...and allows current to flow through the galvanically isolated relay circuit.
The ATtiny circuit is directly connected to the relay circuit through the power supply, so no galvanic isolation there.You probably need some decoupling on the ATtiny power supply pins to stop this weirdness.Try 100uF+100n ceramic, close to the NRF/ATtiny pins.Leo..
Better remove that useless opto, and replace it with a (270ohm) base current limiting resistor (pin to base).The diode is indeed drawn backwards.Leo..
You CANNOT use resistor for regulating voltage of a digital circuit. If I understand it correctly you are trying to do so.
I've done numerous projects with nRF24L01s powered with the 3.3V pin on a Nano or UNO as well as external 3.3V power and never had the problem you describe. What distance is between the radios and do you have the power level set to MAX while working on your project?
Do I see 5.17k as base resistor? I said 270ohm (for a good reason).When a transistor is used as a switch, base current needs to be 5-10% of the collector current for full saturation.A 3volt sugarcube relay draws 100-120mA. So ~10mA base current is needed (Hfe does not apply for saturation).~3.3volt(suply) - 0.7volt(BE junction) = 2.6volt across the resistor, so 270ohm is close enough for 10mA.Remove R1, and also add some ceramic capacitance across the supply of the NRF (I said 100uF+100n).Leo..
Your circuit should work well from 3V3. Some PS supply higher voltage than rated when load is too light. But I don't think this explains your problem. If you need to reduce the voltage you should use a LDO. Or at least resistor + Zener diode. But resistor alone is not able to provide voltage regulation for digital circuits and only the fact you do not need voltage regulation saved your circuit from damage.