For awhile, I was providing the 'pin-fin' BGA type heatsinks on my rectifier boards. These are the short ones with an array of heat dissipating 'spikes'. After giving them a second look - I found that these are optimized for forced air cooling rather than convection. Need heatsinks more optimized for convection rather than forced air for pinball machines. The basket type heatsinks tended to have convection thermal resistance values (about 9.4) much lower than the pin-fin heatsinks (15 or higher) ... the lower the better.

Hard to tell on the extruded heatsinks such as those used by Williams on their WPC boards as there is no thermal resistance value listed. These vary widely depending on construction and size.

Some people use the TO-220 folded type heatsinks such as the 563002 from Aavid. These work but aren't real good... thermal resistance of about 13.

Attached are pictures of the pin-fin, basket and folded type heatsinks.

Thermal compound also makes a difference but not a huge one. Depending on the type of compound - this adds a thermal resistance between 0.8 to 1.2. Good stuff runs about 0.8 whereas the cheapy paste with silicone runs about 1.2 with a myriad of versions between.
Can get Hi-Flow heatsink pads (stick on pads such as Berquist HF115TAAC) - these are about 0.9. Not bad but cost much more for the convenience of not being messy and coming in the form of a sticker. Complete lack of heatsink compound results in a huge thermal resistance rise but actual amount depends on surface roughness.

Now... the real question is how good is good enough.
So many variations, impossible to answer but shoot for a thermal resistance as low as you can go and you should be ok.

To cover a tiny bit about thermal resistance - this tells you how much a part will heat up while dissipating a certain amount of wattage. To get total thermal resistance - add parts junction to case resistance (bridge rectifier = 1.2) plus case to sink resistance (thermal compound roughly = 0.9) plus heatsink-to-ambient resistance (folded heatsink = 9.4) for a total thermal resistance of 11.5.
Multiply this times the wattage dissipated to get temperature rise above ambient temperature (25C). So with two diodes always on within a bridge rectifier with a current of 3 amps x forward voltage drop of 1.1v... bridge typically dissipates about 2x3x1.1= 6.6 watts. 6.6 watts x 11.5 C/watt = 75.9C rise. Add ambient temp. of 25C and you have a temperature of just over 100C... quite hot! We could now go into why Williams went from GBPC type bridges to 6A4's but too much info...
There... clear as mud?

Ed

heatsink1.jpg heatsink2.jpg heatsink3.jpg