PART 3

[ index ]

ABOUT THIS PART :

Contniuing of from where I left off some months ago. What's changed ?

Hmmm. I've finally got a new PSU that is supposed to be really really quiet. Make and model (Fujitsu-Siemens).

Modifications to cabinet :
Gluing. Internal Damping. Making adapter so that  the new PSU would fit into the cabinet.

HD relocation and damping (independently suspended box). See the noisebox project.

Some sound recordings from the different scenarios.

Note ! Systems settings used throughout this part are : CPU=233Mhz, bus= 66 Mhz and Vcore=3.2 Volts.

FAN DOODLING :

The new fan that was replaced some 6 months ago, have become noisy again. Attempts at solving the  problem.

• Cleaning the fan : Effect not noticable, except from that it looked better and reminded me that I should give up smoking.
• Oiling the fan : No effect.
• Vibration damping : Rubberband system on fan .... worked but not as intended, see conclusions futher down on this page.
  Inserted damping material underneath the base of the processor on the motherboard to improve stability. Helped. (I always wondered what those two rubber feets that came with the m.b. mounting kit where meant for. Probably for placement underneath the processor. I put them underneath the PCI slots. Well well, we live and we learn.)
• Other options : Drop the fan ... ? Testing.
  Need some tool to verify the correct operation of the processor. Go to some overcloing site to find tools. tick tick ... www.overclockers.com found the tool CPU Stability Test 6.0 ( www.vtoy.fi/jv16/ ) which worked great for testing.
  The computer ran for 20 minutes with maximum load without any problem. It also passed the programs 5 min. benchmark test without fan. Note : All this was done without the chassis cover on. Don't know how well it would have worked with the cover on. I'll try this in desperation some other time.
  
  OO Max Operating temperature on a  p233-mmx, is said to be 70 degrees Celcisus. This is the same temperature as the maximum allowable surface temperature on a wall-heater in Norway.
  (Just to get a feel for the temperature ).
  
  OO Computer also worked normally for 2 hour with open chassis, without CPU fan & without problems. Not a permanent solution as the HD made to much noise.

SOUND RECORDINGS :

Notes on the recording levels and amplificatoin :

• 20dB+ pre-amplification used on microphone.
• All recordings also software amplified by +44dB, except where otherwise mentioned.
• Microphone placement approximately  30 cm above cabinett front, slightly to the left of the cabinett side. Placed in the typical direction of the user.

Note (1) : Captains log : It's not easy to perform noise measurements in a house full of people moving about. Taps running, cars moving by, da.da.da.

Note (2) : Captains log continues : I think I know now, how the first recording engineers must have felt when somebody screwed up their recordings with the words  : " Are you recording ?". Anyway it doesn't matter that the recordings are of poor quality and relativity. They still give some kind of perspective of changes in the noise level and characteristics.

* GETTING DOWN TO THE POINT

Factors : CPU state HD state General system state  CPU FAN state SW running (CPUIdle) Cabinett state(open/closed)

CPU Idle running.  (GPOS).

All redundant power consuming devices and cards have been removed ( CD-ROM/modem/Diskdrive), except from the sound-card. This reduced the noise level on the PSU noiticably.

SITUATION #1

Comment : Disabled CPU  Fan to see how stable system was without it. Cabitnett cover was removed to improve the airflow, and reduce the load on PSU cooling fan which also needs to cool the rest of the cabinett.
Sound recording : take-1 (2 sec. wave loop) / full take mp3 at 128kbps (mono)

take-1    :    Scenario : case cover open, PSU fan disabled, HD spinning.

SITUATION #2

take-2    :    Scene : now with HD shut down. (might have to retake, not full 10 secs. but got the beginning clean.)

SITUATION #3

take-3    :    Scene : now with CPU FAN, open chassis, HD spinning.

2 second loop-recordings.

SITUATION #4

take-4    :    Scene : as in (3) but With HD spunn down. (Very noisy recording.)

Listen to it :

SITUATION #5

****** Now putting the chassis cover back on *******
take-5 :     Scene : System operates quietly. HD on , CPU FAN ON.
 !  Note on the recording, noises stemming from someone building stuff in the neiboring room.

System is so quite that i can hear the transformer noise from the monitor. Long time since.
Well actually, on opening the monitor and exsamining the source ... the noise was comming from the high-voltage area of the printboard mountede on the back of the picture tube. Not much I could do about it. Tried to dampen the protective metal grid, but no effect. This was not the source of the noise. I'm uncertain as to what the cause of problem is - and as how it can be eliminated. There was no noise comming from the monitor when I bought it and it was new. But it came creeping in gradually with time, that is after a couple of weeks. Well well. I don't want to risk my life exploring this problem any futher, the voltages here are to high so I'll guess I'll just have to get a new monitor without noise. Where ever they can be found !?

Listen : click on filename to hear the recordings.

(Note : level deviation SW amplified by +35dB)

SITUATION #6

take-6    :    Try with HD spun down. (Great ... monitor makes most of the annoying noise. )

Listen : take-6.wav (2 s)

Note monitor have been on during all recordings sofar. And with the HD  spun  down, it is perceived as  the dominant anoying sound.

SITUATION #7

take-7    :    Disabling the CPU-idler ... se what happens ...  [ Amplified (SW) by only +20dB to get recording relative to natural room sounds such as breathing etc. ]
System state : everything is on.
Result :  Noise level increased instantly, as PSU FAN geared up. Listen to it. HERE.:
Recordings split ut in time periods over a total of 28 minutes. (Need to cut this down)

Listen : take-7.wav (several extracts at diffent time points)

1 second snapshot. loop to get the impression. .wav & .mp3 versions.

SITUATION #8

Take-8    :    Now enable CPU-idler software again and observe ...
Result ... Instand drop in PSU FAN noise, slowly gearing down to silence in about 5 minutes.

Listen : ( 1 sec. snap-shots at different time locations. .wav   .mp3 versions)

REMAINING :

Do recording with exhaust pipe on : Several noise trap prototypes are beeing tested. So far with great success with respect to the HighPithced PSU noise, but not so well on PSU Fan noise.

Do recording when HD have been boxed, and relocaded. Results so far in standalone suspension (isolated from the rest of the cabinett chassis, have been great. What remains are the high pitched whining noise; which I hope the "box" will sort out. Seek and vibration noise have almost been eliminated.

OO - Sound recordings are not very informative. Noise levels are to low to be recorded properly. Need to boost the levels. Do the meter reading as well. (at the end).

Other modifications :

Adding of new style exhaust-pipe-muffler-stuff. (Not yet completed).

Noises still present : Monitor (findings see situation #5). PSU (High-pitched ones).

Relocation of HD : The encapuling of the drive to remove the HF noise from PSU (as well as Fan noise).

Cabinet modifications : Adding mass. Damping materials. Gluing. GREAT Success. Link to similar project.
The cabinett damping method have allready prooved sucessfully in a preceeding AT cabinett.

Conclusions :

- The CPU-Idle utility (v.5.6) (at www.cpuidle.de ) really saved the day. It is a must have ... and with the quick and dynamic response of the new PSU, you can hear the fan speed change instantly with CPU load.

- The noise level has now become livable.

- The noise level now relates to the power consumption of the system. Which is good.

- The CPU FAN noise level was reduced by adding rubberbands on both sides of the 4 screws that conected the fan to the cooling element.  The idea was to make somekind of FAN "suspension". This didn't work very well. What did work however was bending the ribs of the cooler, so that the screws fastened more tightly. (It is possible that the rubberbands made the fit between fan and cooling element  better, reducing vibrations).

- The notorious High-Pitched noise have haunted me all along. Even this "High Quality" PSU has this anoying quality. I've come to the conclusion that this must be part of the SWITHCING PSU's nature.
At least I never heard that type of sound in the older AT PSU's . The possibility that it is motherboard dependant still exist, but I don't think so. As mentioned else were it was possible to reduse it by altering the power management something in the control pannel.
The sound dissapears when pressing the mousebutton as before....
Hmmm. It appears that pressing the mouse button, brings the system out of it's idle state. And starts to consume more power.  Wonder wonder ... ???

But the noise characteristics isn't that anoying on this one at least. (Comfort, comfort ...)
Update on the issue [23.10.2000] : The HF noise from the PSU is dependant on the systems power consumption. Problem increase as system power consumption decreases. Pressing the mouse button removes the sound because system is then brough out of it's idle state, and power consumption increases.

- Subjectivly ... Adding the case covering muted  the sound characteristics noticably. Particulary the high frequencies.

- Noise characteristics : It's easier to live with the low frequncy and soft sounds, rather than the hard and high pitched ones.

- A dynamic noise level is also more digestable than a static one. It brings some kind of life to the PC. You can hear that something is going on. When the fan boosts, you know that it's working harder. ( A little positive thinking help a lot too).

[ index ]