I was doing some final checks on a circuit I've been working on the last few months and I realized I made a major boo-boo with my power calculations.
I was planning on using this regulator:
http://www.micrel.com/_PDF/mic29150.pdf
And when I did the power calculations I used the junction to case thermal resistance. But I now know that only tells me how fast the regulator can move heat into a perfect heat sink, and my 4" sq PCB is not a perfect heat sink, by a long shot even though I have a large copper plane on the bottom and thermal vias. (I am using the surface mount version of the regulator.)
In my hours of research into this, I came across this page with someone who had a similar issue with a similarly sized board:
Long story short, they determined they could dissipate around 2W with a board that size.
But with 12V in 5V 3A out, I'm looking at having to dissipate 21W. (12-5 = 7V * 3A = 21W)
Even at 7.4V in 5V 2A out, I'd still be looking at 4.8W.
I'd have to go with the bare minimum of 6V in, 5V 2A out to get the power dissipated down to 2W. And though that might just work for most of my needs, I'd still end up with a fairly warm board.
The reason for such a wide voltage range is I designed this board to be used in large and small devices and with a variety of batteries. For example one might want to place the board in a small handheld device with only enough room for 4 AAA's. Or one might want to place the board in a large prop with an amplifier that needs a 12V input and they would prefer not to have to use two sets batteries to power both the amp and the board.
So, I need to know what my options are. Aside from the requirement that I can't have a large heat sink or fan (a very small heat sink might be okay but they seem expensive and I'm not sure how much of a different one would make in an enclosed space with not much air flow), I also need the design to be fairly simple and low cost.
The obvious solution which came to mind was to use a buck regulator. The problem is, they tend to be complex, expensive, and I can't find one that can output 5V with a 6V input.
For example, the simplest one I've found is the Simple Switcher:
Sticking one of these on my board would end up costing me $1000 more than the regulator I have on there now. And it indicated the minimum input voltage is 7.5V for a 5V output. I could swallow the $1000 cost, but the 7.5V input is a real sticking point.
The reason 7.5V is a sticking point is I wanted to put these in some handheld devices powered by alkaline batteries but there's only enough room for 4 AAA cells and those will drop to 4.8V after 2 minutes when drawing just 1A. And a 9v fares even worse. Another device I want to power would require around 500mA since it would not be playing audio constantly, and I could fit 6 AA's in that, but those would drop below 7.5V after only 18 minutes.
I'd consider the use of LiPos, but I can't find any discharge curves showing what voltage they start at and when they'd actually drop below 7.5V for reasonable discharge rates. The only graphs I've found are ridiculous stress tests in what I believe is the 50A-100A range. (They use C to define discharge rate, and as far as I know the amps would be the maH of the battery * C, which gives some ridiculous amperage numbers on these tests the RC guys are doing.) Plus a 7.5V LiPo isn't gonna fit in the smaller of the two devices anyway. I'd need to use three of these battery packs on Sparkfun and wire them in series:
Two probably wouldn't cut it at just 7.4v.
Anyway back on the subject of regulators, I also found this one:
But that one still requires 7V and has a large inductor that would take up space on my board I really don't have. I realize I may have to give in on the space issue, but the 7V thing is still a problem.
That brings me to this:
http://www.hobbyking.com/hobbyking/store/uh_viewItem.asp?idProduct=4319
I doubt that this UBEC (universal batteyr elimination circuit) is a buck-boost regulator since it needs at least 5.5V, but it lists a 93% efficiency, so it can't be a linear regulator either. And if it were a boost regulator I'd expect it to require a higher input voltage like all the rest I've found. Is this some kind of hybrid linear boost regulator? I have yet to find any chip with which I could make something similar, but I guess I'm gonna go back on Digikey shortly and have another look around.
Another thing that strikes me about this is how cheap it is. It's less than the price of the simple switcher! $1.50 less. And that's at Qty 1. I'm comparing to the cost of 250 Simple Switchers! What's up with that?
Simply purchasing these for use with my boards instead of rolling my own power supply would seem to be a decent solution to my problem. But they are a bit large and I would think I could get them much smaller and cheaper by rolling my own and using ceramic caps. As is it would make things a bit cramped wit the batteries also being stuffed inside my smallest device.
Anyway, if you have any suggestions, please let me know. I wanted to have this board ready to go out to the PCB house by Monday and this has thrown a real monkey wrench into those plans.