Author Options:

LM386 amplifier overheating? Answered

I built a simple LM386 based audio amplifier (schematic attached), but it appears to be over-amplifying, as it makes horrible squeaking noises and picks up a lot of interference. It also gets very hot in use, so I only switch it on for a few seconds at a time. I'm guessing this is because of the high amplification as I have a commercial amplifier that uses an LM386 and it does so without a heatsink, although unfortunately I can't reverse-engineer it and even if I did, my circuit would almost certainly have to be remade. I tried changing the potentiometer today from 1k to 10k and even 100k, but they didn't have much affect. What component(s) should I change to give ti less amplification, I only need about 2-6 times amplification, preferably adjustable but it doesn't matter too much.


You have set the gain at 200 !!!
Remove the 1uF from 1 to 8

and add a 0.047uf in series with a 10 Ohm from 5 to ground, for stability.

The MINIMUM gain of a 386 is 20, so you have to do some attenuation on the input - mind you, you have a 40dB attenuation there right now. Ditch the 100K and put a 100K LOG pot on the input instead. See if that's where you need to be.


Ahh thanks, that would explain a lot. It would speak if you touched the battery, and pick up interference from just touching different parts of the circuit, and even my pulse. I will try this tomorrow and post the results. I don't actually have any log pots apart from a tiny 1k round one I pulled from some old circuit, so I might have to buy one, but for now I will try it with a gain of 20, It might be ok. I have a 1.5W speaker I want to drive from the output of a sound recorder which is normally connected to a 0.25W speaker, so I am assuming to get it to drive a 1.5w, I want 6 times amplification.

No, its not that simple. You want POWER amplification, from a very tiny input power, to 1.5W, you may still find 20 is enough !


I just removed the 1uF, and it is definitely a lot quieter, so thanks for that, but it still overheats. Do I still need to add the resistor and capacitor? You said "for stability", so I guessed it is just for noise cancellation and I was going to add it tomorrow.

No, a lack of stability is high frequency oscillation, which leads to catastrophic overheating.


I just added the cap and resistor and it still overheats, I am using a polymer cap marked with 473J100, which is a 0.047uF capacitor with 5% tolerance and a brown black black gold brown resistor which is 10 ohms with 1% tolerance, so I don't see a problem here. I have now done everything you said except the log potentiometer, but I thought that was just for volume control, and it is actually the right volume as it is, but it still overheats.

Could be damaged I suppose, check what current it draws with no input signal. It should be ~4mA


Yea, it varies from about 3-6mA and doesn't heat at all when removed and stuck in a breadboard on it's own. I thought it was possible that it could be damaged, but it did appear to still work fine.

Yes, but it seems to heat up a lot slower. I can keep it on for about a minute before it gets really hot instead of a few seconds with the speaker attached. It is a 50mm 8 ohm 1.5W mylar speaker, if that helps.

Sure your output capacitor's OK ?

Do you have an oscilloscope or access to one ?


It should be fine, it's unused and in no way damaged. I did have a small 1MHz digital oscilloscope, but unfortunately a friend broke it and I haven't bought a replacement yet. I may remake the circuit on breadboard as per the datasheet as my matrix board circuit is looking rather messy after so many modifications.

I suspect its oscillating somehow. Just try 100uF across the supply, and a 100nF in parallel with that.


I'll try that in the morning, but I have very little faith left in this circuit. I wouldn't have thought it would be oscillations, as I am powering it directly from a 9v battery and it is on a single board with no AC supplies nearby. I may just go out and buy a new oscilloscope, I was looking at the UNI-T UT81B, it's an 8MHz digital scope with a autoranging multimeter with capacitance and frequency testers, and a PC interface for about £130.

My board is becoming very cluttered now so I will probably remake it if it does work. I will try and breadboard the datasheet schematic now as I think I have enough parts inside to do so (I keep most of my components and soldering iron in the garage, and it's nearly midnight). Thanks for all the help so far, hopefully I can get this working soon.

I just breadboarded the circuit pretty much as per the schematic, I didn't have any 250uF capacitors or 10 ohm resistors inside, but I made a resistor from a long piece of wire and stuck two 100uF and two 10uF capacitors in parallel, giving 220uF, and it still overheats. Maybe a faulty or damaged chip? Then again, it did give the correct current readings. I may just buy a few more, they are only about 30p each if you buy 10 on eBay. I don't have a stock of them as I don't use them much, and instead I use 555s, 741s, CMOS, PIC, PICAXE, and AVR chips.

Not. 220uF 100uF in parallel with 100NANO farads (0.1uf) usually a ceramic cap !

Try another chip - I'd never buy off ebay for ICs there have been too many scams were the ICs are essentially scrap devices that failed testing in some way.

Yes, I know this, I was breadboarding the "gain 20 minimum parts" circuit from the schematic, I haven't tried your suggestion of the capacitor filter on the supply yet. I usually buy most components I want in bulk from eBay as they are a lot cheaper and I haven't had any problems with them so far, and individual components from Maplin.

This LM386 is actually from Maplin, so it should have been working correctly when purchased. I do remember breadboarding it actually, and I seem to remember it working fine, otherwise I wouldn't have continued to matrix board it. It could well be that it overheated at some point, but I am still curious as to how this happened so I can prevent it in the future.

When soldering it, I used an IC socket, so there is no danger of it overheating, and I've checked the tracks and they are definitely as per the schematic.

I think I'll go out again today or tomorrow and buy a few more chips and make the database schematic with a gain of 20.

I just got back from Maplin and bought their 1W LM386-based amplifier kit, assembled it, and it works fine, so it must have been a faulty chip.

It's not a solution to your question but take a look at the PAM8403. It is tiny, more poweful (2x 3W), good quality sound and it's very efficient and it's cheap (around $1 for complete board)! You will be amazed by this board.


2 years ago

Hi there. I too built a LM386 audio amplifier circuit on breadboard, set the gain to 20 and hooked up an 8 ohm 3W speaker. The chip started too heat up until i can't even touch it. Then I replace the 3W speaker with an 8 ohm 0.5W speaker. The chip didn't heat up.

So i suppose the 3W speaker is drawing way too much power from the chip, thus overheating it. I checked up the datasheet of LM386 and found out that when Vcc is 9V and speaker is 8 ohm, the typical output power is only 700mW. So I think your 1.5W speaker is drawing too much power from the chip. Change it to a speaker with lower power rating.


7 years ago

If Steve's suggestions haven't worked, check the output coupling capacitor (C1).

Someone here had a similar problem because they omitted that cap (which blocks DC bias voltage from the speaker). Could be yours is damaged...