Regulating 22V down to 3.3V

[rea5245]

Hi,

I'm designing a board that will fit inside my old Heathkit clock, get the time from GPS, and set the clock. The circuit is designed and tested on a breadboard, the software is written, the PCB is designed; I'm almost done. But then I thought: what about the heat?

The clock, a GC-1107, runs off 8V to 21.2V. The voltage is varied in response to the room's ambient brightness so that the display brightens and dims. I need to get 3.3V, 80mA from that for my board. My first thought was to use a linear regulator, rated for at least 150mA, in a SOT223 (I'm using SMD since space is tight: my board has to fit inside the Heathkit case). But then I started looking at heat and it looked like that poor little SOT223 would be generating about 6W. I'll bet it wouldn't be generating it for long. :-)

How can I sink that kind of heat? Is there a ghost of a chance my PCB traces would provide sufficient sinking? Should I switch to a TO-220 with a heat sink and just try to fit it in? Is there a more heat-efficient way to cut 21.2V down to 3.3V without taking up too much real estate?

Thanks,
   Bob

[JoeyP]

Use a switcher. Take a look at the On semi CS51413. Cheap, small and can handle a very wide input range. Data sheet includes an example for 3.3v supply.

[Psi]

This is about as good as it gets

RECOM  R-78E3.3-0.5

- US$2.80 from digikey in qty=1
- 6V to 28V input
- 3.3V output 500mA max
- 75- 88% efficiency
- Throughhole and with 78xx regulator pinout
- Short Circuit Protection

http://www.digikey.com/product-detail/en/R-78E3.3-0.5/945-1661-5-ND/3593412

[planet12]

Definitely go the switcher route - however if you've never built one before and don't want to dive down a rabbit hole of learning (hey it's fun, but not always the goal), go with something like the prebuilt module Psi suggested. No need then to find suitable inductors, capactors, and to learn how to layout a switcher so it not only works but works well.

Also, a note on your dissipation calculations - at 21.2V in, 3.3V out and 80mA, it's (21.2 - 3.3) x 0.08 = 1.432W, at 150mA it's 2.685W - still a lot, but not near 6W.

[rea5245]

Also, a note on your dissipation calculations - at 21.2V in, 3.3V out and 80mA, it's (21.2 - 3.3) x 0.08 = 1.432W, at 150mA it's 2.685W - still a lot, but not near 6W.

Hmmm.... I think I see where I went wrong. The formula in the data sheets includes the term I_G x V_IN (ground current times input voltage). I read the I_G off a chart, but as I look at it again, I see the chart is in microamps; I thought it was milliamps. That'll make a big difference.

So now that I'm down near 1.5W, I have to ask: do I still have to worry about this? Would a linear regulator be OK? Junction-to-ambient thermal resistance is 64.7C/W and maximum junction temperature is 150C. It sounds like I'm safe at 80mA even on a hot day, especially since I'm in the US and we don't have any Celsiuses here.

- Bob
 

[Psi]

With a small heatsink attached to a TO220 regulator 1.5W is ok

I wouldn't run it with no heatsink, 97deg above ambient is too much.

You want to keep the regulator below the point where you have to let go within 5 seconds of touching it.

[rea5245]

Thank you JoeyP, Psi, and planet12!

- Bob

[planet12]

You're welcome. Also remember when doing thermal calculations that if you've got something inside a case, unless there's sufficient airflow through it you need to take into account that the internal "ambient" temperature will also rise vs. outside the case.