My first LM386 build went like this. I had a bad salvaged capacitor that'd failed short in the output filter, and it worked out just like Joe said and gave the chip a viking funeral. This taught me to always invest in chip sockets.
Definitely start with the speaker and speaker connections as per Joe's instructions so you're sure there isn't a short in there.
With the battery and speaker disconnected (to be safe) you should see a very high resistance or open circuit between pins 4 [gnd] and 5 [out] (either side of the arse end of the chip from the dot at pin 1). If you don't see at least 1MΩ then it may be toast. If you're measuring with a meter that has a multi-megaohm range you might get some odd readings, especially if the chip's in circuit or you're wearing a static strap. That's normal, as high impedance measurements are difficult to do well. If the chip failed as a short (hard to tell without removing it) it might hamper your ability to find the short elsewhere in the circuit.
The big output cap (220µF) and the low pass filter cap (47nF) block the output from having a DC path to ground. If either of those caps goes short, or if there's any way for the output to just dump current to ground, it'll be likely to toast the output stage of the chip.
On these chips too much gain won't hurt anything, it just sounds horrible. You can stick any resistor you like between pins 1 and 8 to change the gain, and for super duper gain you can stick a 10μF cap in there, but that's only really useful for ultra low signals (RF mixers and magnetic tape heads, etc.). I've never seen an LM386 with open gain pins, so if it were me I'd try 10k in there just to have something to make me feel better. The volume pot's also optional, as it just provides an adjustable parallel load to go with the speaker, so it's absence won't really contribute to your problem.
Hope that helps some.
Definitely start with the speaker and speaker connections as per Joe's instructions so you're sure there isn't a short in there.
With the battery and speaker disconnected (to be safe) you should see a very high resistance or open circuit between pins 4 [gnd] and 5 [out] (either side of the arse end of the chip from the dot at pin 1). If you don't see at least 1MΩ then it may be toast. If you're measuring with a meter that has a multi-megaohm range you might get some odd readings, especially if the chip's in circuit or you're wearing a static strap. That's normal, as high impedance measurements are difficult to do well. If the chip failed as a short (hard to tell without removing it) it might hamper your ability to find the short elsewhere in the circuit.
The big output cap (220µF) and the low pass filter cap (47nF) block the output from having a DC path to ground. If either of those caps goes short, or if there's any way for the output to just dump current to ground, it'll be likely to toast the output stage of the chip.
On these chips too much gain won't hurt anything, it just sounds horrible. You can stick any resistor you like between pins 1 and 8 to change the gain, and for super duper gain you can stick a 10μF cap in there, but that's only really useful for ultra low signals (RF mixers and magnetic tape heads, etc.). I've never seen an LM386 with open gain pins, so if it were me I'd try 10k in there just to have something to make me feel better. The volume pot's also optional, as it just provides an adjustable parallel load to go with the speaker, so it's absence won't really contribute to your problem.
Hope that helps some.