Cocoon Culture No.47



Measurements of HDD noise hot-spots etc. :

First prototype of HD noisebox.Came to think about it. This document is mainly related to the stuff that resulted in the construction of the HD-Noise Box. Rename and reorganize this later.

[ Index

Testing of equipment :

System state : CD/Floppy dissabled. HD Spinning. CPU 233@66 & Vcore 2.5

Mesurement on two targeted point on the PSU,  refrence point A. tube directly on fan grid (no air turbulence). reference point B. tube directly in air-stream.

on ref. point A. apprix. -57dB (cool-edit scale) 
> m-001.wav | m-001.mp3 (10 sec)

on ref. point B. appres. -39dB (cool-edit scale) in air-flow of PSU fan)
> m-002.wav | m-002.mp3 (10 sec)
OO - recording makes nice reproduction of the fan noise characteristics.

- Recording noise-level with meter (windcap on) immediately behind the PSU fan grid. 51,7 dBA(max) and 61,4 dBC(max).

- Recording noise-level with meter (windcap on) from the standard listening position. 29.0 dBA & 35.7 dBC (approximately).

Not sensitive enough through the soundcard mic.amplifier.
> Try using external mic. amplifier.

Using an external mixer pre-amplification to maximum,.I managed to get the levels up to - 11 dB on the Cool-edit meter.

Test recording at ref point B.
> m-003.wav | m-003.mp3 (7 sec)

Noise introduced due to the heavy amplification :
Peak level is -48 dB with mic. turned off.
(10 meter cable, well it's supposed to be a good one).

Turning the microphone on again, didn't change the noise level. So guess that's good. But hasn't become as focused as I had hoped for. Hiting the enter key on the keyboard 1-meter away-from it makes the level meter plonk up to -30 dB.

Starting probing

Now, to see if it can be of any use in locating sound.

Removed the noise-box around the drive. and took it outside. Observed that the drive was somewhat hoter than normal. apprx. 45 degrees maybe. Warm but no problem with holding the palm of my hand on topp of it without hurting. Moved the probe around the top surface of the disk. probing. trying to find something. And I did.

Lets first make a refrence recording from a more silent part of the disk.

m-004.wav | m-004.mp3 ( not much )

m-005.wav | m-005.mp3 (one spot found) -

m-006.wav | m-006.mp3 (measurement around breather hole, not outstanding in anyway).


> Now flipping the thing upside down. Test the other side (bottom) of the drive.

(Please note that the mid- and treble range from here on have been boosted to focus on the high-frequency components. Apprx.16 dB)

m-007.wav | m-007.mp3 (an average recording of the flip.side noise level)

m-008.wav | m-008.mp3 (probing in the hollow space under the print-board)

m-009.wav | m-009.mp3 (a typical surface level spot)

m-010.wav | m-010.mp3 (a bad spot)

Learnings from the probing :

I didn't find what I wanted to find. There were no particular areas that pointed it self as worse than others. Althoug some areas are worse than others. There were found no bad spots. But bad areas. I don't have any damping material that helped on these area's. Well except that laying my hand on it worked.  So If I can come up with something that emulates the noise damping characteristics of a hand, I've reduced the magnitude of the problem quite a bit.

It appears that all of the metal resonates at this annoying high pitch frequency, but it is not located to any particular area, but distributed evenly throughout. Much like a triangle resonating. It's a sharp sound and lasts a long time. The way of silencing a triangle is also by touching it with your hand, so here the two objects have something in common.

As from what I can see know, the solution must be to add mass. But some kind of watery, jelly-like stuff. Something that can absorb the high frequcy vibrations, and take them down, cancell them out.

The noise probe was working OK. But I think maybe a doctors-scope would have  been suitable. (steto-scope ?) Unfortunately I don't have one.  Swapping the plastic-tube for a solid metal needle like thing directly attched to the mic. whould probbably do the same job. Modificatons to the tool.

marbles-placement-drive.jpg (10900 bytes)Marbels a marvelous idea. (Altering the character of the high-pithced noise)

Notice ... the difference in sound characteristics some 6 sec. into the recording.m-011.wav | m-011.mp3

> Followed up on this. Seven marbles were-placed on top with some sticky stuff. rubber-glue-like.By tuning the placement of theses marbles, the high-frequency noise was improved noticably.

Wrapped up the noise-box. Did some measurements.

Before (with marbles) Average 1 cm above drive top : +55 dBA +60 dBC (+ 50 dBA over the marbels) : Disk idle

After (boxed) Average 1 cm above box top : +33 dBA +39 dBC : Disk Idle state

f1000019-crop-marbles-c15-.jpg (8834 bytes)Picture of the situation with the marbles attached to the top lid of the drive.>>>





Summary / Status on system after probing and HDD modification :

OO - After completing the HD- noise box, and fine tuning with marbles,
the situation now is :

  1. The dominant noise is from the monitor. (eeeeehh) anoying.
  2. PSU FAN : ( With the cardbordbox noise trap on )
  3. The High pitched PSU noise - almost unnoticable, but I can feel it's there. Like psyko-accoustics.
  4. The CPU cooler-fan. Hard to distinguish from the PSU fan. Low-brumble. Muted by chassis. But still there somewhere.


OO - HD Temperature inside the noise-box. Stable but level unknown.

oo - mic. tool worked pretty OK. satisfied with the results.

oo - the new mic.tool now enables me to make more representative recordings of the noise. use in PSU noise rec.

oo - Use of mixer tone controlls helped to focus on sounds of intresst.

Drive Temperature. Follow up of the HD- noise box. Measurements to verify safe operation.

- As mentioned elsewhere I was a little bit woried about the temperature in the box. Might it get to Hot or what ?

So I got hold of a shareware utility SigGUARD, that acesses the drives SMART attributes and gives some kind of monitoring of the intenal state of the drive.

After some plundering with setting up the program, I got some readings.There where two unspecified attributes, which on IBM drives may be temperature related. Great. The attributed changed, expected time of failure tomorrow. I looked at the clock, great it's 30 minutes to midnight. After some messing about, the attribute changed and had it's life   expectancy prolonged by 2 days. Reading the documentation for the  program I was told not to panic, as it would take some time in the  beginning for the program stats to tune in.

Typical Readings during normal use.


Decimal Hex
137 [raw 28 00 0000 0000 ]
141 [raw 27 00 0000 0000 ]
144 [raw 26 00 0000 0000 ] - Drive powered down.
x x
157 [raw 23 00 0000 0000 ] - After 20 minutes HD-power off. Best case recorded.
x- x
Doing some monitoring while I surf.
144 (after 5.min)
x- x
157 (@ 00:34)
144 (@ 00:39)
141 (  00:49)
144 (  00:59)
141 (  01:04)
SYSTEM CRASH AT 1:06 - Systesmet har svært lite resurser.
pluss noe irrecoverable error in VDX VMM sgfasgasg
137 (  01:14)
134 ( 01:34)
134 ( 02:03)
130 ( 02:33)
134 ( 03:07)
130 ( 03:17) [ 2A 00 0000 0000 ]
134 ( 03:47)
137 ( 03:57)
134 ( 04:02)
141 ( 04:17)
144 ( 04:27)
141 04:47)
141 ( 04:52) New expected time of failure in 2 weeks.
134 ( 05:47)

[ End of testing. System in normal use, with 1 minute HD power-down setting ]

Ideas : To make some sense out of these numbers, get reading when drive is super-cool (i.e. without noise box, outside cabinet) [ Or better remember to record value at boot time; with cold cabinet ]

> Follow-up : Initial reading 253 [Raw $ 15 00 0000 0000], room temperature approx. 16-18 degrees (C)

Time Value Hex Comment
19:15 239 - normal system load.
19:18 229 - internett use.
19:23 220 [$19 ..]
19:28 211 [$1A ..]
19:33 196
19:43 189
19:48 183
19:54 177
20:13 161
20:23 157
20:49 144

Typical start up value [250] $16 00 0000 0000

and value when it is heavily loaded (15 min burn in),
> Follow-up : see measurements above : lowest value recorded 127 and standard idle-value.
> As long as the values reaches a stable value, it should be OK.

With worst value beeing 100. I don't quite understand this but, from what I can understand now. The higher the better. The lower the worse. As drive power-down value-reading increases. in steps of 1 [hex-decimal ]

Recording values with heavy disk load.

Using Scandisk, thorough surface-testing option.

Time Value Comment
0:35 [134]
0:41 [130]
0:51 [130] Program appears to be more CPU than disk intensive.


Running defrag, which is more seek intensive.


Time Value Comment
08:13 [127 ] $2B 00 0000 0000
08:23 [127 ] Still not changed.
08:28 [127 ] Assume that this is the noramal maximum level; and that anything beoynd this is critical.
08:30 xxx  Defrag cancelled. See reding on next 5 min. poll.
  08:33 [127 ] Poll. stays the same.
08:38 [130 ] First drop in temperature reading.
08:39 HD-powerdown
08:43 [134 ] Poll.
08:44 HD-powerdown
08:53 [137 ] Poll.
08:54 HD-powerdown
08:58 [141 ] Poll.
08:59 HD-powerdown
09:18 [144 ] Poll.
09:19 HD-powerdown
09:28 [144 ] Poll.
09:29 HD-powerdown
09:33 [141 ] Poll. ??? An increase !!! Assume border; stable.
09:24 HD-powerdown

Drive idle temperature recorded with auto-powerdown feature disabled and no/low disk activity.

Summary of disk temperature values :

  • Average idle temperature : 144
  • Maximum work temperature : 127
  • Minimum boot temperature : 253

Update : Drive temperature is well within operating range of 0-50 C.

Reading at value : 137, translates to 34.9 C using a termometer probe inside the box. Cabinett exhaust air-temperature 30.4 C. No reason for worry.

Searching for a tool / information about reading values

Would be nice to know what these codes means, recomended/critical vales etc.
Do some searching ... Maybe find some IBM Deskstar spesf. utillity.

Tom's hardware page, good info on HD temperature and noise levels.
I'm put a little at ease seeing normal operating temperatures
beeing in the range 37 to 50 degrees Celsius.
Noise levels on newer drives : 27.3 to 29.4 dBA -
Measurement not specified, so assume this to be typical levels.

IBM download section for the drive DJNA-351520 - monitoring utility.

From looking at the drive specs for my drive - operating temperature was from 0 to 55 degreec (C) Think it's withing this range.

downloaded utility : EZ-SMART5_0_24_IBM.exe

Nice utillity, not as informative as SigGuard more like "trust us everythings ok",
and it had a check-box for drive-temperature, but guess what : It was grey !!!
Now that's strange, cause I have an unknown parameter that changers with drive temperature
from the SigGuard utillity. Hummm ... creator of this tool located at ...
lets go have a look ...

Humm this is also a downloadable utillity v5.0 that retails for $29, could it be that the IBM free version have the temperature option disabled ?

Cause there is a tick in the box, but the box is greyed out.

Soddit, back to the search engine.

Not easy to find something ...
IBM : Drive Fitness Test (version 2.00) - 1913 KB

Using this utillities Quick Test - The ready chewed diagnostics said 0x00 ; which meant that evereything was fine.

That's it .

Table - Relationship between HDD SMART values & Noise Box Temperature.


SMART Value (Hex) SMART Value (Decimal) Temperature (Celsius)
$15 253 19.5
$17 239 21.0
$19 220 22.2
$1A 211 23.2
$1B 203 24.3
$1C 196 25.3
$1D 189 26.2/27.6
$1E 183 27.1
$1F 177 27.8
$20 171 28.4
$21 166 29.0/29.7
$22 161 30.1/30.5
$23 157 31.2/31.1
$26 144 (Normal idle temp) 31.7/32.0
$27 141 33.1/33.5
$28 137 34.1/34.9
$29 134 35.1/35.4/35.8/35.7/36.1]
$2A 130 [36.2/36.4 / 37.2
$2B 127 (Peak temperature) [37.1 /37.3

Latency on Digital Termometer is 10 sec. HDD monitoring polled every 5 minutes. Different temperature readings for same SMART values. Latency between internal drive temperature change and external temperature measured. Not so important. What's important is to see that the operation temperature range doesn't exceed specs. All recording Temperatures recorded with an (ambient ?) room temperature of 22.0 C.

As of now it may look as a $01 change in SMART value, correlates to a temperature change of 1 C.

[ Index ]

Copyright © 1999-2002 Cocoon culture no.47 ®. All rights reserved.
[ ? ] [ Top ] [ Back ]
Hit Counter