The gap between the CPU & the heat sink came in at 1.25mm.  It was spanned by 5 layers of thermal tape.  This gooey compliant mechanism held the heat sink in pretty securely while not transferring vibration.

Tried an aluminum plate 1.5mm thick, with thermal paste on 1 side & thermal tape on the other side.  This expanded the gap to 1.75mm  in the hope it would conduct better.  The heat sink was no longer held in by the tongue & groove or any adhesive.  It was less secure but more vibration would be transferred.  There was no obvious improvement in cooling.

Thermal tape & aluminum have different tradeoffs.  Aluminum should be giving it a hair more margin, but noted as it's handled & the heat sink is bouncing around, the thermal paste probably isn't staying optimally distributed.  You wouldn't want just 2 pieces of adhesive & metal.  It needs to be more compliant to avoid breaking a solder joint.

Verified reducing the battery voltage doesn't affect it.  It still bottoms out at 4.5V because of wire resistance.  The only other ideas are a bigger fan or heat pump.

It will never support any modes above 8W but what does work is still better than the gopro 7.  The gopro 7 maxed out at 4k 4:3 30fps with no stabilization.  Now, we're maxing out at 4k 4:3 30fps with stabilization.  The great task is using it outside, without a fan.  Tests without a fan indoors failed in all modes.

-----------------------------------------------------------------------------------------------------------------------------------------------

The trick with the truck is if the camera breaks free, the heat sink is going to separate & rip the flex cable off.  The strapping mechanism needs to be permanent.

Kicked things off with a fanless design & aluminum wire farsteners to keep things together.  Wire has proven easier to use than flex PLA & more reliable than paperclips.  The hope is flexible wire can stay in place.  This quickly overheated at 6W in 80F ambient, in very little wind.  A stiff tailwind bought it more time.  So it would need the 30mm fan at minimum.

The full enclosure with 30mm fan came in at 114g.  The gopro 7 with its full enclosure was 111g.  Note the 7 had a full dummy battery while the 11 had directly soldered leads, but the 7 was driven with a real battery for years.  The 7 with a real battery & enclosure was 129g.

8W has been the maximum in 80F ambient.  Maximum wattage decreases if the ambient temperature is over 80F.  86F ambient basically kills it with any wattage & the 30mm fan.  Noted the standby power is 3W & it still roasts without a fan.

There's a chance running the mobile app makes it a hair hotter.  The 86F hard limit indoors doesn't bode well for any outdoor use.

----------------------------------------------------------------------------------------------------------------------------------------

More optimized mounts & further experiments followed, running 5V fans at 7V, 12V fans at 14V.  Noted a few useless 50 fps modes by setting the flicker speed at 50Hz.  The only useful modes seem to be 6W.  All we're doing by testing higher wattages is increasing margin.

At this point, the original reasoning of making a paw held mount to run in a race was dead.  The cooling system & batteries would come to the size of a Hi8 camcorder.  Also, lion fitness was looking promising enough to justify going as light as possible.

----------------------------------------------------------------------------------------------------------------------------------------

It managed to go for over an hour in outdoor high 80's, with the 30mm fan, at running speed.  The fan was running on 7V from the headlight circuit.  The rotor fell off all the time until it was lost.  After that, the camera quickly overheated.  The fan overheats too, without a rotor.  Once the lubrication dries out, there's no way to add lubricant without destroying the C clamp so they're landfill.  There could be a strap to keep the rotor in.

The gopro also needs the fan when connected to USB.