New Case

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

On Sat, 08 Jan 2005 21:00:26 -0500, Bryan Hoover
<bhoover@wecs.com> wrote:


>> If I were to cut a hole on one of those, I'd put in a
>> 120x38mm Panaflo (FBA12G12L1A)
>> undervolted, or the current cheaper option, FBA12G12M
>> ["M"=Med instead of "L"ow speed, just undervolt more/less]
>
>Yeah, I'd rather bigger, quieter fan too.
>
>> http://www.bgmicro.com/prodinfo.asp?prodid=FAN1080
>
>That's a sweet deal!

Yep, 120mm fans are often higher priced but sometimes it's
just silly what a place will charge for a Panaflo, not to
mention Papst.

>
>> Installed about 2/3 of the way down towards the back such
>> that it blows under video and also above towards upper mobo
>
>I don't know -- I want that VRM to be "suffocating" with air flow, and it's a little higher up. I think the
>video's getting enough air from the intake, exhaust, and cpu as it is. Certainly it's the hotest of any card
>that'll ever be plugged into the board (don't you think?), and it is not, by any means, too hot to touch as it
>is. Higher placement would get the chipset too.

Is the chipset instable?
There isn't necessarily any need to cool the northbridge
beyond having a good margin of stability.

Unless you were overclocking, the VRM FETs should stay cool
enough too, providing overall chassis airflow is sufficient.
To a certain extent it comes back to basics, that the same
parts most vulnerable for most, might be for you. That
might include video (caps & fan(?)) and motherbaord caps.

Touch-test the motherboard caps... forget about the temp
measurements the system provides as they're not of a part
that has substantially short lifespan in the ranges
mentioned. The CPUs on the other hand, have a lower stable
temp, but it goes without saying that they must be kept cool
enough.

So my take on this is that until you know the temps of the
caps it's not really known whether there is any particular
need to go to elaborate measures putting more air on that
region, beyond simply increasing overall chassis flow. If
you really wanted highest rates on the VRM you'd not use a
side-panel fan at all, rather a bracket-mounted one just a
few cm above the area.

The suggestion for location of side-panel fan also took into
consideration that ultimately, you'll end up with a case
having good/quiet airflow that can be reused for the next
build, as the platform is upgrade-limited at this point,
eventually you may upgrade for performance reasons.

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

kony wrote:

> On Sat, 08 Jan 2005 21:00:26 -0500, Bryan Hoover
> <bhoover@wecs.com> wrote:
> >Yeah, I'd rather bigger, quieter fan too.
> >
> >> http://www.bgmicro.com/prodinfo.asp?prodid=FAN1080
> >
> >That's a sweet deal!
>
> Yep, 120mm fans are often higher priced but sometimes it's
> just silly what a place will charge for a Panaflo, not to
> mention Papst.

Prolly a stupid question, but that doesn't need the com line on it in order to control speed through the board does
it? I mean, such control does not generally relie on rpm feedback information to do the job does it? I don't mind
not being able to rpm info, but I would like to be able to control the speed via the SpeedFan monitoring program.>

> >> Installed about 2/3 of the way down towards the back such
> >> that it blows under video and also above towards upper mobo
> >
> >I don't know -- I want that VRM to be "suffocating" with air flow, and it's a little higher up. I think the
> >video's getting enough air from the intake, exhaust, and cpu as it is. Certainly it's the hotest of any card
> >that'll ever be plugged into the board (don't you think?), and it is not, by any means, too hot to touch as it
> >is. Higher placement would get the chipset too.
>
> Is the chipset instable?
> There isn't necessarily any need to cool the northbridge
> beyond having a good margin of stability.
>
> Unless you were overclocking, the VRM FETs should stay cool
> enough too, providing overall chassis airflow is sufficient.
> To a certain extent it comes back to basics, that the same
> parts most vulnerable for most, might be for you. That
> might include video (caps & fan(?)) and motherbaord caps.
>
> Touch-test the motherboard caps... forget about the temp
> measurements the system provides as they're not of a part
> that has substantially short lifespan in the ranges
> mentioned. The CPUs on the other hand, have a lower stable
> temp, but it goes without saying that they must be kept cool
> enough.
>
> So my take on this is that until you know the temps of the
> caps it's not really known whether there is any particular
> need to go to elaborate measures putting more air on that
> region, beyond simply increasing overall chassis flow. If
> you really wanted highest rates on the VRM you'd not use a
> side-panel fan at all, rather a bracket-mounted one just a
> few cm above the area.

Yeah, spot cooling. That's a thought too. Figured against it because the VRM does not really lend itself to
putting a heatsink or anything on it. But maybe a bracket. How would you do that? How do you mount a bracket --
from the edge of the board? Drill the board?

> The suggestion for location of side-panel fan also took into
> consideration that ultimately, you'll end up with a case
> having good/quiet airflow that can be reused for the next
> build, as the platform is upgrade-limited at this point,
> eventually you may upgrade for performance reasons.

I'm gonna keep this baby for the museum -- Tyan s2462 -- world's first dual AMD platform!

Bryan

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Bryan Hoover wrote:

> David Maynard wrote:
>
>
>>Bryan Hoover wrote:
>>
>>
>>>David Maynard wrote:
>>>
>>>
>>>>Bryan Hoover wrote:
>>>>
>>>>>David Maynard wrote:
>>>>>
>>>>>
>>>>>>"ONE of the LM75s"? What does that mean?
>>>>>
>>>>>I think it's one of two cpu temp sensors. There may be some boggle in
>>>>>how the sensor information is given in the winbond chips -- on this
>>>>>board, according to SpeedFan documentation, the LM75 readings are
>>>>>aliased windbond chip information -- which begs the question, why are
>>>>>these temps different from what is otherwise given as cpu temps. In
>>>>>short, I don't know. But SpeedFan lists two of these temps, which
>>>>
>>>>I've,
>>>>
>>>>>after some research in a crash course of which I can now only remember
>>>>>about what I'm telling you now, taken to be cpu temp sensors. SpeedFan
>>>>>lists two temps labeled as LM75-1, and LM75-2, and other board temps,
>>>>>which can be, by deduction (comparison with Tyan System Monitor, and by
>>>>>which winbond chip address the temp is associated), mapped to the
>>>>>corresponding item being sensed.
>>>>
>>>>Well, part of my problem is I have no idea what 'other temps' are nor
>>>>what
>>>>you're monitoring other than partial tidbits like 'one of the LM75s"
>>>>and I
>>>>can't see it, ya know
>>>>
>>>>I'm guessing you mean you have two CPU temps for both CPUs? I.E. 4 total?
>>>>But the 'two' for one processor don't match each other?
>>>>
>>>>My off hand stab in the dark would be that one is an in socket thermistor
>>>>and the other is monitoring the die diode.
>>>>
>>>>Which tyan board is this?
>>>>
>>>
>>>Tyan s2462 (without manufacturer's onboard scsi option).
>>>
>>>Seperate sensor placement is what I was thinking (hoping actually 'cause
>>>that'd be rigorous) too, but I just discovered I'd been using the wrong
>>>command line switch to SpeedFan :|. Instead of TIGERMP, I'd been using
>>>TIGERMPX. Now I've got 4 LM75 temps, along with the readings I had
>>>before. The 4 LM75 readings appear to be duplicates of other readings
>>>-- so it's as the SF documentation said -- these are aliases of other
>>>readings. The readings are:
>>>
>>>VRM-1: 54C
>>>VRM-2: 52C
>>>CPU-1: 52C
>>>CPU-2: 51C
>>>DDR: 43C
>>>AGP: 41C
>>>LM75-1: 51C
>>>LM75-2: 52C
>>>LM75-3: 41C
>>>LM75-4: 43C
>>>HD0: 42C
>>>
>>>These are while running ATI tivo application.
>>>
>>>Incidentally, last evening running this app, and having turned down one
>>>of the fans so that VRM1 temp had started peaking up around 61, I got
>>>that memory error as I was moving around in the video record, and
>>>playback. So there's seems to be an association with the error, and the
>>>temp, but it's not "automatic" as far as reproducing the error on
>>>demand. Does seem to be a connection though. I agree that 61 does not
>>>seem like it would cause problems. Tyan monitor program gives normal
>>>range for all temps, up to 65 with low as -10.
>>>
>>>SF 12V at the moment is 11.07. Tyan monitor shows it at 13.9. Not much
>>>help, 'til have multimeter.
>>>
>>>Bryan
>>>
>>
>>Ok. I have some info, then. (Searching for related Linux info can be real
>>useful in things like this because they're always stuck doing it themselves
>>and you can often find discussions, or a solution, posted)
>>
>>http://www.vikora.com/doc/mbmon/ReadMe.gz
>>
>>There is a list of the motherboard sensors because the Tyan is a 'special'
>>case with a particularly peculiar setup procedure.
>>
>>1st sensor (SMBus, W83782D)
>>
>> temp0 VRM2 temperature
>> temp1 CPU1 temperature
>> temp2 CPU2 temperature
>> Vcore0 CPU1 Vcore
>> Vcore1 CPU2 Vcore
>> Volt 0 AGP voltage
>> Volt 1 system 5V
>> Volt 2 DDR voltage
>> Volt 3 -----
>> Volt 4 standby 3.3V
>>
>> 2nd sensor (ISA IO, W83627HF)
>>
>> temp0 VRM1 temperature
>> temp1 AGP temperature
>> temp2 DDR temperature
>> Vcore0 CPU1 Vcore
>> Vcore1 CPU2 Vcore
>> Volt 0 system 3.3V
>> Volt 1 system 5V
>> Volt 2 system 12V
>> Volt 3 system -12V
>> Volt 4 -----
>
>
> Also, if you're interested, check out the info here:
>
> http://www2.lm-sensors.nu/~lm78/news.html (seach page for '2462', and you may
> want to look at associated (related Tyan boards) links too).
>
> I'm sure you can make more sense out of than I -- that is, there's is some
> context lacking unless you are familiar with probing around these
> chips/addresses, I think.
>
> I'm still looking at these pages -- I'd pulled them up when I was first looking
> into this. Maybe they'll make more sense to me now that I've had time to digest
> a little of that over the past few months.
>
> Bryan
>

Yeah. I found half of those through other links but unless you want to know
how to write your own monitoring program you don't really need the info
they're dealing with there.

I just thought the description of the sensor inputs was interesting because
earlier you had been talking about 'case temp' and, regardless of what
speedfan might assume, there isn't one. CPU, VRM, AGP, and DDR but not 'case'.

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

Bryan Hoover wrote:

> kony wrote:
>
>
>>On Sat, 08 Jan 2005 21:00:26 -0500, Bryan Hoover
>><bhoover@wecs.com> wrote:
>>
>>>Yeah, I'd rather bigger, quieter fan too.
>>>
>>>
>>>>http://www.bgmicro.com/prodinfo.asp?prodid=FAN1080
>>>
>>>That's a sweet deal!
>>
>>Yep, 120mm fans are often higher priced but sometimes it's
>>just silly what a place will charge for a Panaflo, not to
>>mention Papst.
>
>
> Prolly a stupid question, but that doesn't need the com line on it in order to control speed through the board does
> it? I mean, such control does not generally relie on rpm feedback information to do the job does it? I don't mind
> not being able to rpm info, but I would like to be able to control the speed via the SpeedFan monitoring program.>

Speedfan controls fan 'speed' by open loop PWM modulating the fan 'on/off'.

RPM monitoring is simply RPM monitoring and is uninvolved in the fan speed
control.


>>>>Installed about 2/3 of the way down towards the back such
>>>>that it blows under video and also above towards upper mobo
>>>
>>>I don't know -- I want that VRM to be "suffocating" with air flow, and it's a little higher up. I think the
>>>video's getting enough air from the intake, exhaust, and cpu as it is. Certainly it's the hotest of any card
>>>that'll ever be plugged into the board (don't you think?), and it is not, by any means, too hot to touch as it
>>>is. Higher placement would get the chipset too.
>>
>>Is the chipset instable?
>>There isn't necessarily any need to cool the northbridge
>>beyond having a good margin of stability.
>>
>>Unless you were overclocking, the VRM FETs should stay cool
>>enough too, providing overall chassis airflow is sufficient.
>>To a certain extent it comes back to basics, that the same
>>parts most vulnerable for most, might be for you. That
>>might include video (caps & fan(?)) and motherbaord caps.
>>
>>Touch-test the motherboard caps... forget about the temp
>>measurements the system provides as they're not of a part
>>that has substantially short lifespan in the ranges
>>mentioned. The CPUs on the other hand, have a lower stable
>>temp, but it goes without saying that they must be kept cool
>>enough.
>>
>>So my take on this is that until you know the temps of the
>>caps it's not really known whether there is any particular
>>need to go to elaborate measures putting more air on that
>>region, beyond simply increasing overall chassis flow. If
>>you really wanted highest rates on the VRM you'd not use a
>>side-panel fan at all, rather a bracket-mounted one just a
>>few cm above the area.
>
>
> Yeah, spot cooling. That's a thought too. Figured against it because the VRM does not really lend itself to
> putting a heatsink or anything on it. But maybe a bracket. How would you do that? How do you mount a bracket --
> from the edge of the board? Drill the board?

Don't even THINK of drilling holes in a multilayer circuit board.

>>The suggestion for location of side-panel fan also took into
>>consideration that ultimately, you'll end up with a case
>>having good/quiet airflow that can be reused for the next
>>build, as the platform is upgrade-limited at this point,
>>eventually you may upgrade for performance reasons.
>
>
> I'm gonna keep this baby for the museum -- Tyan s2462 -- world's first dual AMD platform!
>
> Bryan
>

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

On Sun, 09 Jan 2005 04:58:38 -0500, Bryan Hoover
<bhoover@wecs.com> wrote:

>kony wrote:
>
>> On Sat, 08 Jan 2005 21:00:26 -0500, Bryan Hoover
>> <bhoover@wecs.com> wrote:
>> >Yeah, I'd rather bigger, quieter fan too.
>> >
>> >> http://www.bgmicro.com/prodinfo.asp?prodid=FAN1080
>> >
>> >That's a sweet deal!
>>
>> Yep, 120mm fans are often higher priced but sometimes it's
>> just silly what a place will charge for a Panaflo, not to
>> mention Papst.
>
>Prolly a stupid question, but that doesn't need the com line on it in order to control speed through the board does
>it? I mean, such control does not generally relie on rpm feedback information to do the job does it? I don't mind
>not being able to rpm info, but I would like to be able to control the speed via the SpeedFan monitoring program.>

I can't speak for your particular board, but fan control is
not usually tied to RPM, I dont' recall it ever being tied
to RPM but it is certainly possible that some board would,
some day... if not today. Generally the control is dumb
(perhaps user setting for ratio) or based on temp, with no
need for RPM feedback. As for Speedfan control though, I
actually MUCH MUCH prefer a hardware solution, permanent
speed reduction. One never knows when a crashed system or
other bug will cause a fan not hard-wired to a power rail to
stop spinning. To me fans, even those very low RPM, are a
basic safeguard that should not depend on "anything" else,
except of course power going to the system and basic
direct-connect 12V rail traces on a motherboard. This is
drifting a bit, I'm thinking mostly of essential fans. This
side-panel fan certainly isn't essential to operation as the
machine already runs ok, albeit questionable temps.



>> So my take on this is that until you know the temps of the
>> caps it's not really known whether there is any particular
>> need to go to elaborate measures putting more air on that
>> region, beyond simply increasing overall chassis flow. If
>> you really wanted highest rates on the VRM you'd not use a
>> side-panel fan at all, rather a bracket-mounted one just a
>> few cm above the area.
>
>Yeah, spot cooling. That's a thought too. Figured against it because the VRM does not really lend itself to
>putting a heatsink or anything on it. But maybe a bracket. How would you do that? How do you mount a bracket --
>from the edge of the board? Drill the board?

It's a whole lot harder to generalize about without having
the system in front of me. Sometimes I'll epoxy heastinks
onto FETs, those 'sinks having tapped or finned allowance
for screws to mount a fan. Other times a bracket, attached
wherever is convenient, depending on the available space and
time + tools available to do it. There might be other parts
facilitating the mount instead of directly to the board, yet
still putting the fan close enough to be very effective.

I would not try to drill into the board at all, and around a
VRM, it would be good to avoid permanently blocking access
to capacitors... despite all cooling measures sometimes
you'll just get a bad one or new/unknown variant that seems
defective.



>
>> The suggestion for location of side-panel fan also took into
>> consideration that ultimately, you'll end up with a case
>> having good/quiet airflow that can be reused for the next
>> build, as the platform is upgrade-limited at this point,
>> eventually you may upgrade for performance reasons.
>
>I'm gonna keep this baby for the museum -- Tyan s2462 -- world's first dual AMD platform!


It's a nice board, great for it's time... the industry just
moves on too fast but we can't necessarily call it a bad
thing, nobody is forced to upgrade. I do agree with your
approach to trying for maximum lifespan, and find focusing
on fans and caps to be one of the best ways to do so,
ignoring hard drives which one can't do a lot about beyond
the obvious, good power/cooling/spinup-interval.

 

[Guest]

Archived from groups: alt.comp.hardware.homebuilt (More info?)

David Maynard wrote:

> Bryan Hoover wrote:
>
> > kony wrote:
> >>So my take on this is that until you know the temps of the
> >>caps it's not really known whether there is any particular
> >>need to go to elaborate measures putting more air on that
> >>region, beyond simply increasing overall chassis flow. If
> >>you really wanted highest rates on the VRM you'd not use a
> >>side-panel fan at all, rather a bracket-mounted one just a
> >>few cm above the area.
> >
> >
> > Yeah, spot cooling. That's a thought too. Figured against it because the VRM does not really lend itself to
> > putting a heatsink or anything on it. But maybe a bracket. How would you do that? How do you mount a bracket --
> > from the edge of the board? Drill the board?
>
> Don't even THINK of drilling holes in a multilayer circuit board.

Okay. I won't .

Bryan

> >>The suggestion for location of side-panel fan also took into
> >>consideration that ultimately, you'll end up with a case
> >>having good/quiet airflow that can be reused for the next
> >>build, as the platform is upgrade-limited at this point,
> >>eventually you may upgrade for performance reasons.
> >
> >
> > I'm gonna keep this baby for the museum -- Tyan s2462 -- world's first dual AMD platform!
> >
> > Bryan
> >