SparkFun Forums 

Where electronics enthusiasts find answers.

Questions relating to designing PCBs
By Mach3Maelstrom
#148263
I want to provide both 5V and 3.3V to my circuit from two 3V lithium-ion coin cells. I'm thinking of one of two ways.
1) Have the batteries provide power to two voltage regulators in parallel (5V reg and 3.3V reg).
Image
2) Have the batteries provide power to the 5V regulator, which then uses a voltage divider to drop down to 3.3V.
Image

The 5V is being used on a few ICs in the circuit. The 3.3V is used to power an XBee radio.

Conceptually, since the voltage is already being regulated through the 5V regulator, I would think it's okay to power the XBee radio through a voltage divider after decoupling since the voltage is already maintained. Plus it would certainly save battery power. However, I may be oversimplifying this and would like some input from you guys.
By fll-freak
#148270
The interesting part of the second schematic appears to have been cut off. If by a voltage divider you mean "two resistors in such a ratio as the center tap is the voltage you want", than that will not be a good solution. The issue is that as soon as you draw any current from the center tap, the simple voltage divider no longer holds true. Pulling more current from the upper resistor now means a larger voltage drop (Dam that Kirchoff). Now if you happened to have a load that was constant, you could take that into account and burn of the extra voltage in the upper resistor. The problem in the Xbee will draw different amounts of current based on when it is transmitting. Your voltage would then fluctuate mostly likely going over the maximum allowable.

So you will need two regulators. The questions is should they be in parallel from prime power, or should the 3.3V takes its voltage from the regulated 5V. There are pros and cons for each way. Putting them in series dictates ALL the current must go through the first regulator. This may mean a larger regulator. In series the second regulator can be smaller in that it has less voltage to burn off as heat. Your mileage may vary.
By Mach3Maelstrom
#148272
Excellent, thanks fll-freak! That's true, I didn't consider that the voltage drop across R1 would change as the current demand changes on the tap (/feels dumb ignoring Ohm's Law).

Thanks for explaining the two 2-regulator solutions, that was another thing for me to consider. I'll do some research on which I should employ since the two largest concerns with this part of the circuit are battery life and noise sensitivity. Any input on that front while I research would be awesome.
By Mach3Maelstrom
#148277
fll-freak wrote:
Mach3Maelstrom wrote:...battery life and noise sensitivity. Any input on that front while I research would be awesome.
Out of my league. Perhaps someone else will jump in.
I'm just going to test for the noise while I'm proto-boarding it. Otherwise, I'm going to base the design around the heat the 5V regulator is being subjected to.

Thanks for answering my questions for me, fll-freak!
By AndyC_772
#148322
You risk poor battery life with this circuit. Another major problem with using a potential divider is that a current flows constantly from top to bottom, regardless of what flows through your actual circuit. This is just wasted power, and it'll drain your battery.

Two regulators (or a single dual-output device) is a much better solution. You need low dropout, so you can still maintain 5V even when the coin cell voltages start to sag, and you need low quiescent current. This is a measure of how much current flows through the regulator to GND as opposed to going through your circuit. Less is better obviously.

Use a Schottky diode rather than a silicon type; you'll get a smaller voltage drop, which means you'll be able to maintain 5V with a lower input voltage.
By Mach3Maelstrom
#148330
@langwadt: Thanks! I actually got a bit lazy and just used the SF EAGLE library for the 3.3V... just wanted to get my point across in my topic via pictures. I have an LM2937ET in my circuit, which has a v. drop of about 0.5V according to the manual.
(Manual link for reference: http://www.ti.com/general/docs/lit/getl ... leType=pdf)

@AndyC_772: That's true; regardless of demands from my circuit, I'm still draining a lot of power to ground while the circuit is switched on. I'll stick to the two-regulator device or I'll search for a single dual-output device that will have less demand on the battery.

Also, excellent suggestion with Schottky diodes. I'll replace them into my design. The typical current being pulled will be about 150-180 mW, so I'll be looking into different battery technologies to see what can provide long(ish)-term usage for me.

@fll-freak: Well, you certainly helped me rethink the circuit and learn enough to do some damage on the bench! (hopefully not literally :D )
#148336
Mach3Maelstrom wrote:... The typical current being pulled will be about 150-180 mW, so I'll be looking into different battery technologies to see what can provide long(ish)-term usage for me.
Current expressed in miliWatts?
By Dave Mueller
#148423
A third option would be to use 2 boost regulators like the ones SFE sells based on the NCP1402 and either use a larger capacity 3V battery, wire 2 smaller ones in parallel, or use each battery to power a separate regulator.
You'll get better life out of the battery for 5V, since even the best Schottky will lose almost .5V allowing little tolerance for even the best LDO to regulate.
By Mach3Maelstrom
#148572
@Valen: I was revising my post a few times before hitting Submit, so I originally had about 30-40 mA. Doh!

@Dave: The 5V source is feeding a quad-op amp IC (plus an inverting circuit for V-) and the 3.3V source is feeding an XBee. Thus, I think that boost regulator will be great for the XBee, but I fear that using it for the op amp network will mess with the signal. The power supply current for my op amp is about 7mA, so it's certainly within range, but if you could share a bit of experience on this one, I'd appreciate it.

Thanks all!