Why does a spark occur when an SMPS is connected to the load circuit?

I am using IRF3205 to switching GND, and the gate is controlled by the microcontroller for the delay, so when the connector attaches to the load circuit, the GND is already disconnected, so the spark not happens.

Solution: - When the power In signal detected by the microcontroller, it turns on the gate after a few seconds to avoid spark. (Note:- Here I can use a relay for the same purpose, but again relay is the one kind of physical switch, so spark happens inside a relay is again cause a problem after a long time so that kind of solutions are not good, so I preferred this type of solution)

Thank you all for your help.

My suggestions,

1. MOSFET based inrush current limiter: https://github.com/msglazer/Anti-Spark_Switch (Generally used in octocopter)

2. Anti-Spark XT60, XT90 connectors explanation: https://youtu.be/X71Suakve6A (Video) In Anti-spark connectors, they are using small value high power resistors approx 6 ohms. This Resistor helps in reduce inrush current due to higher capacitance in your circuit. But this resistor only contacts in a small period of time. If you connect this loose connector maybe meltdown.

Thank you for reading.

You can get a spark hooking up 12V jumper cables between cars because you are making and breaking a circuit in the many-millisecond time it takes to connect. The inductance in the cables or receiving circuit is surprising in its ability to keep current flowing in a collapsing magnetic field when a circuit is broken. It's nothing to worry about. If you want to reduce it, put a switch in the line and throw the switch after you connect. There will be little or no "bounce" that way.

Your design has excessive inrush current because it has no protection for a bulk capacitive load. Thus the surge current is I=V/ESR for ESR includes cap and contact resistance.

You could consider a PTC rated for the max Joules of charge energy but that adds a minor cost and heat.

Even 2.4A USB power to Apple iPad products creates a black carbon strip in the middle for +5V and burns out one connector pad. With an extremely thin flash coating of gold, the plugs eventually wear out in a year or 2 from a daily connection.

An electric spark is an abrupt electrical discharge that occurs when a sufficiently high electric field creates an ionized, electrically conductive channel through a normally-insulating medium, often air or other gases or gas mixtures. - Courtesy Wikipedia.

A spark could be caused by breaking current flow through an inductor (resulting in a high back emf generated by a collapsing magnetic field to return its stored energy to the source). It could also be caused by breaking a high charge/discharge capacitor current. Other causes could be high surge currents through incandescent lamps etc.

Identical scenarios would be possible, when a contact is made using a switch or a relay, due to 'contact bounce' or 'chatter'.

The solution would be to use RC snubbers or VDR's across the contact or across the load.

This is a very common occurrence if your load contains a large capacitor bank. The problem is that if the voltage over a capacitor changes fast (as when you connect the load to the power supply), this results in very large currents. This can be seen by the following differential equation for the capacitor current:

$$I_c(t) = C frac{dV(t)}{dt}$$

Unfortunately this is exactly the case when you connect your load to the power supply. In this case your capacitors will draw a lot of current (which is often called the inrush current)

You can reduce this by limiting the current with a large enough resistor. This approach is the simplest, but adds additional losses through the resistor and increases the input impedance. A better Option would be to use an NTC Thermistor in that case. These resistors have a negative temperature coefficient, which means that there resistance decreases when they heat up. Therefore, the resistance of the NTC Thermistor becomes very low after it is warmed up by the current passing through it. This achieves a power loss that is lower than when a fixed resistance is used. You can find some information about NTC Thermistors here: How to Use NTC Thermistors for Inrush Current Limiting

Alternatively you could add an electric switch (like a MOSFET for example), which ramps up the voltage and therefore eliminates the spark. An in depth discussion about that topic as well as schematics for how to connect the MOSFET can be found here: StackExchange Electrical Engineering: P-Channel MOSFET Inrush Current Limiting

Furthermore also the geometry of the connector can influence the stark. There are dedicated anti-spark connectors which help reduce it or ensure that the spark can not come in contact with the skin.