Four Layer LM3886 Layout

[simonra]

Is the problem having to use a larger size, or having to change the layout?

There's the idea of replacing it (just glancing, there's barely enough rooom) two 0805s each of half the resistance in series. That way they each get half the voltage and dissipate half the power.

Originally I wanted to stick to 0805 for compactness and I'd managed to get the layout pretty tight for the critical components... I've now got a 1206 for the feedback resistor. I think I'll still try some different components in there to see if I can measure differences in performance.

For the assembly aspect: If one pin of a component has a big fat trace connecting to it and a skinny trace at the other end, the fat trace will remove far more heat than the skinny trace. This may cause the solder at one junction to flow well while the other end has a cold solder joint (or hasn't flowed at all). That's why we use thermal reliefs.

Not an issue with DIY and not much of an issue with passives. With tight pin pitch parts, such as LLPs and BGAs, the part may actually shift during reflow as it gets pulled by the surface tension of the molten solder, which causes all sorts of fun effects.
Tom

Okay so this is DIY and not too much of an issue but there's no harm in trying to follow good practice. I can easily change a couple of the pads so that the traces come in at 90˚, not so sure how I can balance the amount of copper on both ends of the component if one end terminates on a pour and the other with a trace... for example it looks like the left pad of R3 would take longer to heat than the right - how would I go about fixing this (pick a better example if I'm missing something)?

Thanks,
Simon.

 

[tomchr]

Okay so this is DIY and not too much of an issue but there's no harm in trying to follow good practice.

My point exactly.

I can easily change a couple of the pads so that the traces come in at 90˚, not so sure how I can balance the amount of copper on both ends of the component if one end terminates on a pour and the other with a trace... for example it looks like the left pad of R3 would take longer to heat than the right - how would I go about fixing this (pick a better example if I'm missing something)?

As I noted earlier, it's mostly an issue for BGAs and LLPs where the ICs can get pulled far enough to short the pins. I'm not overly worried about it for passives as long as you have reasonable thermal reliefs.

Tom