Voltage Regulation With PWM (Low Pass Filter)

Hello,

I need to obtain some voltage levels below 5Volts from 5Volts power rail. I am thinking about to use PWM with 10Khz modulation frequency. For example with %50 duty cycle i am planing to get 2.5Volts.

In order to this i am planing to use the circuit below.



My load has resistive character. Thermoelectric Cooler (TEC).

According to our design TEC will draw 0.8 amps avarage at 2.5 volts.

Questions:

1_According to your experiences will there be any problem that i should be aware of?
2_In order to get linear voltage at the output of PWM driven voltage i added C16 and C17 as low pass filter. Are they necessery, if so are they ok as 10nf and 100nf or should i use 1uf or higher?


Best Regards:thumbsup:

Hi,

R23 may overheat.

The output MOSET will switch fast to 5V. With the output capacitors this causes high current peaks.
Additionally the TEC will draw a lot of current when switched to 5V.

The result is, that with 50% duty cycle you will see a lot more than 0.8A average, because the TEC is no linear resistor.

The TEC cooling power is about proportional to average current (with constant temperature difference). To achieve 0.8A average you need low duty cycle with high current peaks.
The problem is that the power dissipation (unwanted heat generation) is about proportional to I_RMS^2.
Caused by the high current peaks, I_RMS is higher than I_AVR causing low cooling efficiency compared to 0.8A DC current.

You need at least an inductance between FET and load...and a fast diode from GND (anode) to FET drain (cathode).

Klaus

KlausST said:

Hi,

R23 may overheat.

Klaus

Click to expand...

but if i use higher resistance that may cause slow turn on :/

KlausST said:

The output MOSET will switch fast to 5V. With the output capacitors this causes high current peaks.
Additionally the TEC will draw a lot of current when switched to 5V.

The result is, that with 50% duty cycle you will see a lot more than 0.8A average, because the TEC is no linear resistor.
Klaus

Click to expand...

In our design outside temperature will vary that is why i am trying to make a flexible design. Our product will work under 25 degress, ı do not need to keep it at a fixed degree that is why this current ripples may not be a serious problem, i hope

KlausST said:

The TEC cooling power is about proportional to average current (with constant temperature difference). To achieve 0.8A average you need low duty cycle with high current peaks.
The problem is that the power dissipation (unwanted heat generation) is about proportional to I_RMS^2.
Caused by the high current peaks, I_RMS is higher than I_AVR causing low cooling efficiency compared to 0.8A DC current.
Klaus

Click to expand...

Thank you for your warning, we will set the duty cycle on the fly.

KlausST said:

You need at least an inductance between FET and load...and a fast diode from GND (anode) to FET drain (cathode).
Klaus

Click to expand...

What value can be helpfull here? I did not use any inductance before, just some ferrite beads at MCU's power rail that is all that is why i am facing some difficulties here.

Best Regards and Thank You!

A ferrite bead won't do. You need real storage inductor to reduce the current ripple to something like 30 or at least 50 percent. Ends up with an inductance of about 1 mH for the given parameters.