Improving the cooling: Shuttle Zen, first tinkerings

Info & chat about quiet prebuilt, small form factor and barebones systems, people's experiences with vendors thereof, etc.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Improving the cooling: Shuttle Zen, first tinkerings

Post by Copper » Sat Feb 28, 2004 8:48 pm

The first thing I've done is to put the fan on the outside. I removed the ICE fan housing, flipped the fan around so it blows the other way, and attached the housing to the outside. I remoeved the wire fan grill and if you look closely you'll notice I cut out the grill in the case that usually covers the ICE radiator. This eliminated the turbulence noise created both by the grills and by the fan being slow close to the radiator. There is more than enough room for a 25mm replacement fan using this configuration.

Click on image for a larger version.
Image

Please excuse the rather crude workmanship both above and below. I'm still tinkering at this point so I'm not spending time on craftsmanship details.

One of the things I noticed in the review was that the Shuttle Zen suffers Inlet Deficiency Syndrome (IDS for short). :) I attacked the bottom of the case with wire cutters and an old but sharp knife. I not only removed the seudo grills I enlarged the total area of each grill by about 2 fold.

Click on image for a larger version.
Image

The question is, of course, did this make any difference? Given what little temperature resources are available I have only the following to offer.

The air blowing out the fan is noticably stronger at the same voltage. The top of the aluminum case is actually cool to the touch. The hard drive (A Raptor 74GB) runs idle at 35C, down 6C from 41C, at the same 21-22C ambient. And I was able to run two simultanious sessions of CPUburn for approximately 20 minutes before the fan started to ramp. Previously I got about 1 minute, 2 tops. Both instances were using a 60C CPU temp threshold for the fan ramp.

Even with as much as I cut out of the bottom, the inlet still seems deficient. I will experiment more and post here as time goes by.

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Sun Feb 29, 2004 5:59 pm

I took everything apart and pulled the motherboard so I could work with some beter tools. This time I went after the case with a jigsaw equipped with a fine metal cutting blade. It made short work of the cuts.

I made the bottom inlets bigger:

Image


I also snapped a picture of the rear grill cut out while I had it apart. The grill can be cut out easily with wire cutters if you don't want to take the computer all apart, but if you will have it apart, a jig saw works much easier, faster, and is cleaner cutting.

Image

I've only had the computer running for an hour or so since getting it back together. So far the drive is running a little cooler. I want to give it more time to see if it climbs and I also want to run CPUburn to see if there is any appreciable difference there. I'll post later.

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Sun Feb 29, 2004 7:41 pm

Well, there are no appreciable differences that I can deduce with the second cuts that I made to the bottom of the case. The first cuts seem to be as large as needed to make all the difference. The hard drive is still running at 35C idle and dual simultanious sessions of CPU burn still ramps the fan at about 15 to 20 minutes. The ramp is only bumped one step (up to 6v from 5v or "low" from "ultralow") for 2 or 3 seconds then drops back down, then 10 to 20 seconds later gets bumped one step again. This, the same as with the first cuts.

So if you want to improve your airflow considerably, move the fan and housing outside, cut the case grill that covers the ICE radiator, and cut out an area approximately double the area of the underside grills. (Use a jig saw though. That wire cutter - knife technique I used was fugly!) :)

On to the fan!!

replay0
Posts: 56
Joined: Wed Feb 18, 2004 2:59 pm
Contact:

Post by replay0 » Mon Mar 01, 2004 2:22 am

You should try putting a large airfilter running the length of the base while you're still modding your system.

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Mon Mar 01, 2004 6:11 pm

replay0 wrote:You should try putting a large airfilter running the length of the base while you're still modding your system.
That's an idea replay0. I'm busy playing with the fan now, though. :)

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Mon Mar 01, 2004 6:18 pm

I moved the fan back inside, but this time I put it on the outside of the fan housing. With a little cable management and the right dvd/cd player there is more than enough room to use an 80x80x25 replacement fan with rubber grommets to wount the fan without screws. The fan in the pics is the stock fan. (I don't own another fan yet) :)

Image

Image

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Mon Mar 01, 2004 6:53 pm

This fan configuration does not cool as well as with it on the outside. I ran Dtemp and the drive is at 36C, up one degree - probably statistically insignificant. But there is a big difference with cpuburn. Running dual simultanious sessions netted a fan ramp in under 4 minutes, down from 15-20 minutes.

The external mount seems to have the greatest potential to run the lowest fan rpm, but with the drawback of exposing the fan, possibly making it louder even at the lower rpm. New fans should be here later this week or early next week. If I'm watching my numbers right, they should spin at about half the speed as the stock fan when voltage is matched (stock runs 4000 rpm at 12v, for instance. Most low noise replacements seem to be in the 2000 rpm range at 12v). Which leaves in question how well the fans will cool at 5 volts. I'm wondering if they'll ramp even with the processor idle. We'll see!

mas92264
Patron of SPCR
Posts: 659
Joined: Fri Sep 26, 2003 5:26 pm
Location: Palm Springs, CA, USA

Post by mas92264 » Mon Mar 01, 2004 9:36 pm

Hey Copper,

Nice pics and interesting experimental work. What motivated you to cut the holes in the bottom where you did? Was there a series of holes already in that location?

It'll be intersting to see what your results are when you install an 80x25 fan. Let us know how it turns out.

M

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Tue Mar 02, 2004 7:14 am

This is what it looked like in stock form:

Image

The picture is from silentpcreview's review of the Zen. There really isn't a lot of intake area stock. I kept the intakes in the same area assuming Shuttle felt there was good reason for them being where they were.

Hopefully I find a good fan, good as in little to no motor noise. The stock fan has a noisy motor. And hopefully it will cool at the lower rpms. I think whether it will cool or not is the big question. The slower fan is going to max out in rpms pretty close to where the stock fan is at 5 volts. The main problem I forsee is an inability to monitor cpu temps other than by going into the bios. By the time I get to the bios the processor has cooled.

I thought about putting an adaptor on the fan mount and installing a 120mm fan to improve airflow at low rpm but then I read that connecting a 120mm fan to the motherboard isn't a good idea. Apparently they draw more amps than one would want from the tiny motherboard circuits? Perhaps an adaptor with a 92mm fan will be the sweet spot for the Zen.

We'll see. I'm trying to absorb as much info as I can as I tinker with this thing.

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Tue Mar 02, 2004 7:08 pm

I undervolted the 3ghz cpu down to 1.400v tonight. It ran dual simultanious processes of cpuburn for 2 hours without ramping the fan or causing windows to crash. I just installed prime95 and will let it run continuously through tomorrow night.

It occured to me that a stable undervolt would likely be key to running a low rpm 80mm fan safely. The stock fan runs at 1800 rpms at 5 volts. Since I know it doesn't need to spin any faster, even with dual instances of cpuburn running, if I install a fan that runs at 2000 rpms at 12 volts I can take advantage of the motherboard's automatic fan stepping and end up with a much slower spinning fan most, if not all, of the time, and be covered if the cpu load should get demanding.

I'll go down one step in undervolting everytime prime95's torture test runs successfully and report the lowest stable undervolt I find for the 3 ghz @ 3ghz. It will take a while considering how long the testing process takes. I have time though. I'm waiting for new fans. When I get the new fans I'll run dual instances of cpuburn and see if the fans ramp up to 12 volts or not. If they don't, if they only ramp to say 6v or 7.5 volts then we'll know that there is some headroom left to accomidate an even higher ambient than the 21-22C that I'm testing in. If they do ramp to 12 volts then we know that there isn't much headroom and a 92mm will probably be a better option.

I can stop thinking out loud now. :)

Ralf Hutter
SPCR Reviewer
Posts: 8636
Joined: Sat Nov 23, 2002 6:33 am
Location: Sunny SoCal

Post by Ralf Hutter » Wed Mar 03, 2004 6:44 am

Copper wrote:I undervolted the 3ghz cpu down to 1.400v tonight. It ran dual simultanious processes of cpuburn for 2 hours without ramping the fan or causing windows to crash. I just installed prime95 and will let it run continuously through tomorrow night.
I don't know if you noticed but in the Shuttle Zen review I was running my system at 1.325V during my testing. I tested for stability with Prime95 but I had no temp monitoring app available to me when I wrote the review. After the review was published I obtained a beta temp monitoring app that worked on the ATi9100 chipset of the ST62K. I don't remember the exact temps but the difference in 2xCPUBurn temps between the default 1.525V and the 1.325V undervolt was about 8°C.

I've also run my other 2.4C 100% stable at 1.25V and it ran 13°C cooler under load than when it was at default Vcore. This was on an Albatron 845PE board.

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Wed Mar 03, 2004 8:01 am

Hi Ralf,

I'm anxious to see how low the 3.0C P4 will run stable. The numbers you provide look as though running a single low speed 80mm fan to cool the shuttle is promising. If it works we have to give some credit to shuttle for an excellent platform, a 3.0ghz system cooled with a single low speed 80mm fan!!

When I find the lowest stable Vcore for the 3.0C I'll put in a new low speed fan and check the fan rpms with the bios at each level; 5v, 6v, 7.5v, and 12v. I'll run 2xcpuburn to see how many steps the fan gets ramped, if any, at the low Vcore. I guess what I want to acheive is guiet, but also a system I'm not afraid to walk away from for a day for fear that it might overheat in my absence. I already know that even at 1.400 volts an 80mm fan spinning at 1800 rpms is enough to keep the processor cool under the most taxing load. As long as the slower fan spins at least this fast at 12v, the highest ramp of the onboard fan controller, I know my system wont overheat in my absence. If all goes well, this thing should be extremely quite if I can get it to cool with a single low speed "silent" 80mm at 5v without ramping up to 6 or 7.5v unless it's under a very heavy load.

Before leaving for work this morning I checked the computer to see if prime95 made it through the night ok, it did. Dtemp reports the drive is at 37C. I expect this is because prime95 taxes the ram and chipset as well as the cpu. Not only is the part of the ram directly under drive, the northbrige chipset heatsink is completely under the drive and only about a half inch from it. 37C doesn't seem bad considering the heaters that exist immediately below it. :)

Ps., I read another post where you indicated that fan grommets could be purchased at mcmaster-carr (it may have been an old post cause I've been using search). I ordered them monday and received them tuesday! UPS ground no less! Thanks for the unintended recommendation. :)

Ralf Hutter
SPCR Reviewer
Posts: 8636
Joined: Sat Nov 23, 2002 6:33 am
Location: Sunny SoCal

Post by Ralf Hutter » Wed Mar 03, 2004 8:29 am

Copper wrote:Hi Ralf,

I'm anxious to see how low the 3.0C P4 will run stable. The numbers you provide look as though running a single low speed 80mm fan to cool the shuttle is promising. If it works we have to give some credit to shuttle for an excellent platform, a 3.0ghz system cooled with a single low speed 80mm fan!!
I already did give Shuttle credit (although I was using *only* a 2.4C, way underpowered I realize :) )in the review. They did a great job with this SFF PC, especially when you compare it to what came before it.

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Thu Mar 04, 2004 5:19 am

Edited to correct Vcore numbers, 3/4/04

I guess I should have said that *I* have to give some credit to shuttle. :)

I dropped the Vcore down to 1.3375 last night. This was one step up from the 1.3250 where it crashed with cpuburn running. It, 1.3375, proved unstable twice with prime95, both in under 2 hours. I'm trying the next step up now, 1.3500. It looks like the 3.0C isn't going to be as forgiving as the 2.4C. 1.400 works. We'll see if it will go any lower than that.

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Thu Mar 04, 2004 9:29 am

Just came home for lunch. 1.3500 proved unstable at 2 hours and 13 minutes. Vcore is now set to 1.3625 and testing.

Working your way up is much more effiecient than working your way down when testing Vcore. Working down you have to wait 24 hours during tests until you find one that fails. Working up provides failures quickly allowing you to move up through the Vcore setting much more quickly, 3 or 4 a day if you have time to get to the computer every 3 - 4 hours. If you're starting fresh without any reference where your cpu might fail at I'd drop a quarter to a half volt from its default value and let prime95 run for three hours. If it fails at that voltage you can work your way up reasonable quickly until you find a setting that prime95 runs 24 hours without failing.

The prime95 literature indicates that the torture test will, if it's going to fail, usually do so in under 2 hours. Hence the value of working up rather than down.

Edit: I don't know where I got 2 hours from... I think I've been reading to much. From the prime95 stress.txt file:
I recommend running it for somewhere between 6 and 24 hours. The program has been known to fail only after several hours and in some cases several weeks of operation. In most cases though, it will fail within a few minutes on a flaky machine.
Mine has not failed in less than an hour and a half at the settings I've tried. It must not have been too flaky even at those settings. :)
Last edited by Copper on Thu Mar 04, 2004 6:04 pm, edited 1 time in total.

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Thu Mar 04, 2004 2:33 pm

1.3625 proved unstable at 4 hours and 4 minutes. I'm noticing a trend here. Did I hit the magic number with 1.4000? If so, I should have played the lottery instead. :)

On to 1.3750.

Edit: 1.3750 failed at 7 hours and 13 minutes. On to 1.3875.

Edit: 1.3875 failed at 6 hours and 18 minutes. I'm going to run 1.4000 a second time to insure it's stability.

Edit: 1.4000 is the magic number for this 3.0ghz @ 3.0ghz. It ran stable for ~24 hours during the first test and ~16 hours during this test. At 1.4000 the stock fan does not ramp even with dual instances of cpuburn running. Time to try a low speed replacement fan.
Last edited by Copper on Sat Mar 06, 2004 10:25 am, edited 1 time in total.

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Sat Mar 06, 2004 10:15 am

Big snag!!

I installed a new fan, a low speed inexpensive Aerocool 80mm:

Image

A little work with the voltage tester reveals that the motherboard is delivering ~8.25 volts to this fan even when the fan isn't being ramped. I left the case cover off, went into the PC Health Status of the bios and set the control to smart fan. With the cpu at 45C and the smart fan set to 60C the fan is being delivered ~8.25 volts, up from the 5 volts that the stock fan was delivered. Both fans are running at the same ~1800 RPM at their respective voltages. Leaving the voltage tester hooked up I lowered the ramp temp until I reached 44C where the fan began to ramp, then lowered the ramp temp in 1C steps and observed the voltage at each step. The fan was delivered ~.5 more volts at each step.

I only have one sample other than the stock fan to work with but it appears at this point that the on-board smart fan controller initially sets the fan not by voltage, but by rpm (~1800). From there, at every degree above the ramp temp setting, the fan is delivered ~.5 more volts until 12 volts is reached.

But there at first appeared to be an anomoly. Using the manual fan settings (ultra low, low, mid, & full) the respective voltages and rpms between the two fans at first didn't seem follow any rhyme or reason. Stock is delivered 5v, 6v, 7.5v, and 12v @ ~1800, ~2200, ~2800, & 3800, respectively. The Aero is delivered 8v, 10v, 11v, & 12V @ ~1800, ~2000, ~2200, ~2300. They didn't seem to follow any rhyme or reason until I looked at the percentage of increase for the second and third steps (low and mid). If the first step (ultra low) ramps the voltage until ~1800 rpm is reached it would account for the differing voltage but matched rpm of the respective fans at that setting. The voltage increases from ultra low to low and low to mid are ~1.2 times the previous, with the exception of the 10V to 11V (low to mid - aero) jump which is only ~1.1. The rpm increases are the same, varying from ~1.1 to 1.2. The last setting (full) just delivers the full 12v. It would seem that the second and third steps are a ratio of the preceeding steps with the initial step (ultra low) being whatever voltage is necessary to achieve ~1800 rpm.

So what does this mean? It means bad news. If the bios is looking for rpms then it's going to ramp the voltage being fed to the fan until it sees 1800 rpms. If we install a fan that runs at 1500 rpm at 12v (the nexus I have ordered and expect delivered monday, for instance) the fan is going to get 12v. The problem here is that while the fan may be running slower, and therefore quieter, the fan controller will already be maxed at 12 volts. If for some reason the cpu temps rise the controller has no more juice to give and the cpu fries. Alternatively, placing a voltage reducer between the fan and motherboard (like the fanmate I ordered and expect delivered wednesday) will have the same affect as running a fan that maxes out below 1800 rpms. If the board doesn't see 1800 rpms it's going to send 12v. The fanmate may cut those volts back to 5 and reduce noise, but as the board will be delivering all it can (12v) there is no protection from an overheat.

I will confirm my speculations when both the Nexus and Fanmate arrive. If true, then quieting the Zen will be a matter of going the opposite direction. Reducing fan size rather than increasing it. The test will be to see if a solitary 60mm, or even 40mm fan running at 1800 rpms will cool the beast. The question is, is this a deadend? Is it unreasonable to expect a 60mm or 40mm fan to run any quieter than an 80mm fan at the same 1800 rpm?

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Sat Mar 06, 2004 10:46 am

I don't think it is either voltage or rpm that the motherbaord is using. It's load! The stock fan gets 5v and the aero gets ~8v because they each spin about the same speed at those respective voltages and create the simular loads.

To test this idea I hooked up the voltage tester again and put my finger in the fan, slowing it. Voltage dropped rather than increased. The same thing was observed with the stock fan. Slowing the fan stesses the motor, increasing load. The board sees the increased load and drops the voltage attempting to reach its preset value. Bingo!!!!!!

Increase the load, slow the fan. The controller seeks a higher load at ramp temp, thus a higher rpm will be acheived at ramp temp. The smart fan control will still operate to protect the cpu and we get a slower fan!!!!!

Will a resistor achieve this??? Radioshack here I come!

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Sat Mar 06, 2004 3:50 pm

I picked up a 3 watt 25 ohm rheostat along with some wire connectors at radioshack and we have success. New approximate rpms at ultra low, low, mid, & full are 1500, 1600, 1700, and 1800, respectively.

Image

Image

Time for some cpuburn testing!

mas92264
Patron of SPCR
Posts: 659
Joined: Fri Sep 26, 2003 5:26 pm
Location: Palm Springs, CA, USA

Post by mas92264 » Sat Mar 06, 2004 5:21 pm

The question is, is this a deadend?
Yes. I've got a box full of 60 mm fans. They're useless as mammary glands on a boar hog. There isn't enough airflow through them at anything less than "loud" rpm. Even an 80 mm Panaflo L fan at max rpm is marginal in my 2 "full size" XPC's running folding@home. One of my many fantasies was to attach a 92 mm fan on the outside, similar to what you did. No shroud, though. One small problem is dealing with the smashed in bumps that are in the way on the back of the case. I was thinking about cutting away the offending part of the fan frame with a metal cutting blade on a table saw. Thinking about it was as far as I got. Was going to use an AOC/Evercool aluminum frame 92 mm fan. I did see a pic of an XPC with a 120 mm fan on the outside back using a 92x120 adapter, however, you lose some or all of the pci slot.

FYI, SpeedFan (which I understand will not work with the ATI chipset) doesn't control fan speed via the rpm wire.

Interesting idea and nice work on the variable pot.

M

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Sat Mar 06, 2004 6:45 pm

Thanks mas92264.

Running two instances of cpuburn netted a fan ramp in about 40 min. The ramps are so small now that I had to hook up the voltage tester and run the leads out the pci slot cover so I could tell if it was ramping. Though the fan is not inauble at this level, the ramps practically are.

Letting the cpuburns continue to run, the fan seemed to stablize at the 10.1v level, occasionally bouncing down to 9.75 volts (ultra low with the rheostat attached and turned to 25 ohms) and occasionally going up one more ramp to ~10.5 volts. Ramping steps with the rheostat seem to be reduced to ~.3 volts where without the rheostat they were ~.5 volts. There should be at least three ramping steps left from the peak level acheived with cpuburn. That's good news. If even more headroom were needed one could turn the rheostat back a little too.

I'm excited to get the Nexus fan.
Last edited by Copper on Sat Mar 06, 2004 6:59 pm, edited 1 time in total.

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Sat Mar 06, 2004 6:57 pm

And just to add that to my ear it seems to be all motor noise coming from the fan now. I can no longer hear the sound of moving air. And this motor is much quiter than the stock one.

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Sat Mar 06, 2004 8:38 pm

Ralf,

I tested my power draw and got some numbers that seem pretty unbelievable. I plugged my power supply into a six foot extension cord that has the insulated leads separated so an amp meter can be used on it. I tested the wall outlet with a voltage tester and it gave me ~121 volts.

I got the following amp readings with Vcore at 1.4000:

Desktop Idle: ~.31 amps (it's a digital readout but hops up and down)
2xcpuburn: ~.70 amps (1xcpuburn gives the same reading)

121v x .31a = ~38 watts
121v x .70a = ~85 watts

These are the readings I got with the Vcore set to auto:

Desktop Idle: ~.35 amps
2xcpuburn: ~.82 amps

121v x .35a = ~43 watts
121v x .82a = ~100 watts


How can this be when you got the following with a 2.4C, first undervolted then not:

Idle: 42W 45W
Load: 86W 103W

And my 3.0C:

Idle: 38W 43W
Load: 85W 100W

Lower at every check? Can't be. What Am I doing differently?

It's just wierd. I expected to be giving everyone an idea how much more the 3.0C draws in the shuttle.

Ralf Hutter
SPCR Reviewer
Posts: 8636
Joined: Sat Nov 23, 2002 6:33 am
Location: Sunny SoCal

Post by Ralf Hutter » Sun Mar 07, 2004 6:29 am

Copper wrote:How can this be when you got the following with a 2.4C, first undervolted then not:

Idle: 42W 45W
Load: 86W 103W

And my 3.0C:

Idle: 38W 43W
Load: 85W 100W

Lower at every check? Can't be. What Am I doing differently?

It's just wierd. I expected to be giving everyone an idea how much more the 3.0C draws in the shuttle.
I dunno. I measured mine with a Kill-a-Watt, you did yours with a straight amp meter. Maybe that can account for the difference. Otherwise I have no clue.

And the 3.0C does draw more current than the 2.4C, at least in the testing I did with my 3.0C, 2.4C and Kill-a-Watt meter. I don't have the numbers anymore but, IIRC, the 3.0C used about 20-25W more than the 2.4C, at idle and at load.

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Sun Mar 07, 2004 7:16 am

Everyone should probably take my wattage figures with some reservation.

I did install a bios update trying to solve a usbfloppy issue. I don't know that it could account for the differences though.

Bios update:

"Checksum: C800 Date:01/28/04 1. Update CPU micro code"

http://www.shuttle.com/share/fae/hq/dow ... meft62.htm

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Tue Mar 09, 2004 1:32 pm

The Nexus 80mm and Fanmate arrived. The Fanmate works with this motherboard as well. So there is no need to wire in a rheostat by hand. The nexus runs about 1500 rpm at 12 volt. Using the nexus and fanmate together you could tweak the rpms down as low as possible while still keeping the cpu cool. How low, you'll have to experiment with. I've already moved on to a fanless project with the same computer. :)

Fanless using a major undervolt and underclock. Now my hard drive is driving me nuts! New hard drive and enclosure is on the way. I'll post stats and pics in a new topic when I'm done.

Knowing how obsesive I am, I should have just jumped straight to the fanless project from the get go. :)

spacey
Posts: 81
Joined: Mon Aug 11, 2003 10:31 am
Location: Ontario, Canada

Post by spacey » Tue Apr 06, 2004 7:37 am

i'm thinking of buying a zen for my gf, what cpu do you recommend. she's only going to be using it for school and we want to keep costs down. i was thinking some sort of celeron, they take less voltage than the p4s no? and maybe undervolt it?

also, what fan ended up working best for you to replace the stock with? did that nexus + fanmate work well?

PhilgB
Posts: 389
Joined: Sun Jul 13, 2003 4:25 am
Location: Welland, Ontario, Canada

Post by PhilgB » Tue Apr 06, 2004 9:27 am

I hope they make a SocketA version of the Zen. That passive PSU is brilliant. Would be nice with an XP-M 2500.

Ralf Hutter
SPCR Reviewer
Posts: 8636
Joined: Sat Nov 23, 2002 6:33 am
Location: Sunny SoCal

Post by Ralf Hutter » Tue Apr 06, 2004 1:13 pm

spacey wrote:i'm thinking of buying a zen for my gf, what cpu do you recommend. she's only going to be using it for school and we want to keep costs down. i was thinking some sort of celeron, they take less voltage than the p4s no? and maybe undervolt it?
NO! P4-based Celerons are dogs. Just get an inexpensive P4, something like a 2.4C, although it's out of production. They run very well undervolted

Celerons run at the same Vcore as real P4's, but who knows, maybe they'll run OK if they're undervolted. I don't recall anybody trying that.

Copper
*Lifetime Patron*
Posts: 587
Joined: Sun Feb 22, 2004 8:43 am

Post by Copper » Mon Apr 12, 2004 12:04 pm

spacey wrote:i'm thinking of buying a zen for my gf, what cpu do you recommend. she's only going to be using it for school and we want to keep costs down. i was thinking some sort of celeron, they take less voltage than the p4s no? and maybe undervolt it?

also, what fan ended up working best for you to replace the stock with? did that nexus + fanmate work well?
Like Ralf I don't have much good to say about the Celeron. The only exception being where the computer is being used strictly for office and internet software. In these two areas it doesn't make sense to pay the extra cost for the P4. The only experience I have with undervolting is with my P4 so I don't know how well a Celeron will undervolt. You can pick up a Celeron for 60-$70, so experimenting with one would be pretty inexpensive.

The Nexus and Fanmate worked very well. Though I didn't get to experiment very long with them because I moved to a fanless project with the same computer shortly after receiving them. Running full speed the Nexus is more than capable of cooling a 3.0c P4 undervolted to 1.4000 Vcore, even with cpuburn running. It will very likely cool the same setup at less than full speed using the Fanmate to reduce its speed, but I decided to go fanless right when I got the Nexus in. The Nexus and Fanmate are big improvements over stock. Not to mention that together they aren't very expensive.

Post Reply