Can Cmoy damage earphones?

That setup with two batteries and a centre tap will allways give a loud click, I would build something better with a proper virtual ground circuit.

Power on thump is present in a vast number of amps, some combat it with a delay circuit such as AMB's e12. If it were me I would simply wait a couple seconds for the caps to charge/off set to settle before plugging in my phones.

This is covered in Basic Troubleshooting for Headphone Amplifiers

Yes, that's fine. What you have is just turn-on thump where the DC offset is very high for a split second when the amp turns on. tangent's excellent article covers that:

"If DC offset isn’t the problem, it’s probably just a circuit turn-on transient. If you have an oscilloscope, measure the transient. If it’s less than a volt or two, it’s not worth worrying about."

In your case I wouldn't worry about investigating further - it's just the nature of the design of the amp you've built. It shouldn't damage the headphones but it can't hurt to turn on the amp and then wait a couple seconds before plugging in the headphones.

Witch one ebay kit you use? If it is from Thailand with ground, maid by two battery middle connection, it is dangerous, and you have risk to burn your phones!
However your measured DC offset (4.2mV,2.2mV) are perfect now, this may change during battery discharging, or if you connect different battery, used, with different voltages.
For safety, you need to build standard CMoy railsplitter with two 4.7K resistors, or better with TLE2426

It should be ok to use a single 9V with the TLE splitter, depending on what kind of opamp it's using (if the opamp can use a +/-4.5V supply instead of the +/-9V it's using now). You can read more about virtual grounds here.

Zigis is talking about the fact that the virtual ground point in that amp is set by connecting the two batteries positive and negative terminals. Thus, as the batteries age, the probability that both of them age at the same rate and thus give out the same voltage is very low.

So after a few hours of use, you'll find that one battery is putting out +8.7 volts while the other one is putting out 8.0 volts. The result is that you have now 0.7 volts of offset, and that is dangerous for your 'phones.

For similar reasons, Chu Moy suggested the use of an output electrolytic cap when using resistor based virtual ground circuit and low impedance headphones with his amp. The low impedance load unbalances the virtual ground, causing the DC offset to move into dangerous levels.

Adding a TLE2426 doesn't seem very hard. The battery switch on/off scheme would have to be rewired too, as the virtual ground would now be the output of the TLE and not the batteries' junction point.

cheers!

Not with a CMoy pocket amp it isn't.

We're talking about batteries here, which are floating power supplies. The stock CMoy design doesn't depend on ground being any particularly specific voltage. As long as the virtual ground is approximately centered between the rails, it works okay. This is why the imbalance problem discussed at the top of my Virtual Ground Circuits article isn't a fatal flaw in the CMoy.

In the stock CMoy pocket amp, the input and output are referenced to the virtual ground, the midpoint of the resistor divider. The situation is no different with a center-tapped battery. Since the ground-referenced load -- headphones -- is a purely passive component, there is only one thing to say where ground should be, the music source connected to the input. The source drags the CMoy's ground to wherever it wants it, and life is good.

If the supply is imbalanced, the input signal is indeed offset relative to a precise 50/50 supply split, but the signal comes out the amp offset by the same amount. The headphones see this offset point as "ground", so they perceive no net offset other than what the amp itself added, and that is tiny in a properly-functioning CMoy pocket amp.

That's not to say power supply imbalance is not a problem.

When the imbalance is small, you lose headroom between the signal and the lower of the two rails. If it gets bad enough that the amp starts clipping on that side before the low-side battery is fully depleted, you've just wasted some battery life. More efficient use of the battery is sufficient reason all by itself to fix the imbalance. (This is a central topic in my Op-Amp Working Voltage Considerations article.)

Stock CMoys don't work very well on 1x9V for this reason: they've got so little margin on the low side of the supply that they don't get much use out of the battery. Change it to 2x9V in series, and all is well, because a single 9V isn't much use below about 5V. You can tolerate quite a lot of supply imbalance when the minimum useful battery voltage is 10V total.

This situation doesn't get dangerous until one side of the supply drops below about 1V, while the other rail is still high enough to permit normal operation. (e.g. 0.5V on one side and 5V on the other.) This happens with trivial center-tapped 2x9V setups when they are left on long past when they've started clipping so badly they're unlistenable. Or, if they're left on while no music is playing, as might happen if you fall asleep listening to a CD. When you run one rail so far down, the op-amp stops being a linear amplifier, and it starts producing a DC offset on its output. As the supply on one side gets closer to 0V, this gets worse. I don't know the exact mechanism, but it's no doubt some problem with pushing the transistors on one side of the circuit too close to the rails.

You can play with this phenomenon by swapping the rail splitting resistors in a CMoy for a 1-5K linear pot. Anywhere around the midpoint, the amp works fine. Turn the knob farther, and the amp starts clipping, but DC offset remains low. Turn it even farther, and at some point DC offset does start rising, but this won't happen until pretty near the full turn of the knob.

Incidentally, this problem is what scuttled the successor to the PINT amp. I threw a lot of time at trying to design protection circuitry that would let me use center-tapped battery packs instead of active virtual grounds, but couldn't make it small or efficient enough to be suitable. I might return to it, but it almost certainly won't use a center-tapped battery.

Thank you very much Tangent ! That was certainly a complete and thorough explanation.