Prev: Problems with badly burnt DVDs
Next: Salts and farming
From: Moe Trin on 15 Oct 2007 20:29
On Sun, 14 Oct 2007, in the Usenet newsgroup alt.os.linux.mandrake, in article
<AZxQi.2320$Oy1.1476(a)trndny08>, Adam wrote:

>Moe Trin wrote:

>> Again, we have no idea where the temperature is being measured. Is
>> it possible to touch the "top" of the drive (or at least get the
>> fingers close enough to detect if the drive is really that toasty)?
>
>hda apparently gets cooler under stress, which is impossible. I just
>stressed hda again, and again it reported 57C before it locked up.
>However, a household thermometer placed on top of hdb only went up to
>28C, which is about how warm it felt.

Assumed typo. I would sort of expect the thermistor to be inside the
drive near the platters, as that is where most of the heat is going
to be generated (spin motor and head positioner). You say it went up
to 28C (82F) - from what quiescent?

>I'm considering both HD temps, as reported by SMART (and therefore
>hddtemp), to be inaccurate.

I'd have to agree.

>Hmm. Let me think about that. Obviously the temperature measurement
>is incorrect, but SOMEthing locks up when the drive is stressed up to
>a certain point.

"locks up" meaning...

Old guy
From: David W. Hodgins on 16 Oct 2007 06:59
On Sun, 14 Oct 2007 19:53:04 -0400, Adam <look(a)bottom.for.address> wrote:

>> Also, I recently cleaned the fins on my cpu heatsink, and reset it on
>> the cpu with some more thermal compound (after cleaning off the old stuff),
>
> I did that last weekend, because my search for the temperature sensors
> involved taking out the power supply and then the CPU heatsink/fan. I
> gave it about a 1 mm coat of the expensive silver thermal paste. I
> think CPU temp as measured has been slightly lower since then, though
> I'm pretty sure the value is inaccurate and only good for showing
> relative changes.

Again, just a reminder. Make sure the fan/heatsink combo is tight
against the cpu. Otherwise, the heatsink, actually ends up just blocking
the air flow from the fan.

My biggest problem is the cat fur. At least they've learned to
avoid getting too close to the external fan :)

Regards, Dave Hodgins

--
Change nomail.afraid.org to ody.ca to reply by email.
(nomail.afraid.org has been set up specifically for
use in usenet. Feel free to use it yourself.)
From: David W. Hodgins on 16 Oct 2007 07:08
On Sun, 14 Oct 2007 19:53:04 -0400, Adam <look(a)bottom.for.address> wrote:

> Well, both drive temperatures seem to be inaccurate. hda seems to get
> cooler under stress, which is almost impossible, and hdb claims to get

If the sensor is near the hottest part, and spinning up the drive
temporarily causes better heat distribution, it would be possible,
although I've never seen that myself.

> Hmm. Let me think about that. Obviously the temperature measurement is
> incorrect, but SOMEthing locks up when the drive is stressed up to a
> certain point.

Given that the drives are locking up, I suspect those temps are close
to accurate.

As Moe said, could be a grounding problem, or it could be just bad
air flow, around the drives.

Check hdparm -I /dev/hda, etc. Read the man page carefully. Ensure
dma mode, 32 bit access, and readahead are on. Also check the acoustics
settings. I strongly suggest turning off write caching, usless you have
a tested ups system. The risk of using it, isn't worth the benefits, for
most home systems.

Regards, Dave Hodgins

--
Change nomail.afraid.org to ody.ca to reply by email.
(nomail.afraid.org has been set up specifically for
use in usenet. Feel free to use it yourself.)
From: Adam on 17 Oct 2007 22:43
As usual, I'm replying to everything in this thread in a single message,
so I don't have to repeat myself, and to make it a little easier for
those who are following, or ignoring, this topic. Is this a good idea,
or a bad idea, or doesn't matter much either way?

Wes Newell wrote:
> [lm_sensors] doesn't care about MB's. It strictly looks at the outputs of
> the sensor chips on the board. Sensors.conf tells it what is what on the
> sensor chip which can vary from one MB to another. It's up to you to
> straighten it out if the default doesn't match the way your MB handles the
> sensors.

Yep, I've been trying to figure out just what my MB has connected to the
IT8705F's inputs. I think I have it mostly figured out, except for in7
which reports about 1.65-1.74V and temp3, which if calculated as a
thermistor reads from 14C to 69C and may or may not be actually
measuring anything, and if it is it may or may not be measuring anything
useful.

>> gave it about a 1 mm coat of the expensive silver thermal paste.
>
> LOL. I use 30+ year wheel bearing grease.

NOW you tell me. :-)

Moe Trin wrote:
>> hda apparently gets cooler under stress, which is impossible. I just
>> stressed hda again, and again it reported 57C before it locked up.
>> However, a household thermometer placed on top of hdb only went up to
>> 28C, which is about how warm it felt.
>
> Assumed typo. I would sort of expect the thermistor to be inside the
> drive near the platters, as that is where most of the heat is going
> to be generated (spin motor and head positioner). You say it went up
> to 28C (82F) - from what quiescent?

Yep, meant "just stressed hdb again." Today, same household thermometer
placed on top of hdb (bulb 1/4" above top surface of HD) stayed around
33C no matter how much (or how little) the drive was accessed, even
while hddtemp was reporting up to 60C. It felt more like 33C than 60C.

>> I'm considering both HD temps, as reported by SMART (and therefore
>> hddtemp), to be inaccurate.
>
> I'd have to agree.

One of these days I'll look into exactly how it's measured and whether
recalibration is possible and/or worth the effort.

>> SOMEthing locks up when the drive is stressed up to
>> a certain point.
>
> "locks up" meaning...

Meaning hddtemp (reads SMART data from drive) and zcav (stress drive by
reading blocks) stop producing output, and stay that way until the drive
apparently cools down. Further investigation revealed that it was NOT a
matter of drive stress, but an attempt to read something between blocks
5000 and 5100 of hdb, or between blocks 2900 and 3000 of hdb2, that
causes this to happen. No problem when it reads anything else on hda or
hdb, even all 7100 blocks of hdb6. So... the problem has been narrowed
down to something between block 2900 and block 3000 of hdb2, which is
W2K which seems to run well enough. Another problem for investigation
later on.

David W. Hodgins wrote:
> Again, just a reminder. Make sure the fan/heatsink combo is tight
> against the cpu. Otherwise, the heatsink, actually ends up just blocking
> the air flow from the fan.

Thanks, I remembered. I think CPU has been running a few degrees cooler
since I replaced the thermal paste.

> My biggest problem is the cat fur. At least they've learned to
> avoid getting too close to the external fan :)

I wish I could have a cat. Back when I lived with my parents, Allegra
(large tan tabby) once stood on <ctrl> and <c> for so long that the
succession of ^Cs caused Compu$erve to disconnect me!

>> Obviously the temperature measurement is
>> incorrect, but SOMEthing locks up when the drive is stressed up to a
>> certain point.
>
> Given that the drives are locking up, I suspect those temps are close
> to accurate.

As explained above, lockup only occurs when one particular part of hdb
is read. And hdb's internal sensor reports up to 60C but the
thermometer on top stays at 33C, which is about how warm it feels.

> Check hdparm -I /dev/hda, etc. Read the man page carefully. Ensure
> dma mode, 32 bit access, and readahead are on. Also check the acoustics
> settings. I strongly suggest turning off write caching, usless you have
> a tested ups system. The risk of using it, isn't worth the benefits, for
> most home systems.

Thanks, will do!

Adam
From: Wes Newell on 18 Oct 2007 01:51
On Thu, 18 Oct 2007 02:43:59 +0000, Adam wrote:

> Yep, I've been trying to figure out just what my MB has connected to the
> IT8705F's inputs. I think I have it mostly figured out, except for in7
> which reports about 1.65-1.74V and temp3, which if calculated as a
> thermistor reads from 14C to 69C and may or may not be actually
> measuring anything, and if it is it may or may not be measuring anything
> useful.
>
in7 could be vdim, but I doubt it. I ignore everything from in5 on. I
guess you might want those if you have a laptop. Temp3 isn't used more
than likely. At least it isn't on any boards I have.

--
Want the ultimate in free OTA SD/HDTV Recorder? http://mythtv.org
http://mysettopbox.tv/knoppmyth.html Usenet alt.video.ptv.mythtv
My server http://wesnewell.no-ip.com/cpu.php
HD Tivo S3 compared http://wesnewell.no-ip.com/mythtivo.htm

First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4 5
Prev: Problems with badly burnt DVDs
Next: Salts and farming