This is a very trivial question but I am lost.
What I am doing: Just for fun, I did a Peltier water heater (the idea was to cool the driving MOSFET with the Peltier cold side, which pumps said heat to the hot side: I keep tge MOSFET cool and get extra efficiency.)
See the rough schematic below.
The basic components are:
- Selectable voltage reference to get various temperatures: Selector position 1: gets a fixed resistance set to 50C. Selector position 2 gets a starting temperature of 40C then a clock divider (working kind of a NE555) closes a BJT changing the final resistance and raising the temperature to 60C.
- A comparator (LM358 1/2) triggers an oscillator at 10-12kHz (duty 50% roughly) created using the LM358 2/2.
- The oscillator acts as PWM which drives the gate of an IRF520 MOSFET that is directly connected to Peltier mains.
- The CD4060 here is used as a NE555 timer (the original idea was to get a multi stage heating, but then I dropped it.) I use the other outputs of the CD4060 to light some LEDs indicating the time elapsed (e.g. 6 minutes=red LED on.)
Now the problem:
The power supply (here a bench power supply) is not infinitely stiff so I get basically the PWM noise to the main voltage, messing up with the CD4060 (the LM358 does not seems affected strangely.)
What solutions are there to fix the induced noise of such setup?
I tried a beefy bypass capacitor, but currents of 2-3 amperes makes it irrelevant. I tried a boost converter to attempt regulation but things got worse.
Final solution would be to have a split supply: a stable 5V for CD4060 and LM358 and a 12V for driving the Peltier.
Any input is appreciated.
Image: Q_C, Q_b and CLK are the CD4060 output. S1 is the selector between the two modes. U1a and U1b are the two op-amps in the LM358. In reality the comparator output "enables" the oscillator, so here I depicted it as an AND gate. The oscillator is at 12kHz, not 25kHz as indicated.