If a midtower case, has a fan in the front (in) and fans in the
back(out) and AIO fans on top (out), does it generally need additional
fans (and if so, where would you put them---to cool what?)?

I was thinking of getting an NVidia 1660 Super or RTX 3070 if that
changes the equation by a great deal.

Thank you for your suggestions.

On 1/11/2023 9:24 AM, Bill wrote:
>
> If a midtower case, has a fan in the front (in) and fans in the
> back(out) and AIO fans on top (out), does it generally need additional
> fans (and if so, where would you put them---to cool what?)?
>
> I was thinking of getting an NVidia 1660 Super or RTX 3070 if that
> changes the equation by a great deal.
>
> Thank you for your suggestions.

If the fans are arranged to directly route the hottest air out of the
case, great numbers may be excessive, and will increase dust collection.

My approach is to have 1 big fan with a foam filter pulling air into the
case, and the back fan and power supply fan carrying the CPU heat out
with cardboard baffles to direct the hot air directly to the exhaust fans.

When I had a high power GPU, I cut a slot in the case side over the GPU
and baffled that also to direct it's exhaust directly out the side of
the case.

On 1/11/2023 12:24 PM, Bill wrote:
>
> If a midtower case, has a fan in the front (in) and fans in the back(out) and AIO fans on top (out), does it generally need additional fans (and if so, where would you put them---to cool what?)?
>
> I was thinking of getting an NVidia 1660 Super or RTX 3070 if that changes the equation by a great deal.
>
> Thank you for your suggestions.

Case cooling equation.

For air cooling a 182W PC, allowing a 14F degree internal
case temperature rise...

CFM = 3.16 * Watts / Delta_T_degrees_F

= 3.16 * 182 / 14F = 28CFM

One exhaust fan (35CFM "medium" fan) should move that much air on its own.
An AIO group of fans on top, could be easily helping with this.

Some motherboards have an ambient sensor, and that can
be used to get a "case temperature" for the "case temperature rise"
measurement. But you need agreement with how you are measuring
room ambient, for that sort of thing.

The video cards have a power spec, and they will draw
exactly that power when maxed out. My 180W video card draws
183W and reduces the clock to make it stay that low. It's
because of that, if you measured the PC power with a Kill-O-Watt
meter today, before the video upgrade, you will have a pretty
good idea of a ballpark power number for when the video card
is fitted. You can watch the behavior of the card with "GPU-Z"
which displays the "state" of the card -- power limited or
clock limited or "idle". The NVidia smoke particle demo is
"clock limited" and card power is 1/3rd of rating. Not all the
shaders are engaged.

As it turns out, the PC in the above equation, can draw about
400W now, so it needs more fan-age than the above example. It
switched from a 13W video card to a 180W video card, which is
why it huffs and puffs now. The 13W video card was actually nice,
but it no longer got driver updates and I had to get rid of it
so I could run newer OSes.

Pull or push cooling, makes a difference to dust precipitation
inside the PC.

If you use filters on a PC, they have to be cleaned. This is
only practical, if the filter is easy to slide out. The filters
that exit from the bottom of a PC case, may not be all that convenient
to work with.

By measuring delta_T between inside the case and outside the case,
you don't need the cooling equation, and you can keep increasing
the airflow out of the case, until the delta_T is where you want it.
This requires running Prime95 and Furmark, to heat up the PC so you
can measure the worst case. On my current PC, I tested with Prime95
about a week ago, and discovered no cooling air is hitting my VCore
heatsinks. A problem now, I have to fix. I haven't located an easy
and ideal solution yet, either. The two Noctua candidates that
might have worked, are offset designs, and may not fit the way I
want to fit them. One of the Noctua designs, lacks a dimensioned
engineering drawing on the web site. How lazy can these people be ?
They could not *make* the thing, without that dwg.

Paul

On 1/11/2023 2:25 PM, Paul wrote:
> On 1/11/2023 12:24 PM, Bill wrote:
>>
>> If a midtower case, has a fan in the front (in) and fans in the
>> back(out) and AIO fans on top (out), does it generally need additional
>> fans (and if so, where would you put them---to cool what?)?
>>
>> I was thinking of getting an NVidia 1660 Super or RTX 3070 if that
>> changes the equation by a great deal.
>>
>> Thank you for your suggestions.
>
> Case cooling equation.
>
>    For air cooling a 182W PC, allowing a 14F degree internal
>    case temperature rise...
>
>    CFM = 3.16 * Watts / Delta_T_degrees_F
>
>        = 3.16 * 182 / 14F = 28CFM
>
> One exhaust fan (35CFM "medium" fan) should move that much air on its own.
> An AIO group of fans on top, could be easily helping with this.
>
> Some motherboards have an ambient sensor, and that can
> be used to get a "case temperature" for the "case temperature rise"
> measurement. But you need agreement with how you are measuring
> room ambient, for that sort of thing.
>
> The video cards have a power spec, and they will draw
> exactly that power when maxed out. My 180W video card draws
> 183W and reduces the clock to make it stay that low. It's
> because of that, if you measured the PC power with a Kill-O-Watt
> meter today, before the video upgrade, you will have a pretty
> good idea of a ballpark power number for when the video card
> is fitted. You can watch the behavior of the card with "GPU-Z"
> which displays the "state" of the card -- power limited or
> clock limited or "idle". The NVidia smoke particle demo is
> "clock limited" and card power is 1/3rd of rating. Not all the
> shaders are engaged.
>
> As it turns out, the PC in the above equation, can draw about
> 400W now, so it needs more fan-age than the above example. It
> switched from a 13W video card to a 180W video card, which is
> why it huffs and puffs now. The 13W video card was actually nice,
> but it no longer got driver updates and I had to get rid of it
> so I could run newer OSes.
>
> Pull or push cooling, makes a difference to dust precipitation
> inside the PC.
>
> If you use filters on a PC, they have to be cleaned. This is
> only practical, if the filter is easy to slide out. The filters
> that exit from the bottom of a PC case, may not be all that convenient
> to work with.
>
> By measuring delta_T between inside the case and outside the case,
> you don't need the cooling equation, and you can keep increasing
> the airflow out of the case, until the delta_T is where you want it.
> This requires running Prime95 and Furmark, to heat up the PC so you
> can measure the worst case. On my current PC, I tested with Prime95
> about a week ago, and discovered no cooling air is hitting my VCore
> heatsinks. A problem now, I have to fix. I haven't located an easy
> and ideal solution yet, either. The two Noctua candidates that
> might have worked, are offset designs, and may not fit the way I
> want to fit them. One of the Noctua designs, lacks a dimensioned
> engineering drawing on the web site. How lazy can these people be ?
> They could not *make* the thing, without that dwg.
>
>    Paul

Thank you Paul. You shed a lot of light on the issue!
FWIW, while doing a bit of research on my case a few months
back, I recalled that it had a slide out "filter/screen" on the bottom,
for the power supply I expect. I checked the filter and (much to my
surprise) it did not have
much dust there--whereas the filter on the front of the case draws a
fair amount of dust and I clean it every couple of months. The case sits
on a wooden board, on its feet, so I am not blocking the filter on the
bottom.
I have a Corsair RM750 PSU, and that particular model is designed to
"run quietly"--so maybe it doesn't use it's fan that often. I just
copied the following from the Corsair website: "The RM Series' cooling
fan doesn't spin until you need it, for near-silent operation at low and
medium loads." I don't push my system often. I bought a higher
capacity PSU than I really required to allow the possibility of a GPU
upgrade.

Thank you for helping to "continue my education"!