DC spikes?

30 mV is fine

If there is a hard and fast rule on what isn't fine I am not aware of it. It depends a lot on the sensitivity and type of your headphones. I think once you start getting to the mid-100s of mV that is pretty universally bad. Sensitive IEMs you will want lower, and robust technologies like planar magnetics might be able to handle more. 30 mV should be fine for just about anything though.

Quote:

Originally Posted by G.Trenchev 

OK thanks a lot
 

The Turn-ON is usually a lot worse.  That's why there are relay-delays built into a lot of headphone amps.  Many OTL and hybrid tube amps are biased with the signal's zero voltage reference actually a significant offset from true ground.  For instance, on a basic Millett Hybrid that could be as much as 12VDC direct to the headphone.  Output capacitors are used to block that DC.  However, large capacitors take time to charge, sometimes on the order of a half-minute or more.  It's during that period when headphones might be in danger if they are plugged in at the same time.  The relay is set to delay the signal circuit from kicking in to the headphones until the capacitors are charged.

Turn-Off spike is usually a symptom of the onboard capacitance.  If there's a delay-relay (the delay is at turn-ON), then it depends on how quickly the relay can open the circuit.  Just as an example, we switched to a 24VDC relay in the MOSFET-MAX and found that it opens faster on turn-off and the spikes are much less than with 12VDC relays.  (It also closes with much more authority in the Turn-ON mode.)  YMMV, of course, depending on the equipment. 

If you see DC transient spikes in the amounts of whole volts - 3, 5, 7, etc., then it's a concern for your headphones.  That assumes that the transients are quick, of course.  The slower the transient, the lower the voltage necessary to cause concern. BCG27 is probably more prudent to suggest voltages in the 100's of milivolts, but some relays will let more than that through with no ill effect.  It depends on how fast your DMM is, too.  Some DMM's may not register anything, whereas others may show whole volts.

With your circuit, the power supply has already created a + and - supply, so there is no bias to worry about at turn-on.  As stated previously, in such an instance the onboard capacitance will pretty much dictate the turn-off spike.  Those 4700uf capacitors you have in the power supply are the culprit.  They're not going to drain down very quickly, so voltage/current will be available to the output for some period after you turn the switch off.  Hence, a voltage spike and current pulse.  It's not surprising that it lasts for up to a second.

Still, even 30mV at 1 second is probably not enough to worry about.