[morzh]

MrFixr

Well, to know exact requirement one has to buy the standards (national standards, to me, must be available for free, but no....those are agencies, and agencies want money) and read them.

For the US those are UL 60384-14, UL 1283 and UL 1414; there are also EU EN 60384-14 and EN 132400; these specify "across the line" applications and touch on the requirements to the caps.

The most rigorous ratings are Y1 and X1.

X-cap is supposed to filter differential noise, and Y-cap filters common mode noise.
Example: the noise on the electric line has differential (current) component and this will be short-circuited from L to N through the X-cap; this cap has to withstand disturbances (that is, a finite number of them, after which it is allowed to fail), which are inherently present in any AC line, industrial or residential. Of course, various things like Transorbs, MOVs, fuses, gas tubes, heavy inductances etc are utilised to attenuate the disturbances before the X-cap and the rest of the circuitry would see them.
The thinking is, well, OK, if the cap fails open, then it filters no more, so, so be it. If it fails short, this is what fuses are for. (it is incorrect that X-cap is required to fail short; it is not).

Y-cap filters common mode (voltage) noise. This noise, usually a higher frequency one, makes things radiate. An example is a switcher: the AC voltage developed across the flyback primary will, through capacitive coupling, induce the common mode noise on the secondary winding. This means, that the noise is present, more or less as the same voltage (hence "common mode"), at any point of the secondary, as it simply acts as the second plate of the capacitor. The Y-cap (in this app it is called "bridge cap") will provide the short circuit between the primary and the secondary (or rather the secondary and the GND), limiting the current flow to a small loop, rather than letting the whole secondary and the rest of the circuitry after it radiate as an antenna.
Across the line, if the common mode noise is present on both L and N, obviously X-cap will not do much; this is when Y-caps are installed from L to GND and from N to GND.
In both cases the Y-cap connects to GND and therefore failure as shorted may lead to electrical shock.

The value of the Y-cap in universal switchers is rarely more than 1.5nF. This is to limit the touch current in case the GND connection from the case to the GND fails, unbeknownst to the user, and the said most unfortunate user were to complete the circuit between the case and the GND. TYpically required max is not to exceed 0.5mA for plugged in equipment.
It is allowed to be 3.5mA for the permanently wired one, with the label warning of the high leakage current.

Example: a 1nF capacitor from L to GND will create 0.045mA of leakage current. However 10nF (used in our radios) would create 0.45mA, and two of them will create 0.9mA, which exceeds the maximum requirements.
A typical switcher with flyback topology uses 3 caps: L/N to GND plus the bridge, so due to three caps their summary value could be uqite high, so it becomes a compromise to find the values that would work for EMI suppression while passing the safety leakage current test.
This is where additional things like common mode chokes etc are popular additional suppression measures.


This is not a treatise, this is an explanation; the only document that can explain every small detail is the agency requirement, as this is what drives the design and the safety marking of the Y/X caps.


PS. As for the Phorums "expert" rating, I personally am no expert at anything, l just know some things at a better than average level, plus I can repair anything electronic, short of the parts that require programming or pick-n-place line for repair; when it comes to old radio equipment, folks like Terry and some others know about 1000 times more than I do.

PPS. Solen (and film in general) became popular because they do not have limited lifespan, as the electrolytic caps do, and they in fact do have a decent ripple current value.
What ripple current does, it heats up the capacitor due to the AC component (charge-discharge) with Joule heat, with the charge current heating the ESR (equivalent series resistance) of the capacitor. Electrolytic caps (the GP ones) are known for high ESR at smaller capacitance values. Higher value caps (hundreds and thousands of uF) have fairly low ESR, but we cannot use them as the first rectifier filter cap, as the tubes have limits of what capacitance they can handle. Old capacitors, however, had huge dimensions and huge mass, and this kep them from heating too badly; todays caps of the similar values are small, and the same amount of heat will make them quite warm. And in case of restuffing, this small cap gets to be sealed inside the older case with no air circulation whatsoever, which does not help either.
So, modern technology came with so called Low ESR caps and High Ripple Current caps. I am still not sure what the difference is, but basically the idea is, a cap with high ripple current rating wil withstand much higher ripple current. Which in turn means that it is either somehow resistant to that heat way, way better than a GP cap, or has low ESR, which makes the heat way, way less at the same ripple current. In either case the cap runs cooler.

Now, the formula that gives us the life of the capacitor involves WV and temperature. And while WV (that is the ratio of the app voltage to the rated WV) is at the base of the power expression, the temperature (that is the difference between the rated Temp and the ambient plus the temp rise) is at the exponent.
That is, your electrolytic cap could outlive multiple generations of people, if it is employed cool and under the rated voltage.
And, to this day, nothing really beats the electrolytic cap when it comes to achieving high capacitance and high working voltage.