Too many Y caps means tripped RCD?

[cupoftea]

Hi,
if i plug in loads of mains devices which contain Y caps from live/Neutral to earth, then at some point the RCD will trip?

 

[Easy peasy]

Yes, yes it will.

 

[betwixt]

A rough calculation says the value should be less than 0.8uF, assuming the capacitor and RCD are in series and directly across 240V 50Hz. I would expect it to be a little higher when other filter impedances are also present.

Brian.

 

[KlausST]

Hi,

It surely depends on RCD trip current.

Klaus

 

[cupoftea]

Also, its the y cap between live and earth that gives the most current to earth........the one across the transformer isolation barrier gives much less, and the y cap from neutral to earth gives virtually nothing.
The Y cap across the transformer, especially if sec gnd is earthed, actually gives a noisy current shape..."sharp cornered"...probably no good for emc...what do you say?

 

[cupoftea]

If you run the attached simulation of a rep of an offline SMPS with three 2n2 Y capacitors, then you can see that theres only actually 261uA of earth current from 240VAC input.
The Y cap from neutral to earth contributes nothing, and the Y cap across the transformer contributes less than the one from live to earth....the striking thing is that due to the secondary being earthed, the y cap across the transformer actually has a sharp current waveform, which woudlnt be good for emc.

If common mode inductance is added, then it gats worse, as the cap across the isolation barrier...well, it then puts a ringing current into earth...which is bad for conducted EMC...so i see the cap across the isolation barrier comes with disadvantages.

 

[FvM]

How can we relate post #5 and #6 to the original question? RCD trip current is at least in a several mA range, it can't be achieved by nF Y capacitors.

 

[c_mitra]

How can we relate post #5 and #6 to the original question? RCD trip current is at least in a several mA range, it can't be achieved by nF Y capacitors.

Right; also several mA (I guess 10mA at the peak is closer to reality) should be maintained during at least one 1/2 cycle. In my guess, it is the total energy that matters in the tripping (delivered over 10-20ms).

 

[KlausST]

Hi,

In my guess, it is the total energy that matters in the tripping (delivered over 10-20ms).

If so, then a purely capacitive current means "no energy". Then an RCD never trips with capacitive current.
I doubt this ... although I don't know for sure.

Klaus

 

[Easy peasy]

the surge into the Y caps at turn on is often a factor ....

 

[c_mitra]

then a purely capacitive current means "no energy". Then an RCD never trips with capacitive current.
I doubt this ... although I don't know for sure.

A purely capacitive current in the capacitor really means "no energy" but when you make the same current flow through a resistor or a transformer primary, strange things can happen. This current is 90 deg out of phase with the voltage only across the capacitor, if my understanding is not rusted.

RCD detects the unbalanced current via a transformer and the trip mechanism is a load on this transformer. All mechanical relays cause a change in the magnetic flux during closing on/off period and the magnetic energy is transferred to the spring of the relay contact (mechanical energy).

Do correct me if I am talking nonsense.

 

[KlausST]

Hi,

I think there is a misunderstanding.

My thinking was the "external energy" or the "load side energy".... like an energy meter does.
Your thinking is some kind of RCD internal energy.

To avoid this confusion let´s go back to RCD trip point definition:
Here in Germany the trip point is defined as "current RMS". Like "30mA".
And I assume this trip point does not care about phase shift related to phase voltage. So it´s a pure current related trip point.
And it also does not care about RCD internal resistor´s (or other circuitry) power loss or energy. I think the RCD internal power loss may be different from brand to brand .. even with identical trip current.

Klaus

 

[c_mitra]

To avoid this confusion let´s go back to RCD trip point definition:
Here in Germany the trip point is defined as "current RMS". Like "30mA".

I am trying the say the same thing.

The 30mA trip current must be associated with a time spec; I assume this should be at least a quarter period or more likely a half cycle (else the RMS definition becomes problematic).

I also assume that the frequency should be around 50Hz (domestic frequency) that means that the 30mA current must last at least 5ms or preferably 10ms.

My point is that the trip switch needs some energy to turn on/off. Why?

Consider the gedanken. Assume you have a good FET switch that can be turned on or off with a single pulse.

How much energy the FET needs to turn on or off? That depends on the junction capacitance and the turn on voltage.

You cannot have a FET that have ZERO junction capacitance and also turns on at a very low voltage. And it also depends on the power you are switching on or off.

Consider a bistable switch state. The pulse must deliver some power to the circuit to turn it on or off.

Same idea goes for a mechanical relay.

When the current changes from 0-30mA, the RCD trips. Will the RCD trip if the current pulse is 1ms wide? My answer is no; the pulse must supply some energy to trip the mechanism and that is some finite energy.

How about a circuit in which a small and sensitive relay trips a bigger relay and that in turns turns a still bigger one?

 

[KlausST]

Hi,

I agree that there needs to be a trip time specified.

I´ve done an internet search. In one document I found it says:
200 ms with 1x I_deltaN
40ms with 5x I_deltaN
Don´t know wheter this is a common specification.

the RCD is manually switched on ... the energy to switch OFF is stored in a spring.
And the trip current just needs to release the spring energy.
(Very small energy to release ... compared to the spring energy)

Klaus

 

[c_mitra]

the RCD is manually switched on ... the energy to switch OFF is stored in a spring.
And the trip current just needs to release the spring energy.
(Very small energy to release ... compared to the spring energy)

True; but.

When you turn on the switch, you need to push the spring against a peak. The switch spring settles in the on position which is a potential energy valley on the other side. The on position of the spring is a potential energy valley which is much higher in energy compared to the off position. To turn off the switch, you need to push the lever across the peak (which is small but must be much greater than the noise threshold) and after this level is crossed, the manual lever disengages and the mechanism turns off. This is to prevent (rather reduce arcing) the contacts from getting damaged.

But this is very clear that the few uAs of current flowing through Y-capacitors for less than a millisecond cannot trip a RCD even if 1000 such devices are turned on exactly at the same time.

Gas centrifuges used for isotope separations are run by individual power supplies and they must not be turned on at the same time and there are 10,000 (or more) of them running in series with in one place. And they run hours and days and months continuously at a time. And they can be replaced individually without stopping the whole setup.

 

[Easy peasy]

a repair place next door has no end of issues with RCD tripping due to > 10nF Y caps on 3 ph chargers, 10mA RCD, always trips at turn on ....

chop out the Y caps - no issues ....

 

[c_mitra]

A rough calculation says the value should be less than 0.8uF,

Surely you mean more than 0.8uF? My estimate is in the same range, it must be greater than 0.47uF across the line and earth (net unbalanced)

 

[betwixt]

sorry, misunderstanding, I meant if the the value is more than 0.8uF there is a risk of a standard 30mA RCD tripping. As most 'Y' capacitors are only a few nF it would take lots of equipment containing them to cause a false tripping condition. By the time enough capacitor current had operated the RCD, the chances are the overload breaker would already have operated.

Brian.