Recommendations for dual-feed switching regulator

Go To Last Post
4 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'm looking for an elegant way to power a project, something that doesn't take up too much board space ... It's an LED controller that's powered from 12V normally, but also has a USB interface - planning on the uber-cool V-USB stack to drive it. The Tiny861 will run at 3.3V (or maybe 3.6V if that works out better ?)

The box will normally be powered by 12V coming into a switching regulator solution, but has to be able to accommodate being powered by USB for configuration. Code will detect USB commo and cut power to the LEDs so we don't try to pull too much power from the USB line.

There are a couple of ways to do this. I suppose the simplest is to use a typical 3.3V buck switcher to handle the 12V supply, and a couple of diodes in the USB power pin to drop that to 3.6V, followed by a 3.3V zener. Does it need that zener ? Would there be problems with a 3.3V switcher and the 3.6V from the dropping diodes at the same time ? Could use an adjustable switching regulator to put out 3.6V, but that takes more board space from the samples I've been looking at from Linear and National.

This feels kludgy - there must be a better way. I remember seeing several regulators in EDN or somewhere that take multiple Vin inputs, just haven't found the blurb articles yet. Ideally, I would like something that can take +12V on Vin-1, +USB5V on Vin-2 either individually or both at the same time, and spit out 3.3V.

Any good ways ? Anything to avoid ?

Dean 94TT
"Life is just one damn thing after another" Elbert Hubbard (1856 - 1915)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Let me see if I follow you completely: do the LEDs run off the 3.3V supply from the switching regulator?

Can you use two schottky diodes to select the switching regulator's input from the USB input or the 12V input?

Alternatively what I might do would be to use the switching regulator power just the LEDs from the 12V power, while the attiny has a separate linear regulator (at 3.3V, you're only going to use a max of about 3-4mA, so the power loss is pretty small). The linear regulator will either take its input from the 12V power or from the USB using the schottkys. There are many sot23 linear regulators that are cheap as chips.

- S

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Have you looked for dedicated power mux chips that might do the job? I had a very similar situation to yours -- an LED controller, normally run from a dedicated PSU, that had to be able to run from USB power for config -- and settled on a TI TPS2111A power mux. But then, my board runs at 5V, so I just needed to switch between USBV+ and the output of a 5V LDO. The LEDs were connected straight to the upstream side of the LDO, so with no external PSU connected they simply didn't get power. The TPS2111A is only good for 5.5V, but there may be other power mux parts good for up to 12V that you could sit between the 12V and USB supplies and the AVR.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yup - want to run the LEDs from as low a voltage as possible. Power is sourced via high-side P-fets, and controlled via a TLC5916 (on/off and constant-current). To minimize the dissipation in the tlc5916 one wants to run the LEDs as close to forward voltage as possible. 4 banks of 8 LEDs at up to 50ma each. Only one bank on at a time, so about 400mA draw.

I was originally going to run the LEDs off the 12V supply and the attiny off regulated 5V. The hassle comes in the P-fet control of the source for the LEDs. Turning them on is easy - pull the fet gate to GND. Turning them off is tougher - the gate has to go up to +12V, which means an N-fet driver each, which means more board space, which is of course at a premium.

There's more discussion about the driving possibiities at
https://www.avrfreaks.net/index.php?name=PNphpBB2&file=viewtopic&t=114951

I think the options are :

1) Two schottky diodes to feed a 3.3V switcher from USB and from 12V. Don't know why I didn't think of that before - so simple, sheesh. Switching regulator takes more board space - inductors are big. High-side P-fets can be driven directly from attiny. I think this is cheapest and reasonable board space. A really high-freq switcher means a smaller inductor.

2) Feed attiny via LDO linear 5V regulator, tied to USB, maybe a diode to isolate. Power LEDs from 12V. Needs P-fet drivers (quad N-fet or similar). Smallest board space for regulator. Possible problem of dissipation in tlc5916 - I don't know how it regulates the constant current per channel. The similar tlc5940 is linear ... The tlc5916 says it can handle 120mA per channel, 8 channels - that's pretty hefty.

3) Find a dual-input 3.3V switcher. I know they're out there - I even tore out a Design-Idea article or something from one of the magazines. I just have no idea where I put the damn thing.

Meh - I think the two schottkies is the way to go. Now I need a high-freq 3.3V switcher that's hand-solderable :) Dang it - board layout has USB diametrically opposite the 12V input. Back to shuffling components around the board.

Dean 94TT
"Life is just one damn thing after another" Elbert Hubbard (1856 - 1915)