If your 5Y3 is directly heated (filamentary), it'll probably handle around 40uF maximum without any great loss of tube life, as long as it's operated at modest voltage and current, which is the case in your headphone amp. If it's an indirectly heated 5Y3 (cathode sleeve type), it'll take 40uF at much higher current and voltage than your headphone amp requires. In your headphone amp, probably 60uF wouldn't hurt anything with an IDH 5Y3.
5U4s are directly heated, and just don't like a lot of input capacitance (more on why later on). Ruby 5U4s are notorious for arc-over at powerup with only very modest voltage/current, unless the input capacitance is held below say, 20uF at maximum. An American-made NOS 5U4 will withstand more input capacitance, but it's still not the best idea in the world to run over 20uF on it. Originally, 5U4s were only rated for something like 5 to 10uF of input capacitance, depending on the particular variant. In those days, electrolytic capacitors were beyond terrible, with very poor electrical characteristics. Today's electrolytics are much better, and film caps even better yet. Still, 40uF is a good bit of input capacitance for a 5U4 to deal with. You likely already know this, but for those who don't , here's why big input cap values are stressful to tube rectifiers:
When a capacitor is completely discharged, there is a short period during its initial charge cycle (called the transient period
or transient state
) where it has almost no resistance to DC. During this time frame, the rectifier tube's cathode is literally directly shorted to ground. The charging current during the transient period is, not surprisingly, called the transient current
. The transient current can easily exceed the rectifier's maximum forward current rating, until the point in time where the capacitor is charged up enough for its DC resistance to become established. The larger the input capacitor value, the longer the transient period and the greater the transient current necessary to charge the cap up to the point where it no longer passes DC through to ground. This is a chief cause for arc-over on startup with large input cap values.
Directly heated rectifiers come on near instantaneously, which means they are conducting significantly during the transient period. OTOH, indirectly heated rectifiers heat up slower and therefore ramp up their output voltage much more slowly. This allows them to smoothly charge the cap up past its transient state while their output voltage is still rather low (too low to arc over).
Also, with a capacitor input filter section, the peak RMS current the rectifier and power trafo see may exceed the amplifier's actual nominal current draw by a factor of 1.5. This must always be taken into consideration. What happens is that the input cap does just what any capacitor does, which is store DC voltage in an electrical field. When the AC sinewave going into the rectifier peaks at its maximum voltage value, the rectifier's DC output voltage peaks at the same moment along with it, and the input capacitor is charged to that voltage. When the AC input voltage passes its peak value and begins falling back toward zero volts at the center point of the sinewave, the DC voltage output from the rectifier falls along with it. The input capacitor then releases its stored voltage in an effort to keep the rail voltage from dropping.
As long as the rail voltage is maintained by the capacitor's stored energy being released, the rectifier will not conduct. But, the amplifier is constantly consuming current, and therefore the cap is always being discharged and depleted as the rectifier's DC output ramps down towards zero volts. When the cap becomes discharged enough that it can't maintain a voltage equal to or higher than the voltage coming from the rectifier, the rectifier will conduct.
OK, down to the nitty gritty of it all. The fact of the matter is that on every half-cycle, the rectifier must supply 100% of the current necessary to feed the amplifier and charge the input cap back up. With a small capacitor value, the cap is quickly charged and quickly depleted. The rectifier fires over over a relatively large part of the half-cycle. It has more time to supply the necessary current, so it does so in a more smooth and uniform manner.
But as the input cap value is increased, the extra stored energy tends to maintain the voltage high over a larger percentage of the cycle. Therefore, the rectifier can only conduct at the very peaks of the sinewave, in short, violent, "wide open throttle" bursts that supply all the necessary current all at once, in a very short time frame. You can see how although the amplifier's nominal current consumption has not increased, the peak current the rectifier must supply is increased with greater input cap value, because it must do it all at once. This not only places more stress on the rectifier tube, but also on the power transformer (on which it has the same effects).
Unfortunately, the smaller the input cap value, the greater the residual ripple and the poorer the overall voltage stability. But, because the rectifier fires more smoothly and for a larger percentage of the cycle with small value input caps than with larger value input caps, there is also less chance of the ringing oscillations that are caused by a sudden, giant burst of current.
BTW, here's some related technospeak you may see sometime when reading about power supplies, so I thought I'd throw in the definitions. The radian degrees of the cycle through which the rectifier conducts is called the conduction angle
. The radian degrees of the cycle where the rectifier does not conduct is called the dead angle