Maintaining supply voltage on load

I have designed a circuit to keep the heating element at a constant temperature. The digital part of the circuit is being feed from a 5V LDO which is powered by two 18650 batteries in series and the heating element is directly connected to the output batteries. Also heating element wire(approximately 2ohm) is controlled with an N channel low side mosfet by the microcontroller.

What i noticed is that the LED and the seven segment display(connected to 5V output of LDO) seems to flicker when controller toggles the mosfet because of the voltage drop on the supply. Is there anything i can do to isolate these rails so that current draw for the load does not affect the 5V components?

The highest supply voltage for the system is 8.4V and currently the load roughly uses 2A. The problem is becomes more significant when the supply voltage gets lower. I don't think that any inrush current situation occurs because voltage continues to stay below 5V at the output of LDO when the load is on after few seconds. Also i tried with both batteries(which i am sure that can supply a lot more than 2A) and a lab bench power supply with no luck. I read that bulk capacitors can fix instant voltage drops but can they prevent constant voltage drop? If so where should i place these bulk capacitors?

Here is my schematic and pcb:

Are your cables thick enough to carry the current without dropping too much voltage?

Sorry to say you have a miserable problem but I know you can solve it. From what I can see by looking at your board layout and schematic I believe most of the problem is in the MOSFET portion of the circuit. This type of current needs large traces which I saw on the drain but nothing close on the source connection. Remember the source handles all of the load current plus the gate current, this can lead to some peculiar results when switching. Be sure to use at least 18 AWG or larger wire for the battery connections. If using a common battery connect the MOSFET ground and the micro grounds at the battery, this will help eliminate current feed back into the micro power supply.

Measure the voltage drop from the Battery to the source and drain of the MOSFET and load, I think you will find your problem in that area. It appears you are not fully enhancing the MOSFET with your gate drive. I would change the 10K to 20 Ohm then on your redesign consider adding a gate driver, you have a relative high gate charge to dissipate before the device is fully on and no current to do it. The gate can draw well over an amp if it is turned on fast. The slow turn on and slow turn off is probably causing the MOSFET to heat. Your JA is 107oC per watt. The hotter the FET the higher its resistance then the hotter it gets, eventually if out of control thermal runaway but not in your case.

If the FET temperature is rising more then about 15C you found your problem. The voltage drop from the battery ground to the load should be less then 0.1 Volt with the fet on. You will need a scope to measure the voltage and a current probe for the current while cycling the output.

Another thing to check is the source to gate voltage when turning on, you need a minimum of 3.5 Volts even during switching turn on.

Good Luck and have fun!

There must be a ground plane but the PCB view isn’t showing it. Make it display and try to follow that path from Q1 source (pin1) around to the power supply ground. If that has any pinch points then those points will have a high voltage drop. If any of your sensitive components are on the other side of that pinch then they will see a voltage drop because their ground is coming up (instead of 5V coming down.)

Try to eliminate pinch points in the ground plane. Then try to imagine where the current flows in the ground plane and keep sensitive components out of that region. At low frequency, the ground current will follow the shortest path. At high frequency (eg 20kHz PWM) the ground currents will try to follow the matching power wire.

Make C1 bigger and “wider”. Maybe 100uF bulk capacitance and 1uF×5 spread around the board with “decoupling” capacitors adjacent to any fast-switching digital components.