Almost there

[Sigbjørn Lund Olsen]

Alright, about half a year has gone since I started looking at getting a new computer, and picking up hardware and stuff to put in it. The final(?) specs:

Evercase 4252 (or some such number - I always get it wrong) with blue trim
Aopen AK-79-1394
2x256MB PC3200 GEiL RAM "Golden Dragon"
Barton 2500 MHz
120GB Barracuda V
An old 2x DVD player from Philips
Aopen 40x12x48x CD-RW
Floppy drive
Matrox Millenium P650
MS Internet Keyboard
MS IntelliMouse Explorer 3.0
Philips 109S40 19" monitor
old SB! Live 1024 for kicks, will probably use on-board nVidia SoundStorm primarily
Seasonic 460W Super-Silencer
3x Zalman variable voltage thingies for fans
3x Panaflo 80mm
Thermalright SLK800A
Arctic Silver paste
AcoustiPack Deluxe

Well. First things first. The case isn't that great. In fact, it's not great at all - and if this is the 'best' then there is no such thing as an even remotely good ATX case design in existance these days. The main problem with the Evercase is:

1. It's cramped. A little more space above the CPU would have done absolute wonders.
2. No MB tray. I like MB trays because no matter how airy a case is you simply don't want to be doing all these things inside the case. With an MB tray you get to do a lot more outside the case without worrying whether you're applying too much pressure. And when installing that Thermalright heatsink, you should be worrying
3. Stamped grilles. They should simply be gone, with a supply of 'circular' fan guards instead.
4. 3.5" bays. Any really good silent PC case doesn't need any. You want to have some sort of thing around the HD, while keeping it low and in the airflow of the case, so the 3.5" bays are just pointless.
5. There's no front door to stop optical drive sound. At least in my case, the optical drives, when spinning, are by far the noisiest thing inside the box, and they need to be dampened.

I'm not religious, but Jesus Christ, my CPU must be blessed by divine powers to survive the abuse of mounting that heatsink on it.

The Barton is currently underclocked at 150x11, down from 166x11, so not a whole lot. It's not generating much heat, and I've undervolted it down to 1.25v. Currently idles at around 34 degrees, peaking at 42-ish (running Sandra burnin, presumably there are more intensive burnins, though I don't think it'll be a problem). At any rate, a massive margin there. I could probably undervolt it more, too, I'm just not sure if I'm going to bother with a CMOS wipe. Laziness will prevail

My main headache is the HD. It's running at around 32-36 generally, which in my book is too close for comfort. AFAIK the Cuda V shouldn't go over 39, right? The problem is that I forgot to order one of those bottom mount thingies, so it's currenly hanging just a little over the airflow from the input fan (read: move-around-air-fan-in-the-front). If the case design had been done by an intelligent primate, it'd just have had 5.25" bays all the way down, problem solved. I'd probably not need a fan either since I could put it in one of those Zalman heatpipe mounts.

My second largest headache comes from the fans that are mounted to the case, ie "input" and output fans. I bought some gel thingies from Kool'N'Quiet, but they don't fit the Panaflos (apparently they're made for Papsts), so at the moment they're fastened with a pair of big screws each. They emanate a fain clicking noise while rotating, which I think may have something to do with this - they don't click then unscrewed. I need to order some other flexible mounts for them, basically. Due to this, I've kept the CPU fan running at a higher speed than I need it to - basically in the choice between audible low-quality noise and slightly louder higher-quality noise I'd rather have the second.

I'm also not convinced that the Cuda is silent enough. I'd rather have a regular swoosh than irregular clicks from a HD, so basically it will have to become more silent later on.

I'm happy with the Acoustipack, though it would be a lot better if the case design was more roomy, and this more accomodating for sound insulating materials.

The Super Silencer is really a fine piece of PSU. I'm not sure how much Dr.Cable actually improves airflow in my case, because there's so much in so little space (especially the flat cables get in the way), but it sure as hell looks a lot tidier

The CDRW is noisy, but I'll probably only use it for burning, and since the DVD is silent as a dead man (mostly) I think I'm pretty much sorted there. P650 is nice and passively cooled, with dual output and a wicked 2D crispiness you just can't get anywhere else than Matrox.

The worst and best buy I made was the monitor. It's got a fantastic image crispiness, and brilliant colour controls. It does however sometimes emit a loud high discrete tone, probably from resonance mechanics in the CRT array. Switching the update frequency eliminates the noise (untill it appears for that frequency). It is *very* annoying. It'll probably go before the next PC, though I imagine the next PC might appear within a year. Those 64-bit AMDs look delicious.

Thanks for the authorative help and resources made available to me here at SPCR. This is now my one-stop resource for all things PC hardware. I'll be joining F@H/G@H efforts in a little while, also putting my old P266/64MB laptop on 24/7 (seeing as I'll have little use for it now) folding.

Cheers,
Sigbjørn Lund Olsen

[EndoSteel]

Restore the CPU clock - 42 degrees is COLD .

[ruprag]

I support EndoSteel (my temps are 50 idle and > 65 when running burn)

also < 40C for a harddrive is no problem,
most harddrives are rated for 50C and if I recall correctly the cuda is rated for 60C (the cuda runs hotter than most other drives)

[Sigbjørn Lund Olsen]

I restored the clock just a little after posting that. Still running fine at 1.25v - you sort of have to wonder what the point of the 1.65v core voltage default really is.

[EndoSteel]

you sort of have to wonder what the point of the 1.65v core voltage default really is.

Undervolted CPUs lose stability at lower temps, probably that's the reason. Yours will most likely freeze at ~65c instead of 85 at nominal voltage.

[Rusty075]

Well. First things first. The case isn't that great. In fact, it's not great at all - and if this is the 'best' then there is no such thing as an even remotely good ATX case design in existance these days.

I don't think anyone would describe the Evercase as the "best" case for a quiet PC. Usually its described as "a pretty good cheap case"
All of the "problems" you have with it are things that you really should have noticed during your year of research.

I'm not religious, but Jesus Christ, my CPU must be blessed by divine powers to survive the abuse of mounting that heatsink on it.

Yeah, plenty of first-timers are surprised my the amount of tension put on the CPU. It's perfectly normal; the CPU isn't as delicate as you've been led to believe.

The Barton is currently underclocked at 150x11, down from 166x11, so not a whole lot. It's not generating much heat, and I've undervolted it down to 1.25v. Currently idles at around 34 degrees, peaking at 42-ish

Most of your wattage reduction is coming from the undervolt, so you were right to kick the speed back up to full. Your temps are fine, low actually. Too many people obsess over their temps. The rule is: if your CPU is below 60°-ish at load (idle temps don't matter) and you're stable, then everything is OK.

My main headache is the HD. It's running at around 32-36 generally, which in my book is too close for comfort.

Your temps are fine there too. For a 'cuda, anything under 40° is actually pretty good. I wouldn't worry unless you were in the upper 40's.

If the case design had been done by an intelligent primate, it'd just have had 5.25" bays all the way down, problem solved.

Ever seen a case with only 5.25" bays? There's a reason why you haven't: 99.99% of the buyers out there wouldn't want it. Read the articles on here about suspending your HDD's. That would both quiet them, and let you move them down into the airflow path better.

[Rusty075]

you sort of have to wonder what the point of the 1.65v core voltage default really is.

Undervolted CPUs lose stability at lower temps, probably that's the reason. Yours will most likely freeze at ~65c instead of 85 at nominal voltage.

I've never seen a correlation between dropping the vcore and a reduction in the max temp at which the CPU is stable.

The chip companies set the default vcore for a given speed CPU at the levels they do in order to ensure that the highest possible percentage of the chips produced will be stable, and thus sellable, at that speed.

The individual variation between chips means that most of them will still be stable when undervolted. How far you can drop the vcore while maintaining stability is dependant upon the individual chip (and mobo) that you have.

[MikeC]

3. Stamped grilles. They should simply be gone, with a supply of 'circular' fan guards instead.

The need to meet EMI regulations & agency approvals ensures that NO manufacturer will or can release cases that have blank holes in them. The wire fan grill does not meet specs for such regulations & approvals.

[zuperdee]

The need to meet EMI regulations & agency approvals ensures that NO manufacturer will or can release cases that have blank holes in them. The wire fan grill does not meet specs for such regulations & approvals.

I'm curious now--in that case, why is it that some PSUs like the Antec SL300S and the True380S come with wire fan grills?

[MikeC]

I think it has to do with the fact that in cases, those wire grills would be removable -- ie, the user can easily discard them. In the PSU, generally it is pre-mounted and "sealed in" so to speak. That's just a guess.

Re -- the EMI for cases: I'm really just parroting a case manufacturer's rep met at a trade show. Maybe he knew nothing & was giving me BS.

On the other hand, have you seen a single case that uses wire grills & not some kind of cutout mesh? It made sense to me whewn the fellow told me, I didn't question it at the time.

[Rusty075]

Mike makes a good point. If there wasn't a good reason not to, by now somebody would be making one.

The only cases I've seen that come with wire grills are the clear acryllic ones. Obviously they've given up on getting UL-listed from the very beginning.

(if you read the manufacturer's literature, almost all of them have disclaimers stating that they are for use as a "demonstration tool" only, and not as a regular PC. It kinda reminds me of companies that sell battery-operated appliances as "neck massagers")

[EndoSteel]

I've never seen a correlation between dropping the vcore and a reduction in the max temp at which the CPU is stable.

I haven't yet gathered enough statistics to be sure about all of them, but both of my processors (AXP 2100+ and a CuMine P3 600) do have this 'feature'.

[silvervarg]

My experience with undervolted processors is that they get unstable at cold temps. So the biggest problem is to make them cold boot the computer or bring it out of S3 state.
Soon we are going back to the ages with old disel cars where you had to turn on pre-heating and wait 1 minute before you could start the car (during winter time).

[Rusty075]

silverarg's cold-start trouble may be more motherboard than CPU. I've had a couple of mobo's where the vcore fluctuates for the first minute or 2. (it probably fluctuates alot more during POST and BOOT, when you can't monitor it, as the peripherials put sudden draws on the voltage regulators) Because you're operating the CPU closer to its limits, it'll be much less tolerant of these fluctuations.

To test my hypothesis you could set MBM to start at start-up, and record a log of the voltages. If your board is like mine, the graph of the vcore will look like a seismograph for the first minute, then settle down.

[Sigbjørn Lund Olsen]

Hmm.

Well, I experienced the weirdest problem, as it was. The comp was working perfectly allright for a while, but at the moment I've been working on getting my home network working with internet connection sharing, and so I put it in the 'office room' in front of the TV there. And worked fine at 1.25v. Then I wanted to see a TV program yesterday, unplugged the computer and moved it away for the duration of the program. Then I plugged it back in and it wouldn't boot. It wouldn't boot at 1.3v either, but would at 1.4v. I also set the RAM CAS/RAS latencies to 2-3-3-5, or rather it used the settings stored in the RAM EEPROM chip. After that it's been giving me BSODs, so I've increased the CAS/RAS to 2.5-3-3-6 now, and the core voltage to 1.45v. It's better now, though I'm not sure if it's completely stable yet. As a consequence the CPU now is idling at 38 degrees. Need to check the load with 1.45v. But I thought it was a bit funny that it suddenly wouldn't boot after I'd unplugged it. Anyone got any good explanation? Was there some resident current that made it boot?

EMI certainly is a possible explanation for not using wire grilles, but frankly I'm doubting. Consider the Seasonic Tornade PSUs, which afaik are within EMI limits(?). It has an enormous amount of open area, and EMI, while interference from grille patterns should matter a bit, should also be dependant on the amount of exposed area, and the insulating properties of the materials. Also a number of PSUs *do* use wire fan grilles. Like my Super-Silencer. And its screws are on the outside, so it's not unscrewability that's the problem either. Furthermore, if EMI was seriously much of a problem you'd have a lot more aluminium, as it is by far one of the best metals for EMI-shielding.

Furthermore they could use stamped grilles better if they rounded the edges. Frankly this sounds more like: "It's more difficult for us to do this properly, it costs us a little more and frankly 99% of our buyers are dumb as shit so we're going to blame EMI."

Presumably some server boxes would have 5.25" bays all the way down (due to increased HD cooling needs), or perhaps some 'CD burner tower'... I dunno. I would much prefer a load of 5.25" bays. But I suppose its a buyers market... *However* there does seem to be enough of a market for 'silent' cases these days that they are actually being produced. So if there's enough of a market to make cases like the AcoustiCase and the Antec Sonata, and market them as silent cases, there *ought* to be a market for a case with only 5.25" bays.

[trodas]

Just short note about the HDD temps
BarracudaIV is rated that it's operational temps are up to 85 degrees. When i put it into non-ventilated removable ide rack, i hot 62 degrees during backuping Don't worry about HDD temps up to 65, maybe even 70 degrees.
Limit is 85, is there is still plenty of room.

About the CPU - try moving the machine back to see, if it boot with the undervolted CPU again. If does, then the voltage or phase into wallet where it boot probably differ in a way that the CPU need to boot from the new place, where it don't boot.
Sounds crazy, especially considering there are voltage regulators and so on, however i lack any other explaination.

General rule is - never touch anything that working well. It might change - to not working well