5v rail to run a
IGHz AMD CPU or 2 x Intel FC-PGA's satisfactorily, together with all cards & memory you're likely to have: the 12v rail will run all the 12v devices you've places to fit into a midi-tower.
Under the quite high total load at which we tested it, the PSU's supply to each CPU sagged 0.04v under 100% CPU load; we noticed a smaller but similar drop when loading a single CPU [to 32.5 Watts]. This
is a normal enough with any PSU; but may affect overclocking performance if you load this PSU as high as us & are running near the vcore adjustment limits of your CPU[s].
We have no reason to believe a 400Watt+ PSU would perform better in our system - unless such a PSU had superior regulation/smoothing circuitry on the +3.3v/+5v rail. The [theoretical] extra power,
especially on the 12v rail, is unnecessary & would merely provide more waste heat - & noise.
We started off saying most PSU's deserve to be grey: this one deserves to be . . . . hmm . . . green?
Burningissues would like to point out that opening a PSU [as above] will both void your warranty & may expose you to dangerous voltages.
We would like to thank Glenn at www.quietpc.com for encouraging us to try to break this product. We failed.
|
|
