Attached is my attempt at a block diagram.

The regulator I am considering is here. Your opinion would be appreciated.

That's a pretty handy regulator! It looks like it will do the trick. Remember to turn your device off, though, as its quiescent current is quite high (in the mA's).

AA's are 1.5V charged, discharge between about 1V and 1.1V: http://data.energizer.com/SearchResult.aspx . I would say that using 4xAA's, you could drain them all to 1.1V each, which is basically dead, and still replace them 1 at a time and maintain your 5V or 12V desired output (the regulator would boost the 3.3V). Let's crunch some numbers...

The most power you could pull from your specs and this regulator is either 103mA at 12V (1.24W), or 248mA at 5V (1.24W), both corresponding to maximum currents at those output voltages for 3.3V input voltage. The regulator won't accept more than 500mA input, so let's see what the battery drop is. The max IR is around 300mohm for Energizer, but lets say you're using a cheap brand that has 400mohm, which makes its voltage drop by \(500mA \times 400m\Omega = 0.2V\) per battery. The actual input is then \(3V-3\times 0.2V = 2.7V\), which is still 0.2V above the regulator minimum.

Now let's see what you can do with fresh batteries, or 6V. Looking at the "5V in" table you could output 156mA at 12V (1.87W) or 375mA at 5V (1.88W). The 12V output current is limited by the 500mA input restriction, corresponding again to battery drop of 0.2V; the 5V output current is limited by the conversion efficiency and thus an input of 500mA, for another 0.2V battery drop. The battery output is actually \(6V-3\times 0.2V = 5.4V\), which really doesn't make a difference.

Now let's compare the performance of this regulator against that of a LDO regulator, like the LT3080, which has a dropout of 350mV. First, you won't get 12V with a linear regulator. Second, you are not able to "hot swap" one of 4 AA's -- you would need to use at least 5. With 5 batteries, you can drain them to about \((5V+350mV)/4=1.34V\) before having to swap one out, which doesn't give much run time. With 6 batteries, you can drain them to 1.07V each, which is basically dead, so let's use 6 batteries. Matching the available output power with the switching regulator, 1.88W to 1.24W, the LDO will pull \(1.88W \times 9V/5V = 3.38W\) (56% efficiency; 375mA) to \(1.24W \times 5.35V/5V = 1.33W\) (93% efficiency; 248mA) from the batteries. If the load follows this power curve, the batteries will last about 2.5 hours (look at "Constant Current Performance: Typical Characteristics" and interpolate). The switching regulator (SwR.) will pull \(1.88W / 75% = 251W\) (75% efficiency; 500mA) to [/tex]1.24W / 70% = 1.77W[/tex] (70% efficiency; 500mA) from the batteries. With the same power curve, they will last about 1.5 hours. Now compare a 1W load, or 200mA at 5V. The LDO follows the "Constant Current Performance" line, allowing 6 to 7 hours of use. The SwR. follows the "Constant Power Performance" line, pulling \(1W / 75% = 1.33W\) or 222mW per battery, which allows 7 to 8 hours of use.

So why'd I go through all this math? 'Cause you really won't know what battery solution is best until you crank out the numbers. When looking for a power solution for your widget, you often have an Excel spreadsheet that you've setup to calculate most of this stuff for you. As you can see the load makes a big difference: in one case the LDO was better, in another the SwR. was better. If you are using less than 1W loads or want 12V output, use the SwR.; if you are using more than 1W loads or want more than 500mA, use the LDO.

Extra battery calculations for 1W load and 4 batteries: \(1W / 75% / 4 = 333mW\) per battery, resulting in about 2 hours of battery life.

 

Looking at your block diagram, here are some new numbers:

The switching regulator will pull about \(500mW / 75% = 667mW\) from the packs, or about 56mW per battery, giving them a runtime well in excess of two days. With only 2 battery packs it would still run at only 84mW per battery giving more than 36 hours. The LDO would only allow the batteries to drop to \(5.35V/4 = 1.34V\), but still give around 30h of service using only \(500mW / 5V / 3 = 33mA\) per pack.

 

There are way too many numbers up there for me to have not made a mistake, so let me know where they are.

 

Based upon all of the input obtained so far, I wanted to summarize my current understanding of the project and the questions that I still have. I have tried to provide links to everything from previous posts so the discussion can continue from here. Please correct any inaccuracies.

• To power the 500mW AV transmitter, the AnyVolt regulator will need a 667mW input because of its 75% efficiency.
• There will be a power loss across the diodes that are needed to isolate the battery packs from each other. How do I determine the appropriate diodes to buy? And how do I calculate their power loss?
• Based on the block diagram, will there be any other losses to take into account for wiring, connectors or fuses? Or would these be considered negligible? Is a 1A fuse at the output of each battery pack appropriate for this circuit?
• There will be a voltage drop at each cell due to the internal resistance of the battery. I couldn't find the IR for the 2900mAh batteries I will be using so I will assume the stated worst case of 400mΩ. Depending on the output current, the voltage drop per cell could vary from a maximum of 0.2V (500mA X 400mΩ) to the more likely (for this load + associated components) 0.1V (250mA X 400mΩ).
• NiMh AA's can vary between ~1V (fully dead) and 1.5V (fully charged). Taking into account the voltage drop, the output of a four cell battery pack in this project should vary between 3.2V (4 X 1V - 4 X 0.2V) and 5.6V (4 X 1.5V - 4 X 0.1V).
• A 3.2V input @ 208mA (667mW / 3.2V = 208mA) should get about 14 hours from each battery pack (2900mAh / 208mA). A 5.6V input @ 119mA (667mW / 5.6V = 119mA) should last about 24 hours per pack (2900mAh / 119mA). Those numbers seem right mathematically but not practically. It doesn't make sense that a pack of four "dead" batteries can run anything for 14 hours. Is there something about current that I am missing? (like dead batteries can't really supply 208mA)
• After looking at the LCD's suggested, I could see how I might want to drill some holes and mount an LCD on the plastic case of each battery pack. However, I'm still unclear as to how you program the LCD. I have done this with a PC, some software and this LCD display. But that is obviously a completely different situation. What equipment do I need to program an LCD like those suggested? I looked at this Volt Ampere Meter Kit but I don't think I want to spend an additional $40 for each battery pack. I also found these Panel Meters. If you scroll down to "LCD Panel Meter - 5V Common Ground", it looks appropriate to this project but there are absolutely no instructions.

Thanks to all who have responded with your time and counsel. It has been very informative and instructional.

 

To Display character and info on an LCD, u have to interface it with PIC.
write the code for the program, get PIC KIT and program the PIC.

This way you can do almost anything you want.

Is there any size limitations. I mean can you use a Li-on instead of a NiMH.
2 Li-on's will give you 8.4V when fully charged and 7.4V when discharged.
This way you can use tht Anyvolt thingy to get the desired output.

2 Li-on's can provide enough voltage and lasts much longer than NiMH. The diode drop will be negligible in this case.

And designing a Li-on charger is child's play.

 

• There will be a power loss across the diodes that are needed to isolate the battery packs from each other. How do I determine the appropriate diodes to buy? And how do I calculate their power loss?

A diode will have an approximately constant forward voltage associated with it. A common Si has 0.6-0.7V forward drop; a Schottky has 0.2-0.4V drop. Power dissipated is P = I*V.

• Based on the block diagram, will there be any other losses to take into account for wiring, connectors or fuses? Or would these be considered negligible? Is a 1A fuse at the output of each battery pack appropriate for this circuit?

Negligible. 1A will probably be fine. Max current in any line will be ~500mA, when you have only 1 battery pack hooked up and the output shorted, and it doesn't look like the regulator will fry if overloaded for a short period of time. YMMV, of course!

• NiMh AA's can vary between ~1V (fully dead) and 1.5V (fully charged). Taking into account the voltage drop, the output of a four cell battery pack in this project should vary between 3.2V (4 X 1V - 4 X 0.2V) and 5.6V (4 X 1.5V - 4 X 0.1V).

You may want to use 1.1V as "fully dead", as a battery's internal resistance increases at this point (part of why a "dead" battery's short-circuit current is so much less). Even if they can deliver the required current, it is best to underestimate power budget than over! Seems like the packs will power your device for a very long time, regardless.

• A 3.2V input @ 208mA (667mW / 3.2V = 208mA) should get about 14 hours from each battery pack (2900mAh / 208mA). A 5.6V input @ 119mA (667mW / 5.6V = 119mA) should last about 24 hours per pack (2900mAh / 119mA). Those numbers seem right mathematically but not practically. It doesn't make sense that a pack of four "dead" batteries can run anything for 14 hours. Is there something about current that I am missing? (like dead batteries can't really supply 208mA)

You're right, a dead battery won't do that; by the time it has reached 1V, its entire usable energy reserve, or most of "2900mAh", has been used up. The number "2900mAh" is stated for a particular discharge curve, and changes based on the load. It is better to use the discharge curves directly to determine battery life. Since you are using a switching regulator, look at the "Constant Power Discharge Performance", or something similar, since current draw will not be constant (ie: 2900mAh/I is not valid).

• After looking at the LCD's suggested, I could see how I might want to drill some holes and mount an LCD on the plastic case of each battery pack. However, I'm still unclear as to how you program the LCD. I have done this with a PC, some software and this LCD display. But that is obviously a completely different situation. What equipment do I need to program an LCD like those suggested? I looked at this Volt Ampere Meter Kit but I don't think I want to spend an additional $40 for each battery pack. I also found these Panel Meters. If you scroll down to "LCD Panel Meter - 5V Common Ground", it looks appropriate to this project but there are absolutely no instructions.

An LCD has a few different parts: there's the backlight, the contrast, the driver, and the data input. The contrast voltage is often <0V, though you may be able to see it at 0V. Make sure the backlight voltage is within your range (ie: 5V). Also best to get an LCD with a built-in driver, as they are a bit of a pain to program yourself. The last item is the data input, which can be parallel but is predominantly serial, and provided by a microcontroller or other serial bus. You may want to use one of the parametric search tools provided by a few electronics distributors (DigiKey, Mouser, RS) to look through the different options available for LCD displays. For ease of implementation, I would go for one with a built-in driver, uses a single supply (not dual-rail; ie: it generates its own negative contrast voltage), and uses the I2C/TWI serial interface.
In my opinion, the easiest way for you to put text on them is to get an Arduino. They are geared towards beginners, streamlining the process of implementing the micro's kernel circuitry and programming. Another simple option is a small development kit, which is similar to the Arduino platform but is moreso made for professionals wanting to rapidly prototype a design and/or try out a microcontroller family: the PICkit 3 Debug Express (part# DV164131) would do it, and also provide a programmer for future micro projects.
A third option exists: integrated LCD, driver, microcontroller, and programmer packages. I haven't looked around for them, but I'm sure they exist. Something like those programmable scrolling billboards. Maybe someone else knows of a product.