Which pins to dissipate heat for SO-8 regulator?

Nov 26, 2007 3 Replies

First reply

John Larkin

Bust one open and see. I'd guess that the GND pin is the chip paddle, which is where most of the heat comes out. The in/out pins are probably wire bonds, which don't conduct heat very well. In 317 type parts, Vout is the paddle, and the heat comes out there, on 4 pins for SO-8.…

Read the full reply

In the absence of specific guidance from the datasheet, which pins on a SO-8 regulator are the most significant for heatsinking?



I observe that there are more Vin than Vout pins, that Vin is probably the drain of a mosfet, and I conclude that Vin is probably what needs the best thermal connection. Is this sound reasoning?



(specifically in this case I'm using an MCP1726)


Ben Jackson AD7GD http://www.ben.com/

Bust one open and see. I'd guess that the GND pin is the chip paddle, which is where most of the heat comes out. The in/out pins are probably wire bonds, which don't conduct heat very well.

In 317 type parts, Vout is the paddle, and the heat comes out there, on 4 pins for SO-8. On 337's it's Vin, again 4 pins.

How much power do you plan to dissipate? If it's marginal, put a power pad on the pcb below the chip and epoxy it to that. Or use a chip that has better thermals. Their value of 150 K/W for the SO-8 sounds optimistic.

John

The MCP1726 has a DFN package version with an exposed power pad (EP), to remove heat. The exposed pad is connected to the substrate, which is at ground potential. You get a low RthJA = 41 deg-C/W thermal resistance by soldering to that, as opposed 150C/W (3.6x worse) with the soic package. With a 150C maximum junction temperature, and 25C ambient temp., this means you can dissipate 3W vs. 0.8W, without any added safety margin. Call it 2W vs 0.5W, or 4x more.

A lot of times on power devices they use a heavier bond wire. I don't have specs handy, but I recall 1mil standard and 1.3mil heavy. Anyway, the heat carrying ability of the bond is proportional to the cross sectional area, so a slight increase in width is a big increase in heat transfer via bonds.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required