Battery power supply

[Engineer1138]

I'm working on a circuit that has to be powered from a single NiCd cell (1.2V) and I need 3.3 - 5V out at a few mA to run a microcontroller & analog I/O. I found some charge pumps that can give me 3.3V from the 1.2V input, but I was wondering if any of you guys have experience with these devices operating at such low voltages. Any recommendations, or problems I should be on the lookout for?

[Chris Smith]

Chose your parts well, and they are fine.

[philba]

Maxim has several step-up converters that might work for you. a lot depends on your current needs. http://www.maxim-ic.com/Power.cfm look under DC-DC switching regulators - step up

[Bob Scott]

What specifically is "a few mA"?

Did you find this inductorless boost charge pump from TI? TPS60310.
It is SMT, preset output of 3.3V only at 20 mA max. Input 0.9-1.8V.

Bob

[Engineer1138]

I found a few Linear chips that look OK and are guaranteed to start up at 0.9V. I need around 5mA. Maybe a bit more if I decide to pulse an LED everytime the circuit wakes up, but a cap can handle that.

I have never run at such low voltages before, so I was mainly concerned about hidden "gotcha"s that I wouldn't expect.

[Bob Scott]

I have never run at such low voltages before, so I was mainly concerned about hidden "gotcha"s that I wouldn't expect.

I assume that you are designing some sort of circuit for non-hobby use with strict requirements because otherwise, you could probably just add another cell.

If this is the case you don't want any parts that are single source only. I've read horror stories in the sci.electronics.design news group about dropping Maxim as any source of parts due problems in supply.

Analog devices has the same problem of single source and it appears they like it that way. Analog also has "gotcha'd" (if I can use it as a verb) many engineers because unlike normal IC makers such as TI and National, the pinout of Analog SMT parts varies from the same device number with a DIP pinout. Oooh, I've seen 100 pin+ SMT Analog Devices ICs hanging off hundreds of kynar wires reworked to fit on a large prototype PCB just because yet another engineer got the pinouts wrong.

I stay away from Maxim and Analog. They are user hostile.

Bob

[philba]

it may not be an issue for you but i found that some PICs (12F675 to be exact, I'd also expect the 629 to work the same) won't deliver that much current at 3V. I was trying to light an LED and it came on very dimly. Unloaded voltage was correct. Even with out the dropping resistor, the LED was dim. I measured the current through the LED and it was pretty low. Loaded voltage was well less than 2V. I used an NPN as an LED driver and it worked just fine.

I had a long discussion about it in several forums and the answer appeared to be that at the lower voltages, the Rds of the mosfets was much higher. kind of makes sense. Microchip didn't have anything obvious that I could find on this topic.

[haklesup]

The most obvious hidden gotcha I can think of would be to ignore the converter's stand by power when estimating the life of the battery.

5mA for a momentarily active function is reasonable, that much current continuous would kill a battery in a couple weeks (not too bad).

Did you say NiCd http://data.energizer.com/PDFs/CH15.pdf It seems that for slower discharge rates the voltage is closer to 1.3V. You're over 100mA before it gets close to 1.2V. Possibly a gotcha if the converter is less efficient at higher input voltage. Since your using a converter and not a LDO regulator, this shouldn't be a negative factor. Its a tiny range an unavoidable.

You should als consider what will happen if a user puts an Alkaline or Lithium cell in place of the NiCd or NiMH (similar voltage). Will it cause failure, excessive wear or heat.
http://data.energizer.com/

Well, these aren't big potential problems but if you are engineering every last minute of battery life out of the product, you need to consider everything that can consume power.

[jollyrgr]

Some solar yard lights use a single 1.2V AA NiCad battery and a white LED. One such circuit is in the Malibu brand lights and the output to the LED is between five and six volts. Here is a schematic from an old post which I can no longer locate the original thread to. (I have a link to the old thread but it is no longer valid since the redesign of the we site.) Anyway here is the link to the schematic:
http://www.elecdesign.com/Files/29/5886/Figure_01.gif

and the schematic itself:


This is pretty much what is in the Malibu lights. The Malibu solar light recharges a single AA NiCad. Then at night the NiCad is turned back into a five to six volt signal to drive the white LED.

I could not find the original N&V thread but here is the link to the discussion of the circuit posted above.

http://www.elecdesign.com/Articles/Inde ... cleID=5886

[Engineer1138]

The device I need to connect to is powered by one cell and I do have a little leeway in battery selection, but I decided on the charge pump vs adding more cells because the design is cost and space constrained and the IC was cheaper/smaller than the cells.

I chose a device (forgot which one, but it's by TI) with a standby current of about 100uA. The rest of the electronics will wake up every 10 seconds to do an operation that takes about 0.1second. I figured that lighting the LED for that .1sec was a nice visual indicator that things are OK, but it's not a requirement. I'll have to check into what philba said about the output saturation.

[Ni-Cad Steve]

Both Maxim and Linear Technology have excellent step-up switch-mode converters designed to work from one NiCd cell. I really like the Linear Tech. parts. Unlike National Semiconductor, everything that I have ever built from a Linear Tech data-sheet has worked better than expected. Both companies have excellent customer service and you can actually talk to the guys that wrote the data-sheets!