Hi
I'm working on a project right now related to sonar. So in the hardware I have a signal of around 2KHz given to a filter and then to a LM386 audio amp circuit.

Now at the output I've connected an 8 ohm speaker. Now when I connect the setup, the speaker gives a tone. But to check if all's good I check the output with an oscilloscope and I see a lot of distortion and no proper sinewave. It's almost as if the wave is clipped but its peak to peak isnt greater than the amp supply voltage..so the amp is not saturated. So I disconnect the speaker and check again and then I find a perfect sinewave at the output of the LM386.

So I thought something is wrong with the speaker. So next I connect a signal generator generating a 2kHz sinewave and give to the speaker directly. Then I measure the waveform again at the input of speaker...Now again I get a perfect sinewave on the oscilloscope.

So I'm thinking that whatever distortion I'm getting is coz of some mismatch between speaker and amp output impedance or sort of. Now I have several stages of speaker each having its associated filter and amp. So a few stages a functioning perfectly even after connection. The only difference between the erroneous and functional stages is the IC. The functional IC is LM386N nd the defective one is LM386N-1. Both are of national semiconductor...So I try to replace all ICs with a different make...but i cant get my hands on another make..so I have to make do with this itself. Now since the amps are functionalwithout connecting speaker, so I'm thinking theres some deeper issue. Can you give me some solution to this? Thanks

I've attached an image of the distortedwaveform after connecting the speaker.

The amplifier circuit is also attached.

 

It doesn't look like distortion to me as much as it looks like oscillation.

Some ideas:

• Use a 100 uf power supply decoupling capacitor across the 0.01 uf capacitor already there.

• Make sure the path from in 5 through R1 and C3 back to pin 4 is physically short and that C3 is a good quality ceramic capacitor.

• If those don't correct the problem, try reducing the gain by putting removing the capacitor between pin 1 and pin 8. If that solves the problem but you need more gain, reconnect the 10 uf capacitor but with the largest series resistor that gives the gain you need.

 

What are you using for the power supply? Isn't 18V too much for a LM386?

 

What are you using for the power supply? Isn't 18V too much for a LM386?

Yea my bad
..Im actually using just a 5 volt supply.

 

It doesn't look like distortion to me as much as it looks like oscillation.

Some ideas:

• Use a 100 uf power supply decoupling capacitor across the 0.01 uf capacitor already there.

• Make sure the path from in 5 through R1 and C3 back to pin 4 is physically short and that C3 is a good quality ceramic capacitor.

• If those don't correct the problem, try reducing the gain by putting removing the capacitor between pin 1 and pin 8. If that solves the problem but you need more gain, reconnect the 10 uf capacitor but with the largest series resistor that gives the gain you need.

Thanks. I'll surely try this.

 

Presenting us with false information about the supply voltage can only interfere with getting you a good result.
(ps, 18 VDC is the maximum allowed on the first page of the datasheet, then the spec changes on the next page.)
I checked the p-p ratings and conclude the same as everybody else. This is not a load problem, it's a decoupling problem.

 

Audio power amp ICs are ultra cranky about decoupling and grounding. Keep the decoupling cap leads as short as possible and get them as close as possible to the IC pins. Also, big fat ground runs. Use at least two caps in parallel such as a 0.1uF ceramic and a 47uF aluminum electrolytic.

Plus, 5V is not much headroom for the amp to work with. At 5V Vcc and an 8 Ohm speaker you will have less than 2.5V p-p output swing before clipping.

ak

 

Hey,
I soldered a 100uF electrolytic capacitor between pin 6 (+5 volt) and pin 4 (ground) and it did the trick. Thanks for the help. Can you explain the theory behind this? How does this decoupling work?