I'm having trouble to understand why this is happening when I power my LM386 on a 9V battery with no load.

The schematic is the basic application from the LM386 datasheet with the gain of 20 but without the speaker load at the output.

enter image description here

When I measure the AC voltage at Vs I get a really high frequency signal (10MHZ) of small amplitude (10mv).

Now the thing is that this noise get amplified at the output too. This wouldn't matter much if I was using a speaker as a load (because I shouldnt be able to hear that frequency I guess) but I want to use a vumeter as a Load.

I tried the following: 1. Adding a 220uf cap at the power trail, this helped a bit and reduced the amplitude by 50%, but I'm not sure if that's the right approach and I would like to remove completely this noise.
2. Adding the zobell network, this didn't do anything.
3. Adding a fake load (a 10K resistance) at the output, didn't do anything.
4. Also grounding all the floating wires to ground didn't help

Why is this happening and how can I remove this signal ?

0.2us/div 5mv/div enter image description here

  • \$\begingroup\$ Show some schematic \$\endgroup\$ – PlasmaHH Sep 9 '16 at 8:20
  • \$\begingroup\$ Don't talk, show the schematic. \$\endgroup\$ – Darshan Chaudhary Sep 9 '16 at 8:23
  • \$\begingroup\$ add more uFage to the power rail, then edit with your results. Batteries have a high internal impedance, they need lots of uFs to be stiff into an audio amp. 220uF is nowhere near enough. Try 1000uF, perhaps 4700uF. \$\endgroup\$ – Neil_UK Sep 9 '16 at 8:28
  • \$\begingroup\$ Make the leads as short as possible. Long leads like you have can cause instability. \$\endgroup\$ – Leon Heller Sep 9 '16 at 8:41
  • \$\begingroup\$ Try adding a 0.1uF in parallel with the 220uF - large capacitors don't handle high frequencies very well so adding a small capacitor improves the decoupling range. \$\endgroup\$ – JIm Dearden Sep 9 '16 at 8:41

Your photos do help - at 10 MHz., circuit layout is important. You are likely on the right track trying to ensure that Vs is clean DC. At 10 MHz., a 220uF capacitor might have too much internal inductance and/or resistance to work effectively. Try a smaller capacitor, more suitable for radio-frequency bypass: for example - a 0.1uf ceramic type. Value is not critical, 0.1uf capacitors are very common.

And to reduce the inductive path from Vs to ground, add this bypass capacitor to your breadboard directly from pin 6 to pin 4 of the LM386. You are quite correct to proceed only when the DC supply is free of any oscillations. This problem occurs so frequently, that a bypass capacitor is the very next component that I add to the breadboard after the chip is inserted.

Why is it happening? Gain inside the LM386 chip is quite high, even though it is reduced by internal feedback. Gain only needs to be a bit more than one for oscillations to occur. Normally, we assume that "ground" is always unvarying, and that "Vsupply"(Vs) is unvarying as well. Wiring inductance allows the chip itself to vary these voltage nodes, even though the battery supplies steady DC. Sometimes, a LM386 will oscillate at lower audio frequencies when a heavy load like a speaker is inserted. This is often caused by a feeble battery whose internal resistance is too high, allowing the DC supply voltage to vary. In this case, a large value supply bypass capacitor (like 470uF) may make it good.

|improve this answer|||||
  • \$\begingroup\$ Thanks, the part in bold solved it, putting the cap directly from pin 6 to 4 completely removed this oscillation. I read it was a good practice to do it but I wouldn't imagine that it would be that dramatical ! I learned my lesson. \$\endgroup\$ – foobarcode Sep 9 '16 at 16:45
  1. Bypass the power supply leads.
  2. GIVE IT A REAL LOAD. The LM386 was designed to operate into an 8 ohm load, with roughly 220-250 uF of electrolytic DC blocking capacitor. Operating it into no load is not going to make it happy.
  3. If your plan is to drive a VU meter, you do not need anything remotely resembling the drive that an LM386 can provide. Make a better choice of component. I'd start with a garden variety op amp and a couple of resistors. Verify that it will drive the VU meter.
|improve this answer|||||

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.