Voltage Regulation

[Bern]

Looking for a suggestion on a voltage regulator. It will be used in a system with batteries, and running continuously. It will be supplying about 4 or 5 volts, at a current of less than one milla-amp. Looking for minimal drop out voltage and operating power. The normal regulators have idle current of a few mills, which is too much for this application. Thanks again guys.

[MrAl]

Hi Bern,

Need to know input voltage range and exact output voltage you
want...is it 4 or 5 volts, or does it have to be adjustable from
4 to 5 volts?

[Bern]

MrAl -- Input maximum of about 10 volts, and output in the 4 to 5 volt area. Not critical, but has to stay constant with input dropping as the batteries age, and has to stay solid with changing load.

[philba]

I would definitely use a switcher, probably a buck regulator. A linear will be inefficient, wasting a lot of the battery as heat.

I like national's web bench - all you have to do is plug in your numbers and it spits out a design. (OK, it's a little morte complex than that, but not much). You don't need to understand SMPS design theory.

http://webench.national.com/appinfo/pow ... /index.cgi

[Bern]

A switcher sounds like too much for what is needed. I need many of them, so money is critical. I have very little room. The whole system will be running on micro-amps, so just the control circuits in a switcher, at first, sounds undoable. ---- Thanks, and please keep all ideas coming.

[MrAl]

Here's a breakdown of using a switcher:

E1=10
I2=0.005
E2=5
P2=E2*I2
P1=P2/0.85
I1=P1/E1

E1=input voltage
I1=input current
E2=output voltage
I2=output current
P2=E2*I2
P1=P2/eff, eff=0.85
I1=P1/E1

If we calculate all this out, we get:
I1=0.00294

and if we look at the ratio of I2/I1 we see that we get
about 70 percent more runtime if we use a switcher.

This means if we dont use a switcher (use a linear) and we get
1 hour runtime, if we instead use a switcher we get about
1 hour and 42 minutes runtime. Thus, we almost double the
runtime if we use a decent switcher.

The choice is yours....

[jimandy]

What about just using a 5V zener and resistor combo.

[haklesup]

Maxim IC has some nice Low dropout low Iq regulators. For example the MAX1725 ands MAX1726 will supply 20mA at 5V using only 2uA to power itself. Dropout at 20mA is only 300mV so 4 AAs should last a while.

They come in a tiny SOT-23 package and cost only $0.82

http://para.maxim-ic.com/cache/en/results/5094.html
Many other to choose from but these seemed closest to what you asked for.

THe MAX8880 is worth a look at too as the dropout is only 100mV at 50mA and Iq is 3.5uA and has a slightly larger QFN package

Which is more important, low dropout or low Iq? I suspect the MAX8880 will squeeze more out of a battery.

[philba]

I don't see how different linear VRs will make much difference. 10V in, 5V out, you are wasting half the energy. Yes, a really efficient linear will help some but it's going to be the difference between wasting 50.5% or 51%. The dropout doesn't matter since the difference between Vin and Vout times current is whats being dissipated.

[haklesup]

Low dropout counts when you have a battery source and you want the regulator to work as long as possible as the battery dies and the voltage falls. The spec is at the device's max current so running it at lower current yeilds an even smaller dropout voltage (even better)

Ultra low Iq counts as it determines how much current the regulator uses just to operate (i.e. output current =0, you would get this at the input). This is particularly important to squeezing every but of the capacity out of the battery. Not many used more than 1mA

With fresh 4 AA batteries in series you get 6V, 1500mAh (nominally 9Wh available to burn). With a load of 1mA at 5V you loose only 1.002 mA * 1V = ~1mW in the regulator and 5V*1mA= 5mW in the load. Power loss in the regulator decreases as the battery discharges until about 5.3V when it will quit regulating (MAX1726 example). so 9000mWh/6mW= 1500 hours (assuming you can use all 1500mAh of capacity before dropping out)

Or looking at the current only, if the system draws 1.002mA and the voltage drops below dropout at 33% capacity (just a guess) then I get 1000mAh of useful juice or about 998 hours of operation.

That's pretty darn good in any book and at 82 cents per device it beats any switcher for cost and simplicity.

With rechgargables at 1.4V per cell it gets tighter but still dooable. You need to look at the discharge curves for the battery type you plan to use to make a realistic estimate of operating life given the load and minimum operating voltage (dropout V). It would be useless to try to put 5 or 6 batteries in series as you will still have the same capacity but will burn more power across the regulator. If you want longer run time, parallel another bank of 4 batteries. You want the unregulated voltage as close to the regulated voltage as possible to minimize losses across the regulator without going below the dropout.

Another alternative is to redesign the circuit using CMOS 3.3V locgic chips and use a 3.6V coin cell battery without regulator as the primary source.

[philba]

sure the the ULDO will work a little longer but when the batteries are heading south, there isn't much life left, But we are debating 1% issues. A switcher would run much much longer due to a) much higher efficiency and b) a much wider input voltage range. A small switcher isn't that expensive and it will squeeze even more of the life out of a battery.

Using your example though, why not just run with 3 batteries at 4.5V (or 4 nimh for 4.8V) and no regulator at all. you'd get longer life out of that with no wasted energy. you most certainly can't beat the price since you don't need the VR circuit and you save on a battery. There may be some design issues but in this case you'll get more life with no regulator than a linear.

[k7elp60]

Im on vacation and don't have but my memory to go by, but you might try a TL431 series. My recollection is they have real small idle current and are good for up to a couple of Ma.

[Bern]

Thanks for all the inputs, everyone. I have been gone for several days, so missed a lot of the action. For this application I think I will use a combination of ideas. If I break the load into two sections with a R & C, the critical steady load and the variable load, along with your inputs, I think I will have what I need. That MAX 1726 is what I was looking for, I just didn’t dig deep and long enough to find it.