LM386 headphone amp, i need to lower gain

[Lino22]

Hi guys, i built this LM386 based headphone amp



Now it has gain about 20, which is way too much for my needs. I added the 10k input voltage divider, but it really works just at the end of the path, where the part to ground is about 300 ohms.

My question is: is there any other way i could lower gain apart from shunting the signal down through 300 ohms?
There is an unofficial way to make the gain a bit lower, using the pins 1 and 8, but the official LM386 data sheet talks just about making it higher.

[anotherjim]

I would add a resistor in series with the output to the phones.

[Vivek]

The input impedance of the LM386 is about 50K ohms

It will interfere with the volume pot, changing its taper

A) just to make sure, did you use a log taper Audio volume control ?

B) Try adding 50k or 100K or 200K resistance in series, in between pot wiper and pin 3. That would act like a fixed voltage divider reducing the signal hoing into the LM386, and it would increase the impedance seen by the wiper, allowing it to better follow its taper.

C) In one application where I needed to control the volume of a LM386, I used 5K log pot . This was to ensure that 50K input impedance of LM386 does not have too much of effect on the taper


I am not for adding resistance in series with the headphone, since that changes the damping factor. I would rather try all kinds of tricks to reduce the input seen by pin 3

[Vivek]

I dont know enough of the internals of the LM386

Maybe this works to reduce gain of the LM386

Idea being to feed part of signal to - pin as well as + pin

[antonis]

If you want to reduce GAIN (not input signal amplitude..), place a resistor across pins 1 & 5..
(preferably in series with a DC blocking cap..)
Gain should be calculated according to schematic formula, where Rf should be the parallel equivalent of 15k and shunt resistor..
(by placing a pot, wired as variable resistor, across pins 1 & 5 you could lower the gain down to unity - but gains lower than 9 aren't advisable 'cause LM386 might be unstable for such low gains..)

[Lino22]

Thank you guys. Here is the whole circuit.




It is a charge amp that has a headphone output.

I used a log 100k for the volume pot, but it really feels more log than what i would expect, is that the 10k trimpot that  adds its resistance between the wiper and ground?
And there is a weird crackling noise right before it gets to the maximum volume.

[antonis]

C5 and LM386 input cap polarities fight each other at various pots settings..

P.S.
10k pot severely dominates VR3 effective value and taper...

[Vivek]

Looks like the 10k is a trimmer, meant to adjust gain of headphone section (set and forget)

And then volume is controlled by 100k pot



But 10k trimmer is too small and will alter the taper of the 100k volume pot.

Not very smart design

[Lino22]

C5 and LM386 input cap polarities fight each other at various pots settings..

P.S.
10k pot severely dominates VR3 effective value and taper...

Is there any way to sort out the cap problem?
I will try to change the trimpot to 100k, or even 250k. That with 100k linear volume pot should create about 50k log volume pot. Maybe i can leave it out completely and add just the resistor across 1 and 5? Or just a 100k resistor on the input pin 3?

[Lino22]

Looks like the 10k is a trimmer, meant to adjust gain of headphone section (set and forget)

And then volume is controlled by 100k pot



But 10k trimmer is too small and will alter the taper of the 100k volume pot.

Not very smart design

Vivek i am sorry, i should have posted the whole circuit. It is not very smart, i agree

[anotherjim]

Are the headphones too loud or is the 386 distorting too easily (probably both)? If they are simply too loud, just add a resistor in series with the phones output - that's what commercial products do and it's a safe way to do it too. => 100R. The amp output cap value can then be smaller. If max volume is ok but it still distorts too easily, place a series resistor in the feed to the phones volume pot lug 3. It's going to => 10k.

10kA pot is a good match for the 386 input. Ideally, you would use a 4k7A or 5kA pot but the 50k input of the 386 doesn't change the log taper of the 10k too much in practice and 10k pot keeps the input impedance closer to the expected 10k load for line audio.

The 386 inputs are ground-friendly, you don't need an input coupling cap unless you want to cut bass to the phones in which case the cap will be =<1uF.

Why have the headphone vol dependent on the amp volume? You can take the feed from vol lug 3. Also, the load of the 386 input won't affect the amp vol pot taper this way.

[antonis]

Maybe i can leave it out completely and add just the resistor across 1 and 5? Or just a 100k resistor on the input pin 3?

I'd experiment with both mods simulataneously..

Make VR3 10k, delete LM386 input cap, place a resistor across pins 1 & 5 and place a trimmer in place of 10k (its value should depend on desirable signal attenuation 'cause it should form a voltage divider with 50k internal resistor between pins 3 & 4)..

[PRR]

[Lino22]

Paul, thank you. Thank you everybody, it is working the way it is supposed to, ha ha. I didn't realize how much those 47R on the output help because of the high current through the headphones (right?). The 100-500k before the 100k trimpot wasn't even necessary. I rewired it as Paul suggested and the volume works like a charm. I omitted the 100n cap for now.

[PRR]

...I omitted the 100n cap for now.

You are right; that was pointless.

[Vivek]

47 Ohms in series with headphones can alter the bass response

It changes the damping factor

It changes the resistance that the reverse current created by the headphone sees

Way to test : Play a bass note and suddenly stop the note, See if some sound still lingers on in the headphone.

If you cannot hear the difference, I suppose it does not matter. Im sure some will hear the difference.

Damping Factor (DF) is the amplifier's ability to control speaker motion once a signal has stopped. Technically speaking, Damping Factor is the ratio of nominal loudspeaker impedance (the impedance the loudspeaker is rated at) to total output impedance of the system driving the loudspeaker, including the amplifier and cables.

Damping Factor changes with frequency (as does impedance) and is most noticeable at lower frequencies. A high DF typically results in a tighter, more controlled bass, which is usually (but not always) more desirable from a listener's point-of-view. Low DF results in soft or fat bass.

-https://ca.kef.com/blogs/news/damping-factor-explained

https://www.aes.org/aeshc/pdf/how.the.aes.began/preisman_loudspeaker-damping.pdf

[Lino22]

I didn't know that. Thank you Vivek.

[PRR]

47 Ohms in series with headphones can alter the bass response

I worried about this, like 20 years ago.

I had 300r phones with a 600r peak. Driving them with 680r made a small difference. 33r made no difference at all. (And I was using these for monitoring live recording quality.)

I tested lots of "Walkman/iPod" phones and buds around 28-42 Ohms. None of them showed even 10% rise at resonance. A few showed 10% rise at the top of the audio band. 27 Ohms series made no audible difference on any. (Except a lot of them overloaded easy and some series resistance delayed that to higher applied voltage.)

Since Lino was saying "too loud", I 'erred' on the side of less level even if less "accurate" (no headphones are that accurate anyway).

Anyway all musical speakers have more or less resonances, some quite spectacular in a sine-sweep, and musicians play around that.

[anotherjim]

I would not worry about the resistor upsetting the response. Many amplifiers add phones to the main speaker output with a switched jack that inserts resistors to cut the level in the phones.
There is also a concern for the safety of a person's ears. You could limit the gain/input level all you like, but what about power on/off pop or fault? The attenuation after the amp will cut that too.

[Rob Strand]

I've got similar comments to Jim and PRR.

Many years ago I measured the impedance of a lot of headphones.  Mostly reasonable quality and the response variations were minimal and often at quite low frequencies.    In more recent times, say 10 to 15 years ago  , I checked some ear buds and the impedance was surprisingly flat, what I saw was peaking in the higher frequencies - that's because ear buds are strongly coupled to your ear canal.   In short I wouldn't flinch at 33 to 47 ohm series resistance.

You will always find counter examples.  If you see strong impedance peaks around 100Hz the headphones probably aren't that great anyway and any bass boost from the impedance is probably a help 

The international standard for headphone amplifiers is actually 120 ohm however for low voltage power supplies this tends to be too high, too much loss of swing.

The case for adding the resistor is this:
- prevents very high volumes when you plug in low impedance headphones.
- evens out the level when you plug in different impedance headphones (very handy).
- allows a smaller output cap without losing bass response.
  If you have no resistor and a smallish output cap you will get loss of bass on low impedance headphones.
- for opamp based headphone amplifiers it prevents the opamps cooking with low impedance headphones.

By low impedance I mean 32 ohms or less.