4

I've just bought a PAM8403 pre-built circuit board for $1 on eBay. I was going to use it to drive a set of headphones but I neglected to consider that this is a class D amplifier and will not take kindly to connecting the -ve output terminals together (since headphones are typically 3-wire, unless you have a set of very expensive balanced headphones.)

Is there a cheap solution to connecting the two -ve terminals to drive a pair of headphones without rewiring the headphones themselves? In particular is there any arrangement of inductors and capacitors that will allow the negatives to be commoned?

I am aware that I can use two 1:1 audio isolation transformers but the cheapest I can find these are about $5 each which is not ideal.

I have a feeling that it is possible to connect the two negatives together with some sort of LC filter but I'm thinking that it will probably greatly reduce the usable power output of the PAM8403 chip amp which is already quite low. I would like to hear anyone else's thoughts/ideas.

JRE
  • 67,678
  • 8
  • 104
  • 179
ALM865
  • 143
  • 1
  • 1
  • 4
  • The datasheet says 3W. That isn't enough for you? In my experience, headphones are driven with milliwatts. You might be able to create a common ground after filtering, but since you only paid a dollar, and the PAM8403 is a loudspeaker amp and will probably not sound as good as a linear amp, maybe you should just scrap the whole idea. You could capacitively couple the two outputs, and that would at least provide DC isolation. But you would probably want to use 22uF caps, so that might still over-stress the class D outputs (effectively 10 uF load). – user57037 Jan 18 '15 at 08:23
  • 1
    @mkeith it's 3W for a 4ohm load, typically headphones are up at the 32ohm mark so it will probably only give you about 375mW or so of power. It seems such a shame to scrap the amp, and I don't have any other use for it. There are LM4881 based headphone amps on eBay for just over $10 which are Class AB and would sound better but also cost more than 10x the price. Hmmm I was thinking of using caps but since there is essentially zero impedance between the two negatives (any they could be at the + and - rails respectively) I think it will probably destroy the amp. – ALM865 Jan 18 '15 at 10:56
  • Maybe the isolation transformers are the only way to go? I might hang on to the amps and butcher some old electronics until I find a couple of 1:1 isolation transformers. Some old wall socket powered phones might have isolation transformers that could be used to do the trick... – ALM865 Jan 18 '15 at 11:04

4 Answers4

6

As mkeith says, it may not sound as clean as a linear amplifier, and you have to remember that headphones are a lot more sensitive and can be a lot more critical of low level distortions than loudspeakers.

However if you still want to use it, here's how.

Realise that 3W is waaay more power than you need for headphones - so is 0.375W. I'd go further and say that these power levels can be dangerous. 115dBa through speakers is wall-shakingly, cup-sliding-off-table, stomach-hittingly loud. Loud.

The same level - measured with B&K SPL measuring apparatus) from headphones - without all these other cues - sounds ... well, a little bit loud. And that's probably under 0.1W with many headphones.

In either case, the Health&Safety Executive (UK) recommendation of 90dBa safe exposure over an 8 hour working day, to protect your hearing, translates to listening to 115dBa for ...

nearly 4 minutes.

Without all the other cues, headphones can leave you quite unaware of the danger. The fact they go so loud and stay clean and undistorted doesn't help.

(Still, the rock radio studio techs wanted them LOUDER!)

So it's OK to use this amp below its full power rating.

Now - the PAM8403 datasheet doesn't say what's actually inside it, but it's a reasonably safe bet that it's a full bridge for each channel. As Out_L+ is driven positive, OUT_L- is driven negative, and so on, giving you twice the voltage and 4x the power that a half-bridge will do. (With no input signal, both outputs will sit at half the supply rail ... check this with a meter).

So I would simply ignore OUT_L- and use OUT_L+ as a half-bridge (a normal push-pull amplifier). You will need to AC couple the output via 10uF (maybe 47uF would extend the bass a little) but then you can simply return the common GND from your headphones to GND.

schematic

simulate this circuit – Schematic created using CircuitLab

It can't damage the amplifier and if you have all the bits it costs nothing to try.

  • Yeah I assumed it was a full H-Bridge too but couldn't get my head around connecting the two together, your solution is quite elegant. This is a pretty good idea, I'm keen to give it a go! Thanks for the help – ALM865 Jan 18 '15 at 23:10
  • I have some doubts on this, with a class-d you cannot connect the two grounds and what you actually doing here is connecting the two grounds, the ground of the left and right channel on the amplifier output! I measured 12mV and 1.7mA between these ground leads without any load. I don't think this is a good idea. – Codebeat May 09 '18 at 23:24
  • @Codebeat What "two grounds"? Out_L- and Out_R- aren't grounds - and they aren't connected - they are simply left open circuit here. The "Ground" shown above is the actual ground pin. –  May 10 '18 at 06:47
  • @BrianDrummond : OK, I didn't read the long "story", sorry about the confusion. – Codebeat May 10 '18 at 10:59
  • @ALM865 did you try Brian's suggestion ? It seems to be theoretically clean, elegant - something that should work. I'm just about to go out of my way to procure one of these PAM8304 modules for a quick headphone amp that I need to make. – bdutta74 Dec 26 '18 at 08:36
  • 10uF feeding a 32 ohms headset cuts all frequencies below 500Hz! I would use 220uF capacitors that pass 23Hz and above. – Audioguru Dec 04 '22 at 16:14
3

An alternatative plan would be to place some sort of protection or load on the amplifier. Connect the positive outputs to the left and right headphone and place a four ohm resistor in series with a capacitor (around 200uf) for each channel. The four ohm resistor will limit the current coming out of the negative output and the capacitor will remove any dc component. I haven't tried this but it should work. You may need to experiment with the capacitor value so you have good low frequency response. A 47uf cap may do the job, but I don't know.

schematic

simulate this circuit – Schematic created using CircuitLab

Frank Ward
  • 31
  • 1
  • I made this however replace the 4 ohm resistors with 100 ohm resistors. I discover some heat development and to reduce the power of the output a little more. It is working fine however it requires some precautions. You need a very clean power source and you need to add some pull-down resistors (I used 20K) on the input. Still, it is not ideal, sound quality is decent however, depends on how the source is connected (can introduce noise) and the output can still sounds too hot at some levels. It works however the PAM8403 (board) is not a great candidate to use for this, so better don't try it – Codebeat May 09 '18 at 23:51
1

people. I used the PAM8403 with phones using this circuit:

Phones amplifier

Regards Sergio

user266361
  • 11
  • 1
  • Hello Sergio -- Thank you for this answer. What is the total mW output? Or the proper calculation? – Nicholas Feb 13 '22 at 17:11
  • The PAM8403 is rated "Not for new designs" and produces 3W into 4 ohms per channel with severe clipping distortion when powered from 5VDC. It produces 1.2W into 8 ohms per channel at low distortion which is 9.6Vp-p. With the 22 ohms in series with the 32 ohms headphone making a total of 54 ohms, the power in each headphone at low distortion is 106mW for peak loud sounds but the average continuous power is only 22mW. Nobody listens to continuous peak loud sounds, only the momentary peak levels are at 106mW. – Audioguru Dec 04 '22 at 16:46
0

I did connect the negative from the power supply to the headphone output. No problems and the sound is good.

Nice to see that there are many more solutions.

Voltage Spike
  • 75,799
  • 36
  • 80
  • 208
paulcalmato
  • 1
  • 1