Ive got a smoothingcapacitor(2,2uF) C1 in parallell with relaycoils inside a fullwave diodebrigde.

When circuit is closed trough contacts in pre-relay RLY1 it first gives me, almost everytime because of capacitive reactance in circuit(Ce of C1 and C2), an Arc1 over the contacts of the pre-relay, when circuit is closed. But if I place a resistor R1 infront of capacitor C1 for limiting the inrush charge current of the smoothingcapacitor it will in very short backfires into the resistors R2 & R3 and destroy them. Leading to none zenervoltage.

Placing a diode in parallell with resistor R1 for bypassing of the charge current in one direction will end up in almost the same way with destroyed resistors R2 & R3, but not until after almost 1000 tries.

Why is this happening and how to stabilize the circuit so the smoothingcapacitor isnt giving me an arc in either the pre-relay contacts neither backfire into the resistors. Because of C2 theres around 45-50VDC left inside diodebrigde. Any ideas?

^{simulate this circuit – Schematic created using CircuitLab}

Edit schematic: Added R4, an ICL NTC for inrush current limit

Second edit:

Below answer helped me on the way, thanks! As you pointed out @SpheroPefhany, D6 should be the other way around pointing cathode away(changed in schematics) from smoothing capacitor C1. I've tested some new things to increase the effect over R4(changed to 220Ω) with a TRIAC bypassing(edited as Figure 1) and also found another idea regarding the use of a TRIAC(Figure2 in orange dashed box) or DIAC(in blue dashed box) in serie with the zener D5, without the R4 in the circuit. However, to ensure the heat won't destroy the circuit, it needs to switch fast enough to decrease the heat during opening. Is this a working approch or have I missed something?

See updated schematics.

^{simulate this circuit}