Is my circuit correct?

[Ian.M]

The ULN2003A doesn't need a decoupling cap, as it doesn't even have a Vcc pin!

I'd recommend 1nF to ground at the MCU's switch sense input pin which will form a low pass filter with a corner frequency of 16KHz, filtering out most EMI pickup.  If the pin has a Schmitt trigger input characteristic, you can increase this capacitor for a suitable hardware debouncing time constant as per J Ganssle's article, and make your software simpler.  If not, don't, as increasing the capacitor may cause problems with the digital input not transitioning fast enough.

[redgear]

The ULN2003A doesn't need a decoupling cap, as it doesn't even have a Vcc pin!

I'd recommend 1nF to ground at the MCU's switch sense input pin which will form a low pass filter with a corner frequency of 16KHz, filtering out most EMI pickup.  If the pin has a Schmitt trigger input characteristic, you can increase this capacitor for a suitable hardware debouncing time constant as per J Ganssle's article, and make your software simpler.  If not, don't, as increasing the capacitor may cause problems with the digital input not transitioning fast enough.

Thank you:)

One more question, the relay's datasheet recommends a Zener diode + Diode to supress surges between relay pin 1 and the 5v. Do I need it?

[Ian.M]

No, the diode(s) are built into the ULN2003.  Simply return the ULN2003 common cathode pin to +5V via a resistor equal to the relay coil resistance if you aren't going to switch off multiple relays at the same instant, or via a 4.7V Zener, cathode to common if you are.

[redgear]

No, the diode(s) are built into the ULN2003.  Simply return the ULN2003 common cathode pin to +5V via a resistor equal to the relay coil resistance if you aren't going to switch off multiple relays at the same instant, or via a 4.7V Zener, cathode to common if you are.

Alright, Thanks