LM4780TA: multiple same power pins

Okay so it may work? Here's a picture
IMG_2207.jpeg
 
I guess the bondwires inside the IC package may act as fuses when you don't connect all supply and ground pins.
That would be my guess as well. A bond wire can typically handle about 1 A. The LM3886 (half an LM4780) can source 11 A, typical. Also, there's no guarantee that the various V+ and V- pins are connected inside the package, so you really need to connect to all of them.

This + bypassing to gnd should be fine right?
I don't see any bypassing. Are we talking about the same thing?

Keep in mind that you need to attach the LM4780 to a heat sink, so you'll need some clearance along the back side of the IC.

Tom
 
  • Like
Reactions: catcream
The layout clearly shows how the V+ and V- pins are connected together, as well as where/how to decouple the voltage rails:

1717902383506.png


I highlighted in red the ground trace that should be a bit thicker (the one that leads toward the V+ decoupling caps combo). You could use copper ribbons here (if doing the work on vero-board)... this will help decouple the V+ rails properly.

I strongly recommend that you feed the whole IC / every single IC pin through the vero board, and then start to connect its pins as required on the underside.

Good luck.
 
I meant what I had in pic + some capacitors fron the rails to gnd pins. Is it fine to do small film cap soldered directly to pins and a ceramic + larger electrolytic on perfboard?
Sure. You'll want 0.1-1.0 uF ceramic or film in parallel with 10-33 uF electrolytic as close to the chip as practical. You can solder the ceramic caps directly to the pins of the IC and the electrolytic caps where the two supply wires exit the IC.

My thoughts on decoupling of the LM3886 (and LM4780) are found here: https://neurochrome.com/pages/supply-decoupling

Tom