Lost on LM386

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Administrator
Joined 2007
Paid Member
Do you mean R2 in the diagram ? If so then that is part of a 'Zobel' network and required to keep the amp stable. Most amps have such a network.

Headphones really need a series resistor of perhaps 47 or 100 ohm (for a low power amp like the LM386). Whether you can do that depends on the socket. If you use the socket to disconnect the speaker automatically then you might be able.
 
This resistor is not in the diagramm, because I first built without headphone out.

The Headphone socket is switched and the switched tip is connect over to the speaker out. Non switched tip and ring on the headphone socket are shorted to get stereo signal. And the headphone plug negative has a 120R and this is connected with its own cable to the negative leg of the 470uF.

So far, so good?
 
Administrator
Joined 2007
Paid Member
Fair enough :)

Changing the series ground resistor to the headphones should be dramatically altering their overall loudness. The higher the resistor the quieter they get. I don't know if that happening or not.

The LM386 should easily drive headphones to any possible level needed and so if its distorting past a certain volume setting then that suggests the chip has reached its limits of voltage swing on the output... but that should be way over what any low impedance (are they low impedance ?) headphone should require.
 
So, this is a tradeoff then? I have to find the right value for the headphones not to crackle and not to loose too much volume.

Or can I improve this by going from 9V to 12V?

IMHO this looks way more chaotic now, but HOORAY, it works. I'll trim the cables when moving it to a new housing.
 

Attachments

  • DSC_0669-1587x1190.JPG
    DSC_0669-1587x1190.JPG
    307.6 KB · Views: 142
  • DSC_0672-1587x1190.JPG
    DSC_0672-1587x1190.JPG
    274 KB · Views: 150
Administrator
Joined 2007
Paid Member
Without further tests it is all guesswork.

Ideally you should monitor the output on an oscilloscope to see just how much voltage swing is present at the output. In other words, is the amp clipping. That said, an running off 9 volts should absolutely be able to blow your ears off with headphones.

Going to a higher supply voltage (within the limits of the chip) is usually a good option.

Another thought is that you are feeding the amp very low bass signals that would cause distortion anyway in a headphone driver. As an experiment you could try adding a small cap in series with the input as shown here. This would create a high pass filter limiting extreme bass (that probably isn't audible anyway via headphones or small speakers)

Try a 0.1uF (assuming the pot is 10k) and see if that helps. If it is to bass light then increase the cap to a 0.22uF and so on. A 0.1uF will give a -3db point of 160Hz, a 0.22uF would give -3db at 72Hz.

The cap must be added to the input side of the pot in order to maintain the DC path from pin 3 to ground. Worth trying.
 

Attachments

  • Capture.JPG
    Capture.JPG
    81.3 KB · Views: 149
Many headphone amplifiers have been built with the 386. Musicians love them because you can (or could) buy everything at Radio Shack and stick wire them on perfboard.

On 9 volts, it can provide more than enough power to blow your eardrums with low impedance cans. It's also noisy, high distortion, not very precise. But it's cheap and easy.

Raising supply voltage will increase linearity a little but shouldn't be necessary. Check the max rating on your chip (some are 12 volts and some 18 volts).

The datasheet is always worth a look. http://www.ti.com/lit/ds/symlink/lm386.pdf Figure 10 is fine and dandy for headphone amp, with a resistor in series with the output. Try values from 10-50 ohms. You can even use an array of resistors and a selector switch. Find the values that work best with the headphones you are going to use. You can have a different resistor for every headphone in the house. Old school 600 ohm cans might need 50 to 120 ohms. You can also put a capacitor on the input. Try 4.7 uF for f3 = about 3 Hz.
 
I see what Mooly's talking about. You can put the dominant pole on the input for lower distortion. 0.47 uF gives 34 Hz.

Once you calculate your pole for one capacitor value, then you can scale it up and down and avoid calculating the pole for every value. ;)

Then you have to look at your output pole. Make it lower than the input pole by at least a factor of 2.

If you sort it out like that, then you will have made that turd as shiny as it can get. Careful consideration of values will make a difference!
 
That did not work. I have added a 100nF cap to the input as marked by Mooly. Then I tried a 100nF and a 47nF in paralell (thats 147nF, right?). No change. Then I tried a 1uF, also no change.

Finally I tried what Fast Eddie suggested. A 100nf on the IN and a 47nF on the OUT of the volume poti. No change. Then a 2.2uF on IN and a 1uf on OUT. Nothing.

@Fast Eddie: Where is F3 in Figure 10? I can see only two caps?
 

Attachments

  • Schematic_LM386-Amp_Sheet-1_20180527185534.png
    Schematic_LM386-Amp_Sheet-1_20180527185534.png
    39.2 KB · Views: 151
Last edited:
It could be it sounded different, but it did not stop the crackling at higher volumes on the guitar (80 percent and up).

I tried a different guitar and it is quite the same.

And I am sorry, I did not understand a word you two said after my last post. Neither could I understand the link you gave me.
 
@Fast Eddie: Where is F3 in Figure 10? I can see only two caps?

f3 is standard nomenclature for -3 dB point. It is a consistent and simple standard for calculating poles.

Finally I tried what Fast Eddie suggested. A 100nf on the IN and a 47nF on the OUT of the volume poti. No change. Then a 2.2uF on IN and a 1uf on OUT. Nothing.

No. There should be no cap on the output wiper of the pot. The input bias current has to flow through it. The cap only goes right on the input of the circuit.

And I am sorry, I did not understand a word you two said after my last post. Neither could I understand the link you gave me.

This stuff is confusing enough for beginners without misunderstandings. I was referring to low frequency poles, which are simple but require careful consideration for proper circuit operation.

That article talks about high frequency poles, which you do not have to deal with here. Consider yourself lucky.
 
Administrator
Joined 2007
Paid Member
No problem, lets keep it simple :)

This shows the effect of the 0.1uF added to the volume control. You can see how the bass rolls off. If we make it a 0.047uF the bass is even less. The idea was to see if extreme amplitude (but inaudible) bass sounds were overdriving the amp and/or your headphones.

If you can raise the supply a little and you then find it goes louder before crackling then that would suggest the amp is clipping.
 

Attachments

  • Capture.JPG
    Capture.JPG
    200.4 KB · Views: 125
  • Capture2.JPG
    Capture2.JPG
    193.6 KB · Views: 122
An electric guitar has a very dynamic output. Uncompressed it would surely clip even at very moderate volume.

Small practice amps of yesteryear (look up the Pignose*) used transformer coupled circuits with no global feedback. This provided soft clipping without the horrendous distortion that can result from clipping a global feedback amplifier.

As an illustration of the dynamic nature of an uncompressed electric guitar signal, my buddy in high school ran his guitar through his hi fi system, which was pretty powerful. We cranked that system for hours playing records, but just one note on the guitar instantly fried his tweeters.

* The Pignose used germanium transistors, input and output transformers, and ran off a 9 volt battery. You could crank the crap out of it and it would produce that "wail" that much more expensive guitar amps produced when cranked but SPL was only around 93 dB.:) Musicians loved them and also used them to overdrive the input of their big tube amps.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.