step-down converter recommendation?

[gorsat]

I'm working on a project that includes a microcontroller, an Atmel dataflash memory chip and an audio amplifier. Because the Atmel dataflash chips are only designed for 3.3V, I'm running the whole circuit at that voltage level rather than dealing with the added complexity of running with both 5V and 3.3V. As an added benefit, the power consumption of the circuit is lower than I expected.

The whole project will have to run on battery power. At this point, due to size constraints, convenience and cost, I've decided to use a 9V battery as the power source. The toy is designed to be used in short bursts of only a few minutes at a time. During active use it will consume 50 mA at most. It will automatically go to sleep when not in use and should consume no more than about 10 µA when sleeping. Thus, a typical 9V alkaline battery (with on the order of 500 mAH capacity) should last a good long time.

I've been powering my prototypes from a 3.3V power supply up until now, but it's time to finalize the circuit, so I'm looking for a simple, efficient and low cost way to convert from 9V (really 9V-5V over the life of the battery) down to ~3.3V (and 3V would be just fine, too).

The only chip I've found that meets the criteria (simple, efficient, low cost and consumes hardly any power in standby mode) is the ST763A. I'm ordering a few of these to play with. That said, I've searched the web a bit and haven't found anyone else talking about this chip. Does anyone have experience with it or have a recommendation for another option?

 

[MikeMl]

Try looking up the 34063 chip. It is the basis of millions of 12Vdc input cell-phone chargers.

 

[Mr RB]

Your big problem is going to be the amplifier. Even small low power amp chips have nasty quiescent current, 4mA is a good one. So you will probably need a system to turn the amp power completely off when not making the sound.

I faced this problem recently on a commercial project design, and it required some lateral redesign of the product. Good luck.

 

[gorsat]

Regarding the 34063: It doesn't have a shutdown or standby mode. I need the ability to shut down (or put to sleep) the entire circuit from the MCU. I suppose I could use a 34063 with some sort of beefy switching transistor controlling the supply voltage. Would that be a reasonable approach? If so, then do you have any transistor recommendations for the job? Also, I can't find efficiency figures for this chip (at least not the ST version). If I use it to convert from 4.8V-9V down to 3.3V do you have any idea what sort of efficiency I should expect?

Regarding the amp: I'm using the TDA8551 and it has remarkably low current consumption in standby. It's rated at a max of 10 µA and I'm seeing it lower than that -- ~4 µA according to my old, imprecise multimeter. Surprisingly, the current pig turned out to be the Atmel dataflash chip which runs at 25+ µA in standby. I ended up using an IO pin on the MCU to control the dataflash chip's GND connection. That did the trick.

In general I'm just wondering, with the digital world headed toward lower power (<= 3.3V), and with the need for small, cheap, battery powered gadgets, why aren't there *several* competing alternatives for highly efficient and cheap DC-DC converters? Is everyone content using terribly inefficient linear regulators? Or is everyone using the 34063?

 

[Mr RB]

Yeah the TDA8551 was one I looked at earlier, 6-10mA quiescent. That standby current looks ok though.

Re the voltage regulators, I think as the OPERATING current of so many devices is so low now, in the uA range, there is little to be gained from low power buck regulators vs their big footprint and the cost of expensive inductors and schottky diodes etc etc. Especially with the low battery voltage to device voltage overhead, SMPS work best at 2:1 voltage conversion or more, at 1.2:1 they suck. So most of the new regulators are just very low dropout very low quiescent current linear regs. Like the nice XC6206.

There are still plenty nice little buck devices around for the times when you need larger currents like your 50mA. Sorry I don't have any part numbers handy but you could google "ultra low power buck regulator IC".

(edit) a few seconds on goolge found these that seem good for your 9v app;
**broken link removed**

also it looks like fairchild, nat semi and maxim all have similar products, I also tried "DC-DC converters" and or "switching regulators" in the search.

 

[gorsat]

Thanks.

I'm going to play around with a few of these converters.

You mention the "low battery voltage to device voltage overhead". Is there a common battery choice for powering ~3.3V circuits? Two AAA's doesn't provide enough voltage and three of them takes up too much space in my case.

 

[Mr RB]

That page I linked also had some step-up converters that are ideal for 2 cells -> 3.3v, they were up the top.

Many commercial products just run unregulated from 2 cells, you could do that.

Alternatively the Atmel flash will run fine from 2.5v and your micro might too, so you could use a 2.5v LDO linear reg but your amp chip won't be real happy.

 

[chuddleston]

I use the AP1509 switching regulator for my low-power 3.3V applications. It's very efficient and has a 70uA shutdown current.

 

[gorsat]

Thanks chuddleston. The AP1509 looks similar to the ST763. These seem to be convenient, low component count solutions. That said, the overall cost of these looks quite a bit higher than the 34063, even considering the extra components required. I'm going to play with several of these and see what I think of them.

Mr RB: You're exactly right that the amp is the problem child when it comes to reduced voltage. With a couple of AAA's, the min voltage will get down below 2V even when there is quite a bit of life left in the cells. Do you know of a mono amp that works at such a low voltage level? I haven't found any.

On another note, I did find what might be a better amp choice than the TDA8551. It's the SSM2301. It's very inexpensive, has a quiescent current of 3-4 mA and a shutdown current of 20 nA. Any experience with this one?

 

[Mr RB]

Never seen the SSM2301 but thanks for the heads up, I just got the datasheet now. Looks very nice, 2.5-5v powered, 1.4watt, class D with no external filter components needed. And you can't argue with 20nA !

Are you really concerned about operation at 2v?? If so then just go for one of the step up 3.3v regulators in my link or chuddleston's regulator and foot the extra cost. But if it's for intermittent use shelf life will probable make the batteries spit up before they get to 1.2v (2.4v total). Most people these days buy bulk packs of Chinese batteries in the supermarket and just replace them anytime an appliance gets iffy.

I know you might not be at liberty to discuss what this product does on an open forum, but for most typical apps it might just be good enough to go 2 cells unregulated. Lots of cheap products do.

(edit) Did you have a look at the S-8353 step up switcher in the link I posted? 0.5uA standby, 19uA running, over 100mA out, input voltage 0.9v and up, output voltage 1.5v to 6.5v selctable in 0.1v steps.
And tiny in an SOT23-5 package they cant be too pricy.

 

[gorsat]

Just in case anyone is interested...

I tried the 34063. It's a good, solid chip. It worked as expected, but created more ripple (i.e. >100mV) than I liked even with a large (e.g. >300uH) inductor on the output. I could make it work very nicely at any particular input voltage, but it didn't handle the full input voltage range (5V-9V) well. It also doesn't have a shutdown or standby mode, which made it unusable for me.

So I've switched to the S-8521A33MC. It is GREAT for my purposes. Very few parts required, very low ripple voltage across the full 5V-9V input voltage range with a small (~33uH) output inductor and it also has a shutdown mode. I recommend it for anyone else facing a similar design problem.

 

[Mr RB]

What brand is that chip gorsat? And may I ask the pricing?

 

[gorsat]

It's a Seiko part. I bought it for $0.76/unit. The price goes down below $0.60 in large quantities. You also need a switching transistor. I'm using a cheap little MOSFET.

 

[Mr RB]

Thanks for the info.

 

[bountyhunter]

Does anyone have experience with it or have a recommendation for another option?

Yes, use a linear regulator. It will regulate 3.3V with an input as low as about 5V. With only 50mA draw, power dissipation even at 9V input is only about 0.3W which is pretty low. Using a linear will save you about $4 and endless headaches compared to a switcher. That's the reason you don't find bundles of switcher designs in low current applications. The cost savings of battery life isn't worth the extra cost.

 

[Hero999]

Why not use two AA cells?

I assume it will run down to 2V so you should get good battery life.

 

[gorsat]

I should have mentioned this in my original post: I haven't yet found a good amp that performs well (if at all) below ~2.5V. Otherwise, I'd probably just use two AAA's.