Recommendations - Considerations for Power Supply design

Hi all,

I am currently using the following setup for power supply of a stand-alone arduino and other peripheral devices:

Requirements :
Input:
Upto 30V DC / 24V AC

Output:
12V DC at about 200mA (for relay coils) and
5V DC at about 80mA
3.3V DC

Later I decided to also power a wifi ESP8266 module from the same power source. This increases the power consumption of the 5V rail to about 250mA and hence the power dissipation on the 5V regulator is (12-5)*0.25 = 1.75W. Taking into account thermal resistivity of the regulator etc, the temperature rise exceeds 100 Deg C which results in regulator shutdown.
I am now considering possible changes to address the above issue. One way would be to increase track area which will offer further heat sink capability and hence reduced thermal resistance and lower temperature for the regulator. However, I cant really tell beforehand how effective this method would be. In addition the power loss on the LDO is too just much let alone the thermal issues.

I am looking for suggestions to improve the above power supply system with minimal if possible changes while also keeping the cost down. I know its always a trade-off!

With linear regulators you cannot avoid the power dissipation of P=(Uin-Uout)*Iout, for ANY configuration you can imagine.. The heat will be there. No way to get rid of it.. At least in the near future..

Cooling is the cheapest variant - do decrease the thermal resistances as much as possible.. Bigger packages, thermal pads, lot of copper, heatsinks.

The only alternative is a switcher for 5V and 3.3V.

Why have you drawn the schematic right-to-left? That causes confusion and is plain hard to read.

Read to hard plain is and confusion causes that. Left-to-right schematic the drawn you have why?

Why have you drawn the schematic right-to-left? That causes confusion and is plain hard to read.

Read to hard plain is and confusion causes that. Left-to-right schematic the drawn you have why?

Sorry but the picture is an extract from a much bigger schematic which includes MPU, and others.

The only alternative is a switcher for 5V and 3.3V.

5v to 3.3 is not a problem since the actual voltage difference is small and hence Pd is small. The problem is at the 12v to 5v border or even 12v to 3.3v (the circuit above gives this alternative via SJ5)

A 7808 intermediate step-down linear with an external heat sink might be a simple solution.

When I want cheap and easy, I use the old 78xx linears in TO-220 packages with external heat sinks. A couple of decent bypass caps and you're done. For the 3.3 v, I go with the LM1117-3.3 LDO in a TO-220. Very perf board friendly and bullet-proof too.

A 7808 intermediate step-down linear with an external heat sink might be a simple solution.

I am also considering to add another 34063 switcher stage for 24 to 5v or 12 to 5...

There have been a lot of discussions on various forums on this topic and all been finishing with switchers.

Linear regs are good when you mess with radioelectronics (or other noise/EMI sensitive apps), or you do not care on size/weight and power consumption.

There is a lot of modern switchers (the above one is quite old), working up to 2MHz, with quiescent currents down to 10-20uA, some with integrated inductor.

There is a lot of modern switchers (the above one is quite old), working up to 2MHz, with quiescent currents down to 10-20uA, ..

Yes, I know its old. At the same time is very cheap though :)

..some with integrated inductor.

Could you recommend one?

Not sure it is the cheapest, but 2.8x3x1.5mm large :slight_smile:

tps82130.pdf

In linear version above would dissipate 31W.. Like your soldering iron :slight_smile:

Here is a real app with above switcher - 3x switchers there (red pcb, right top quadrant):

Switchers

Wow dead small! Never soldered one of those before!

Not sure it is the cheapest, but 2.8x3x1.5mm large :)

Real nice this chip... but too expensive .... at about 8 euro each. Guess its because is new.