Regulating a bare trickle of power for a microcontroller?

[Scodiddly]

I've been fooling around with the TI Launchpad w/ MSP430 chips for a bit now, and I'm starting to use it for things. Nothing fancy, could probably be done with simple logic or a 555 chip perhaps. Mostly LED dimming and control.

Anyways, these MSP430 chips barely use any power at all, something like 60uA at the most and usually much less. I made a little Altoids box that does nothing but blink LED's and play annoying noises, and a pair of half-dead AA batteries ran it for almost a month. Now I'm putting together a desk lighting controller that will control and/or dim some 12v LED strip lights on different parts of the desk, and so I've got a 12vdc supply and the controller will only need in the neighborhood of 3vdc. 

So I'm thinking that the usual suspects (LM317 and related) would be overkill. I only need a tiny trickle of current at the most... so maybe just a voltage divider with a big cap to act as a buffer? Or maybe a fairly high value resistor down from the 12vdc to a 3v zener for regulation and a cap to act as gas tank. Definitely don't want the 3v supply to waste more power than its customer actually uses!

 

[dfuruta]

maybe look at micropower regulators?  there are lots and lots of different ones available now.  i used an lt3014 for something, which draws 7uA of current.  i'm sure others draw less.

 

[PRR]

Efficiency is cool, but in a "desk lighting controller" the lights will suck 100 or 1000 times more power than your 60uA controller. You also have to conserve brain-power. K.I.S.S.

LM317 with 2mA bleeder may be 20X over-kill, but if you have LM317 handy that's the way I'd go.

3V Zeners are not often handy to me. And common 3V Zeners need over 10mA to hit 3V. If I had some 4.7V Zeners I might try them at 1mA and 0.1mA, see if they get within tolerance for your CPU. However I'd fear the zener voltage would age badly.

 

[Scodiddly]

Micropower regulators, you say?  8)  I will check those out.

Yes, I'm worrying about the small side of the energy draw. But most of the time the desk lighting will be off, while the control stuff will be running 24/7 (assuming I don't keep obsessively unplugging it when I go to bed).

 

[JohnRoberts]

I use a nice Microchip LDO 3,3V regulator in my battery powered application. The VREG is low drop out and very low current wasted too. Probably CMOS inside.  I don't recall the part number but they make different output voltage models. 

One thing I notice when working with low voltage IC, sometimes you need to worry about your max input voltage. I don't recall what the Vreg will take but one of my other chips gets unhappy at around 6V.

JR 

 

[Rochey]

I second the LM317... your quiescent current of the LDO may be a high percentage of the power consumption.

Your other option is to switch to a USB power supply, and use a few diodes to drop the voltage? The '430's are pretty good at running below 3V (down to 1.8v?).

 

[PRR]

> Yes, I'm worrying about the small

I'm thinking 10 Watts of light 2 hours a day.

And 0.001 Watts of micro-brain 24/7.

Figure the total electric bill for four alternative methods.

Don't forget to pay yourSELF a buck an hour (whatever you think you are worth, perhaps much more) to do this figuring.

My 2 cent/hour brain is saying an eighth-cent a YEAR for the CPU alone. 13 cents if you live another century. (Plus universal inflation.) Maybe I dropped a bunch of zeros on the floor. But hard to justify another 2-cent hour of thinking for a one-off. I'd take the first 1% efficient plan I found in my drawers.

Also: I assume the 12V lamp supply is always-on? It may be 10W max and 0.2W loafing. That's 10X the loss of a LM317 at 2mA bleed.

Salt-water dimmers and knife switches. When the show is over, you pull the paddles, cut the knives, standby power is *zero*.

 

[JohnRoberts]

I second the LM317... your quiescent current of the LDO may be a high percentage of the power consumption.

Your other option is to switch to a USB power supply, and use a few diodes to drop the voltage? The '430's are pretty good at running below 3V (down to 1.8v?).

Not to get pedantic, but the LM317 is dumping 50-100 uA down the regulation string, and spec's minimum 3.5-10mA load for regulation, I will repeat that the Microchip CMOS or at least MOS parts are a low current solution. Looking at the MCP1702 (the first part I found on google.) Specs a 2uA quiescent current.

You are correct that typical bipolar LDO regulators are inefficient and dump a bunch of quiescent current down their ground leg, but using MOS or CMOS pass devices means they are voltage controlled not current controlled like bipolar regulators so can be low current.

I will also second PRR's point to do the math...  A couple mA 24x7 is no bigee... My little smart thermostat runs 24x7 and I could turn off the LEDs display, but that current is insignificant vs the power saved from smart temperature control.  So I leave the display on.

JR

 

[Rochey]

My recommendation for the 317 was based on being able to walk into Radioshack and buy one... even these days.

 

[JohnRoberts]

My recommendation for the 317 was based on being able to walk into Radioshack and buy one... even these days.

Be careful, I hear rat shack is closing a bunch of stores.... due to severe profit issues.

JR

 

[Scodiddly]

Followup - I ended up buying a couple MCP1754ST (they were in stock at Mouser and I had other stuff to buy there). Good thing I bought two, because I cooked the first one somehow - probably a result of overheating it with my soldering iron; SMD parts on perfboard is not the most (or least) elegant thing I do while building things.

Anyways, it seems to work perfectly. A couple little eletrolytic caps on the in and out to keep things smooth and everything is working reasonably well.

The dimmer setup itself is working OK; my code for the microcontroller is not very good (yet) and requires tapping the buttons just right, but everything else is good. I used FQP30N06L MOSFETs to do the 12v switching, right off the MSP430 output pins. I might revisit that part of the design because the dimming seems a little blinky, but overall I'm happy. The 120vac->12vdc switching supply stays cool, the little Bud box with all the guts stays cool in any of the modes, so I'm not wasting any real power. Overall win because I had two lighting areas and generally only use one at a time, so most of the time I've cut power needs in half.

Thanks for the help!

 

[pvision]

If the micro controller takes microamps, why not power it with a photovoltaic cell?  Power the controller via the light emitted by the lights it's controlling…

It'll need some kind of bootstrap technique to make it work but it'd be elegant, albeit somewhat eccentric

Nick Froome

 

[ruffrecords]

All the regular regulators like 7805 come in L versions that used to be available an a little TO92 package. They will give up to 150mA output. Should be ideal for the job. Check out the 78L03.

Cheers

Ian