Driving Headphones from a Speaker Output

That is similar to what I did but in this image both speaker grounds are shorted together. My amplifier does not have a common ground on its outputs which resulted in it being destroyed by doing this.

Your amplifier is trashed, for obvious reasons.Get another one. Maybe two, just in case.
Not an endorsement of Deal Extreme, just an example.

BTW, I know nothing about electronics.

As Oddball said " I only ride 'em, I don't know what makes 'em work."

My guess is that your amplifier channel output is push/pull and not single ended, so that neither lead is ground. When one wire goes positive, the other goes negative and vice-versa. You could connect an earphone channel to one of the leads of the amplifier and the earphone common to the amplifier ground. You will probably want to make sure you use the same polarity for both channels or your stereo hearing will not be right. You may want to put a resistors in series between the amplifier outputs and earphones so you don't overdrive you earphones.

It's too late.

...............................Amplifier.

In addition to that you need to add a series capacitor because on the amplifier output you have half of the amplifier's supply voltage. The capacitor value depends on the earphone + series resistor total impedance and the lowest frequency that has to pass. You can experiment with values between 100-1000μF (connect the capacitor + lead to the amplifier output and the - lead to the earphone) - the capacitor voltage rating should be at least the amplifier's supply voltage.

I tried your idea of connecting the earphone ground to supply ground. With the volume all the way down and no input signal to the amplifier, the earphones burned out. Maybe I should have added a resistor to ground.

You might have needed a capacitor as well as a resistor, the mid point is not necessarily at 0V.

The only method that will work, is the one I mentioned.....you must treat both channels as completely separate for safety reasons on this particular chip.....

or

Buying a better (simpler, cheaper) chip!

A class D amp is not a linear amplifier. It uses Pulse Width Modulation, so tying the left and right channels together (as you now know) is a quick way to burn up the silicon.

Sorry. Try reading some of this: https://www.google.com/search?sourceid=navclient&aq=&oq=class+d+am&ie=UTF-8&rlz=1T4GGHP_enUS517US517&q=class+d+amplifier&gs_l=hp..0.0l5.0.0.0.6846...........0.efwV9s1Z_CQ

The amplifier class is not relevant to the problem. The amplifier was almost certainly a bridge output where the -ve output is an inverted output rather than ground. Shorting the any output of a bridged amp to ground or another amp will cause problems.

Even in a single ended output design shorting the output grounds together can cause problems such as distortion or instability because of high circulating currents.

As y0u now know, and several people here have confirmed this, you cannot connect the "ground" signals, as there are none...

You seem to have managed to burn out a short circuit protected chip! Quite an achievement!!

That throws out all 3 wire connections completely (sorry Lyn!) as you now know.

If you still want to attach headphones, you need to get to the 4 wires (2 to each headphone "ear") point of the headphone wiring and attach two "Cinch" connectors. (if there is no "proper" connector for 4 wire phones!) You may even need to completely rewire the headphones.....

The opposing Cinch connectors can be individually connected to each amp output, once you have dropped the signal levels with a corresponding matching circuit, otherwise you will simply damage the headphones.

There are to my (minimum) knowledge, at least several "standards" for headphones, I do not even pretend to know them all. You can have in my limited understanding 8 Ohms or even 36 Ohms (very common nowadays).

You still haven't told us what yours are!!!!!

So the matching circuit, needs to match the Ohms of your headphones AND keep the maximum power within the capabilities of the Headphones as well. Usually 3 or 4 resistors in a small network will achieve it....(but I am prepared to be over ruled on that point!)

Using an 80 - 100 Watt stereo amp for a pair of headphones (1 Watt? max maybe. Just guessing!), sounds bad news.

Reading here should help you calculate the resistors to match headphones to amp:-

headphone-amp-impedance-questions

Why don't you look for a chip with common ground outputs and far less output power....maybe one that also gives details of impedance and power matching for speakers/headphones?

That way you can generally use standard headphones.....

I was thinking that wiring each earphone to each output would be the best option. I have two very nice audio transformers which could step the power right down.

Transformers would be ideal since the Left and right returns on the secondary windings can be commoned so a normal 3 pole socket can be used. - Just make sure it is an isolating transformer, not an auto-transformer and make sure it is rated for the maximum output voltage of the amplifier.

That's exactly what I tried a while ago but the left and right channels were not properly isolated. When I changed the balance from left to right I could still hear the music in both earphones.

The resistance requirements for many headphones is much greater than the normal 4 or 8 ohm of home speakers.

Oh well try again.

It is as others have said. Each channel is balanced and isolated. Never mind.

I can't picture what you grounded and where as this is not a simple chip and there are so many other ways you could have released the smoke from the package with other boo boos.

Heres the data sheet for it. Page 11 is a good place to meditate.

Whatever you did that released the smoke was probably a good thing for your headphones. Silver lining. 100W needs padding before being presented to earphones.

Earphones are very low voltage & current devices with very low impedance which basically presented a short circuit load on the amplifier.

The headphones need to get their signal from the pre-amp section and not from the final output stage of the amplifier.

You could measure the output voltage & current then buy or build a small transformer of the proper ratio to meet the headphones requirements.

Or go online to Radio Shack and order one already built for a small sum that is much cheaper than replacing the amplifier.

I found another good link for anyone trying to do whatyou want here:-

headphone-impedance-explained

Best of luck.

The chip you used is push-pull but it is also "bridged". This means that both speaker leads are "hot". Connecting the negative speaker leads together will result in letting the smoke out. Transformers will work but you will need 2 of them, you must have complete isolation between channels. The easier way is to make a simple resistor network X 2, one for each channel. Yes, it does matter what the impedance of your headphones are. They vary from 8 ohms to 2000 ohms. A standard reduction network uses 220, 1 watt and 47, 1/2 watt resistors. Connect one 220 resistor is series with each speaker terminal. 4 resistors. Then connect the 47 resistors across the free ends of the 220 resistors, this is the end that is connected to the earphones, in parallel with the 47 resistors. Yes, the headphone wire will have to be cut and each pair of wires connected to each channel output separately. Whether you use p38arpxmllugs and jacks is up to you.

Many thanks from myself for the good information and "support" with the proper concept.

I don't think he believed me alone before (Post #9), perhaps now!!

I personally feel that a less powerful, but simpler to connect modern (headphone?) amp chip could be installed instead/in switched parallel, (with a simple switching between the big chip and one made for driving headphones maybe)...

One where one side of each channel is ground and that can be simply connected together, allowing "normal" headphones to be used.....instead of needing two separate connections.....though probably quality of sound may be even better in such a configuration with the original "BIG" chip, but who would actually notice the difference? No one!

It can't. You've proved it. It's knackered!