When would I use a voltage regulator vs voltage divider?

When would you use a voltage regulator vs a resistor voltage divider? Are there any uses for which a resistor divider is particularly bad?

-

These two circuits types have very different applications.

A resistor divider is generally used to scale a voltage so that it can be sensed/detected/analysed more easily.

For example, say you want to monitor a battery voltage. The voltage may go as high as 15V. You are using a microcontroller's analog-to-digital converter ("ADC"), which is using 3.3V for its reference. In this case, you may choose to divide the voltage by 5, which will give you up to 3.0V at the input of the ADC.

There are a couple of drawbacks. One is that there is always current flowing through the resistors. This is important in power-constrained (battery powered) circuits. The second problem is that the divider can't source any significant current. If you start drawing current, it changes the divider ratio, and things don't go as planned :) So, it's really only used to drive high-impedance connections.

A voltage regulator, on the other hand, is designed to provide a fixed voltage regardless of its input. This is what you want to use to provide power to other circuitry.

As far as creating multiple voltage rails: For this example, let's assume that you are using switching regulators that are 80% efficient. Say that you have 9V, and want to produce 5V and 3.3V. If you use the regulators in parallel, hooking each one up to 9V, then both rails will be 80% efficient. If, however, you create 5V and then use that to create 3.3V, then your 3.3V efficiency is (0.8 * 0.8) = only 64% efficient. Topology matters!

Linear regulators, on the other hand, are assessed differently. They simply lower the output voltage, for any given current. The power difference is wasted as heat. If you have 10V in, and 5V out, then they are 50% efficient.

They have their benefits, though! They are smaller, less expensive, and less complicated. They're electrically quiet, and create a smooth output voltage. And, if there isn't much difference between the input and output voltages, then the efficiency can top a switching supply.

There are ICs which provide multiple regulators. Linear Tech, Maxim Integrated, Texas Instruments, all have a good selection. The LTC3553, for example, provides a combination of a Lithium battery charger, a switching buck regulator, and a linear regulator. They have flavors with or without the charger, some with two switchers and no linears, some with multiple linears...

One of my current products uses a 3.7V battery, and needs 3.3V and 2.5V. It was most efficient for me to a linear for the 3.3V, and a switcher for the 2.5V (fed by the battery, not the 3.3V rail). I used the LTC3553.

You'll want to spend some time on their respective website's product selector tools.

Good luck!

-
I think it's worth mentioning that your discussion of efficiency with multiple supply rails applies only to switching regulators and not to linear regulators. – Joe Hass Apr 16 '14 at 10:47
@joehass Thanks, I added it! – bitsmack Apr 16 '14 at 13:39
"the divider can't source any significant current" Why is this the case? – kmort Apr 16 '14 at 17:24
@kmort Imagine you are dividing 10V down to 5V. You use two 500-Ohm resistors to do the division. So, now, you have 10(V)/1000(Ohm) = 10mA flowing through the divider. Now, add your load. This load goes in parallel with the bottom resistor, which skews the resistor divider calculations, and changes the voltage ratio. If your load is fixed, then you can calc the adjusted divider values. A good rule of thumb is to draw less than 10% from the divider's center node, so you don't perturb the ratio very much. But now, you are using 10x your required current just through the divider! – bitsmack Apr 16 '14 at 17:48
@bitsmack Yes, makes great sense. I should have thought about that a bit more. Thanks for your help. :-) – kmort Apr 16 '14 at 18:29

Since a voltage divider does not regulate, one would not want to use a voltage divider when one wants a regulated voltage.

A voltage regulator will, within its limits, maintain the output voltage at a fixed value even as the input voltage and load current varies.

A voltage divider will not do this. Consider the voltage divider equation:

$$v_{OUT} = v_{IN}\frac{R_2}{R_1 + R_2} - i_{OUT}\cdot R_1||R_2$$

which is manifestly dependent on $v_{IN}$ and $i_{OUT}$ so a voltage divider is not a voltage regulator.

There are, however, plenty of applications for voltage dividers, e.g., attenuation, but voltage regulation is not one of them.

-

A voltage divider is particularly bad at providing a fixed voltage to a variable or low-impedance load. Variable loads are quite common, and include most digital circuits on the planet.

Fixed, high-impedance loads can have a voltage divider in front of them. This is the case when using an ADC to measure or a comparator to fence a much larger voltage, or in the sense input of a voltage regulator.

-
So if I have a board where I need to power both 5v and 3.3v logic it is probably better to just have two regulators one for each voltage instead of trying to power the 3.3v off a resistor voltage divider? – Pete Apr 16 '14 at 3:07
Ideally you would have one voltage regulator that would provide both voltages, but having two regulators is better than having any number of voltage dividers to do the same task. – Ignacio Vazquez-Abrams Apr 16 '14 at 3:09
Do you have an example of a part number off hand that can provide dual voltages? – Pete Apr 16 '14 at 3:11
Nope. – Ignacio Vazquez-Abrams Apr 16 '14 at 3:21
@Pete Ha! Just in case you didn't notice, Ignacio's "Nope" is a link to TI's product finder :) – bitsmack Apr 16 '14 at 4:09

Voltage dividers are not usually used to generate supply voltages because they provide no regulation. Many loads will alter their output voltage anyway, for example a resistive load to ground is essentially in parallel with R2.

Voltage dividers are usually used to provide a voltage to a high impedance input. In this case you can think of impedance as being basically the same as resistance. Having a 10M resistor in parallel with R2 won't affect it much, as long as R2 is itself orders of magnitude lower like say 10k. Of course, using low value resistors for the divider also increases the current flow through it, so cause issues for battery powered devices.

A common example of a voltage divider into a high impedance input is to divide down a high voltage to a range that an ADC can measure. Say your ADC has a 1V reference and you want to measure a 3.6V battery with it. You might use a 4:1 divider to scale that down so it is less than 1V and measurable by the ADC.

Another common example is to provide a secondary reference voltage. Say you have a 3.6V supply and need a 1.8V reference (half the supply voltage, e.g. for biasing an AC signal with a DC offset). Rather than bothering with an expensive voltage reference IC you could simply use a voltage divider to halve the supply voltage and feed that to an op-amp buffer. The op-amp has a high impedance input, and the output can be used for biasing.

A regulator can provide a certain amount of current into a load, with the voltage controlled as best it can be, so is suitable for supply voltages and the like.

-