1
\$\begingroup\$

PCB layout

The 6 SMD MOSFETs in the center generate a total heat of about 80w. I was planning on using this heatsink (ATS-1106-C1-R0) with this fan (08025SS-24Q-AT-D0) mounted directly above it.

Looking at the PCB layout, the fan hub will take up a good amount of the heatsink area. I assume there will be very little airflow underneath it so I was wondering whether this will be an issue.

Do you think I should find a different place to mount it, or maybe leave a gap between the heatsink and the fan? I'm also unsure about whether to mount the fan blowing up or down.

Edit: The rectangle in the center is the heatsink (the inner part is the footprint on the PCB, the outer part is the size at the top of the heatsink since the fins spread apart), the smaller circle is the fan hub, the bigger circle is the fan, the big rounded square is the fan housing.

Edit 2:

I swapped the 120mm fan for an 80mm one, which is the smallest size for which the screw holes will fit around the heatsink.

Some extra info:

  • I'm designing this as a prototype to be used without a case around it. If I ever end up putting this in a case, it will have enough airflow not to affect performance too much.
  • MOSFET's max channel temp is 150°C
  • MOSFET's channel-case Rth is 0.462°C/W

As I understand it:

This means that the max case temperature is 150°C - (80W/6 * 0.462°C/W) = 143.84°C

At 25°C ambient, the heatsink would need to have a Rth of (143.84°C-25°C) / 80W = 1.48°C/W or less.

The heatsink's Rth ranges from 1.3°C/W at 1m/s airflow to 0.6°C/W at 4m/s. The fan has an airflow of 1.55m³/min which equates to 4.04m/s over an area of 8cm*8cm.

I suspected that I shouldn't be expecting an actual Rth of 0.6°C/W with the fan mounted above the heatsink, but I was wondering how bad it would actually be. By 'an issue' I meant that the MOSFET's channel temperature would get close to or exceed the maximum of 150°C.

I mounted the fan on top of the heatsink in the first place since that seemed like the most obvious solution and because that takes up the least amount of PCB area. I'm not sure how I would mount it vertically.

\$\endgroup\$
8
  • \$\begingroup\$ Use a smaller fan, or a bigger heatsink. \$\endgroup\$
    – DKNguyen
    yesterday
  • \$\begingroup\$ @DKNguyen I edited the post to make clear what is what. \$\endgroup\$
    – Cecemel
    yesterday
  • \$\begingroup\$ Understood...... \$\endgroup\$
    – DKNguyen
    yesterday
  • 1
    \$\begingroup\$ This is really a fluid-flow question, and probably belongs in engineering.stackexechange.com (that's a more general group, but lots of mechanical engineers tend to congregate there). \$\endgroup\$
    – TimWescott
    yesterday
  • 1
    \$\begingroup\$ why are you not using horizontal airflow across fins rather than vertical which blocks the air velocity and flow significantly? \$\endgroup\$ yesterday
3
\$\begingroup\$

Will it cause an issue with your intended design? Yes , several, but not unless you have specs or expectations. Will it cool down? yes but sub-optimal in several aspects.

  • heat flux loss from coplanarity and mismatched apertures, and suboptimal velocity and heat extraction. More specs and geometry of enclosure are necessary.

You should start with specs for Tj max, Tamb max and thus Rth(cs,sa) max somewhere <<0.5 'C/W pref 0.2'C/W.

Air velocity up to 10m/s significantly lowers Rth(ca) while plenum area impedes velocity depending on 3D turbulence or 2D laminar flow exponentially. But you ought to achieve at least 2m/s over hotspot surface.

Recirculating the heat raises internal Ta. High CFM or cu.m./s does not guarantee highest linear air speed over hotspots so is less important than air speed.

Thus max laminar velocity is best over many parallel fins used on CPU heat sinks and max turbulent flow over flat hot spots is best on PCB with a plenum cover works better. Quieter CPU heat sinks rely on optimizing velocity flow with minimum turbulence with load curves by choice of fin thickness and spacing with number of fins. Laminar requires great height or length, turbulence is synonymous with eddy currents in short path gradients so shape matters on the edges.

So for lowest cost and highest noise, directly over a CPU heatsink works well. But a rectangular plenum area of fan diameter reshaped to end geometry with laminar flow exhaust pulling turbulent air from heatsink achieves quiet removal of heat from enclosure. Dell did this with in inline fan plenum to case >20yrs ago for effective cooling and office quiet.

These could be small diameter thick blades but high RPM to match enclosure constraints. (e.g. 1U high) If not possible then a copper heat pipe to same is what is commonly used in laptops.

\$\endgroup\$
1
  • \$\begingroup\$ somebody does not like good advice -1 or is ignorant or incompetent \$\endgroup\$ yesterday
2
\$\begingroup\$

Unless the board itself gets hot, any airflow outside of the footprint of the heatsink is wasted.

For a center-mounted fan you'd probably do better with a 60mm or even a 40mm fan because you'd have a smaller dead spot in the center.

Better yet if your heat sink has fins mounted radially, so the air has a short outward path as it's blown onto the heatsink by the fan.

Possibly even better yet if you use such a fan mounted next to the heatsink, with straight fins, and a plenum so the air is forced across the fins.

This is not intended as a product recommendation, but just a note: you can not only buy heat sinks with fan mountings, but you can buy heat sinks with fans mounted on them. So if you're seriously looking at doing this there's a solution that's probably more cost-effective than rolling your own.

\$\endgroup\$
2
  • \$\begingroup\$ Thanks for the note about heatsink-fan assemblies. What exactly is a plenum? Is it like a duct to force the air from the fan through the heatsink fins? Google wasn't very helpful. \$\endgroup\$
    – Cecemel
    21 hours ago
  • \$\begingroup\$ The wiktionary article on "plenum" is interesting. In this context it's a volume holding air at above-ambient pressure, from which (hopefully) it is distributed at the flow rates needed. \$\endgroup\$
    – TimWescott
    15 hours ago

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.