Is it dumb to strap an old stock CPU cooler onto my modem?

I put a stock Intel cooler on a cable modem a couple years back, same situation, cabinet under the TV, case too hot to keep your hand on. Mine's still running, but I made one change from your plan: I pulled the fan off the heatsink and ran it passive. The fan is the part that'll bite you. Those stock fans have cheap sleeve bearings and they're designed for a desktop that gets turned off at night, not 24/7 duty. Give it six months of dust buildup in a enclosed cabinet and it'll slow down or seize, and then you've got a chunk of aluminum sitting on your chip trapping heat instead of moving air. If you really want moving air, plug a seperate little USB fan into the router's power supply or a nearby outlet and just have it blowing into the cabinet. When it dies, you'll hear it, and swapping it takes thirty seconds instead of opening the modem again. On the fit question, don't stress about the overhang. What matters is contact over the chip itself. But also check where the actual heat is coming from before you commit. On my modem the big flat chip on top ran warm but the real furnace was a smaller chip near the power input, barely anything on it from the factory. Touch test after it's been running an hour tells you pretty quick. I put a small heatsink on each.

For mounting, there are no screw holes on a modem board, so plan on thermal tape or a dab of adhesive epoxy. Works fine for something this light.

And do the vent holes regardless. Even with a heatsink, a sealed cabinet just recirculates the same warm air. A few holes low on the door and high near the back gives you free convection, and that alone made a bigger difference than anything I bolted onto the chips.

I find the discussion regarding the thermal interface material quite interesting, as simply applying a substance is rarely sufficient for optimal heat transfer performance. If you are using a thermal pad, you absolutely must ensure the entire surface area of the integrated circuit is covered, and that the mounting pressure exerted by the cooler is distributed evenly rather than concentrating pressure on just one corner of the die. Uneven compression dramatically reduces the effective heat dissipation capacity. For those using a thermal paste, a thin, even layer is key, and I've seen some situations where over-application actually leads to poor thermal transfer due to trapped air pockets under the excess compound.

I'm curious about the mounting piece. When I put a heatsink on a small circuit board, I found that just using tape made it look kinda flimsy. I ended up using a thin metal strap I salvaged to keep the heatsink stable. That really helped prevent the base from bowing or warping under the pressure. Also, I think the main chip didn't even cover all the hot spots, which was annoying. I noticed some little surface components were still baking.

I went thru this a while back on a small server, and the standard thermal pad setup wasn't cutting it. I noticed the contact points were uneven even with good pressure. What made the difference for me was swapping the thermal pad out for a small strip of compressed silicone foam. It really filled the tiny air gaps between the heat spreader and the chip better than the gooey stuff, even if it meant slightly uneven pressure. It gave me much more consistent heat transfer right across the entire component.

Update: went passive like the first reply suggested. Pulled the fan off the heatsink and just left the heatsink sitting on the chip, since the modem barely makes any heat on its own in an open cabinet. I also borrowed the strap idea from the mounting reply, a bit of flat steel from a cheap bracket keeps it from tipping. Case is warm but no longer painful to touch.