LM386 Resonance Mystery

[onoezitasploded]

Hi All,
I'm building a little theremin/synth thingy as a present for my sister, who's a musician. For the I/O and tone generation, I'm using an Arduino microcontroller, and for the audio output I'm using an LM386 amplifier and a little 8ohm speaker I pulled out of a clock radio. I'm not expecting awesome audio quality out of this setup, but I've got this odd resonance thing happening when I connect the speaker.

Here's a picture of the trace on my oscilloscope without the speaker connected:


Looks exactly like the intended waveform. But, once I connect the speaker to the amp, I get this:


That fuzzyness on the bottom of the waveform is audible as a high pitched tone. I'd like to get rid of it, if possible.

I'm using the example circuit in the datasheet, with the addition of a low-pass filter before the circuit to get rid of the pin noise from the microcontroller. For convenience, here it is:


Any help is greatly appreciated, sorry if this is a common question but I don't know what it's called so I don't know what to search for. Thanks!

[DigitalJunkie]

I'm not sure if this will help your issue,but it might help to put a bypass cap on pin7. 10-22uf to ground should be good.

[itsmrdavid]

What about the bypass pin (7)?
Take a look at application notes.

I know you can skip it, but i'm not sure what effect that has.

Also -- bypass the shi-dookie out of the supply; maybe 1000uf or more, plus small ceramic for HF bypass.

Might help

[onoezitasploded]

I tried putting a bypass cap on pin 7 to ground, with a variety of small values-- 2.2uF, 10uF, 33uF with no effect.

I've currently got the power supply bypassed with a 4000uF cap, and I also added a .1uF ceramic bypass, with no effect. I don't think it's a power supply problem, because when the amp is just idling, there's <10mV ripple voltage coming through the audio.

I did notice one thing: when I sweep it through frequency, the bounce or fuzz or whatever is minimal around 250Hz, and increases as I change the frequency away from 250Hz.

Thanks for your help so far, I am ???

[Conrad Hoffman]

The LM386 is famous for this sort of thing. If you can use one of the newer IC power amps, you'll find it far easier. Layout is critical- any path from output to input will cause oscillation so the 386 is very sensitive to grounding. Single point recommended. Think about how current to the two ground points to the right can cause a voltage rise on the pot and into the positive input. All traces have resistance. I have a whole tube of the silly things and shudder whenever I have to use one.

[kaos]

That kind of oscillation is usually related to speaker load inductance and an inadequate output zobel. The original zobel values for this family of amplifiers was .1 uF in series with a 2.7 ohm resistor (as opposed to the .05 uF and 10 ohms shown in the application notes for the LM386). If you mess with these values you may have a fair chance of eliminating the oscillation. Try going up to .1 uF first, if that doesn’t cure it, start lowering the R value in conjunction with the .1 uF cap. Good luck …

[onoezitasploded]

Solved!

Putting a .1uF cap + 10ohm resistor across output and ground, instead of the .05uF shown in the schematic cleaned it right up!

Oh, it sounds so much better. tytytytytytytytytytyty

I'll post back with a link when I finish the project, it's coming together nicely.

[onoezitasploded]

Hey all,
I finished the project and it came together pretty well:



I did a pretty thorough write-up on my blog if anyone's interested, it might be helpful if you're trying to get a similar project off the ground:

http://blog.wingedvictorydesign.com/...l-walkthrough/

Thanks once again for the help!