3.3V (700mA) from a 24V DC source

[ppTRN]

Hi everyone,
I need to design a board with an ESP32, powered from a 24V source. I am looking for a swiching regulator but there are so many that I am worried that i might take the wrong one. First things first, I guess that the best way to get 3.3V from a 24V supply is a switching regulator, not an LDO. Secondly, won't the power loss be huge with such a gap between input and output voltages? Did anyone faced such problem, and found a suitable solution?

Thank you.

PS: till now the best option seems to be a regulator based on the MC34063 or the LM2596

[wasedadoc]

1.  LDO means Low Drop Out.  Only necessary when the input voltage is close to the output voltage.  For example 5v in and 3.3v out.  24 in and 3.3 out does not fall into that category.  Not saying you cannot use an LDO, just that you would not need to discount a regulator which was not a LDO type.

2.  When dropping significant voltage in a linear regulator the dissipation needs careful attention. Every volt dropped with every milliamp of current means 1 mW of heat.  So heatsinking may be needed and be an additional reason with efficiency for opting for a switching regulator. For some moderate currents you could reduce the heat in a linear regulator by using a series resistor to drop some of the voltage.

700 mA times 20.7 Volts is about 15 Watts so you should certainly be thinking of a switching regulator.  A side benefit is that the 24 Volt supply will only need to provide about 120 mA.

[mariush]

While MC34063 and/or LM25x6 would work, they're kind of bad choices because they switch at low frequencies, which means they require big inductors and capacitors to output a smooth voltage, and that means higher costs and more pcb space used.
MC34063 is also fairly basic, with minimal protections, but it's indeed very cheap.

I'd suggest looking for something that can work at least 200kHz or more (but not too much, as the higher the frequency the more important it is to do proper layout on the pcb and the higher the frequency usually the efficiency goes down a little)

A bonus would be synchronous rectifier step-down regulators, which allow you to skip on a diode making circuit cheaper and smaller.

For example, some suggestions :

AP63201 : https://www.digikey.com/en/products/detail/diodes-incorporated/AP63201WU-7/9858425
See figure 10 in datasheet (page 9)  and the recommended values chart on page 13 and the example layout on page 15.

AP63203  : https://www.digikey.com/en/products/detail/diodes-incorporated/AP63203WU-7/9858426
(fixed 3.3v output version of the above, so no resistors to set output voltage required)

AOZ1282 : https://www.digikey.com/en/products/detail/alpha-omega-semiconductor-inc/AOZ1282CI/3973644

If you want through hole (but NOT breadboard friendly, you must solder to some prototyping board or pcb), well, not much besides the classic MC3x063 BUT you could go with packages like TO-252 that are very easy to solder , and even bend the leds to make it through hole.

For example see BD9870FPS  (900 kHz) : https://www.digikey.com/en/products/detail/rohm-semiconductor/BD9870FPS-E2/1816644
or BD9870FP (300kHz)  : https://www.digikey.com/en/products/detail/rohm-semiconductor/BD9703FP-E2/1957242

[ppTRN]

1.  LDO means Low Drop Out.

Of course, my bad, I ment linear regulator.

 they require big inductors and capacitors to output a smooth voltage

Space on the board is actually not a problem. Thank you for your suggestion, i will go through them!

[langwadt]

for a few prototypes or one-ofs it doesn't get much cheaper and simpler than something like this; https://www.ebay.com/itm/224151690136

[dobsonr741]

Look at this TI part - no external inductor needed, has a built in one! I’m using one and totally happy how simple to design it in. https://www.ti.com/lit/ds/symlink/lmz14202.pdf

[Benta]

Yep, that's a nice part.

[langwadt]

Look at this TI part - no external inductor needed, has a built in one! I’m using one and totally happy how simple to design it in. https://www.ti.com/lit/ds/symlink/lmz14202.pdf

it is also expensive and out of stock

[dobsonr741]

My next runner-up is from XP power: https://www.digikey.com/en/products/detail/xp-power/VR05S3V3/13147710  Tried and proven as well.
Or, if 0.7A is a hard limit: https://www.digikey.com/en/products/detail/xp-power/SVR10S3V3/13147726

[AnalogTodd]

1.  LDO means Low Drop Out.

Of course, my bad, I ment linear regulator.

 they require big inductors and capacitors to output a smooth voltage

Space on the board is actually not a problem. Thank you for your suggestion, i will go through them!

Linear regulators are basically controlling a transistor to give a constant voltage at Vout as long as Vin is higher than Vout plus the regulator dropout. They dissipate huge amounts of power based on the equation Pdiss = (Vin - Vout) * Iout. As already pointed out, 24V to 3.3V at 700mA is 15W of power dissipated, basically just becomes wasted heat.

Switching regulators operate by moving electricity through from input to output only part of the time. They will turn on a switch to build energy in an inductor, then turn off that switch and the inductor then sends that energy to the output through another switch. For a buck regulator (switch from high voltage to low) going from 24V to 3.3V, the switch turns on less than 15% of the time. The capacitor on the output smooths the output voltage to keep it close to where you want. I know many purists are going to point out that there are a lot of equations and details that I'm not covering, but this is meant to be a basic concept.

Because the switching regulator only pulls power from the input part of the time, wasted power is minimized. For a buck regulator, 3.3V output at 700mA is a little over 2.3W of power delivered, and the input is going to be 24V delivering 700mA, but only 15% of the time. So while it can be taking 16.8W of power at a given instant when the switch is on, the average power needed is only 15% of this, a little over 2.5W. Instead of 15W wasted, it's only 0.2W.