# Choosing the right power regulator for Battery powered designs

I've come across this problem a number of times in my hobby projects, but I am sick of making an educated guess.

I often do microcontroller projects and often want them to be powered by a 3.7V Lipo cell charged by a standard microusb cable. This means input voltage can range from 3.0V to 5.0V and I want an output voltage of 3.3V. With thousands of voltage regulators that can fit those requirements, how do you choose?

I can ask about each specific project to get the right regulators, but I would rather have the knowledge I need to find the right regulators on my own. I'll add edits as I find more answers.

Edit1:

Switching Regulator - This is the only choice if you need to boost voltage. They are the most efficient and produce less heat than LDO's, but produce noise not usable with RF applications including bluetooth and wifi and are generally more expensive. If you want to use this with RF then you will need proper filtering.

LDO - LDO's are cheap and are preferred for RF applications because LDO's produce no EMI interference. They generally have poor efficiency but the efficiency depends on the ratio of the input to output voltage. The closer the two are, the more efficient the output will be. In high current applications they can produce a lot of heat so proper cooling may be required.

Charge-pump - A subset of switching regulators that don't require an external inductor. They generally have worse efficiency than inductor-based switching regulators but are better than LDO's. They can also only output relatively small amounts of current. These should be used when you need better efficiency than an LDO but are constrained on board space.

Hybrid - There are a few companies who make hybrid Switching/ldo regulators. These provide the efficiency of a switching regulator with the noiseless output of an LDO. The downside is there are very view of them and they don't have high output current in most cases. These still produce heat like an LDO.

Quiescent Current - In battery powered applications you need to be concerned about the amount of current the regulator uses to function. The more power it uses, the less battery power is left for your device. LDO's generally have much smaller quiescent current than switching regulators, but since switching regulators have the ability to boost voltage when input voltage drops below desired output voltage, which to use depends on your devices voltage requirements.

• If you must know the specifics, then I am making a project using the STM32F446 using the RN4871 bluetooth module and the LS013B7DH06 display. I need 3.3V and 5.0V rails but most of the power consumption draws from the 3.3V rail. – Nate Dec 18 '16 at 7:32
• Most of the power draw will be drawing from the 3.3V rail. Because RF is involved a switching regulator is likely to cause noise interference. Efficiency is definitely preferred. – Nate Dec 18 '16 at 7:35
• From what I have read, most people believe that having a noiseless power output is more important than efficiency in an RF circuit like mine. Many people with either use an LDO and sacrifice the efficiency or use something like the TPS61130 which uses a switching regulator for desired voltage and incorporates an LDO to reduce noise. Since LDO efficiency is determine by the ratio of input to output voltage this makes an efficient, noiseless voltage output. – Nate Dec 18 '16 at 16:59

Battery powered projects (particularly those with periodic events spaced quite a bit apart) usually benefit from using a linear regulator.

Looking at your requirements (LiPo 4.2V to Vo + dropout voltage) a linear regulator will be (on average 3.7V battery, regulated output 3.0V) 81% efficient which is close to the SMPS solution anyway.

A switch-mode device uses a lot of power (relatively speaking) just to power itself up; it is overall more efficient where a relatively large load current is being used.

What you need to look at carefully is $I_q$ for the regulator; this is the amount of current it draws simply to power the internal circuitry.

The regulator with the lowest $I_q$ that can handle the input and output voltages and required current is usually the best, but temperature rise is something to be considered - the small devices cannot dissipate much heat (there simply is insufficient physical packaging to permit it).

As most modern microcontrollers and many interfaces will run quite happily at 3.0V or even 2.7V, you should consider using that as the regulated voltage. If you really want Vo = 3.3V for Vin 3.0V <> 4.2V you would need a buck-boost or some such arrangement. I would stick with a simple LDO regulator and drop the regulated voltage.

Note that regulators designed for large load currents (not the usual case for battery operated projects) will typically have a relatively high $I_q$

The most astounding part I have ever seen for $I_q$ : TPS783xx

Compare that with a buck SMPS that has a relatively low $I_q$ (for an SMPS device).

There are other considerations, but the no-load power dissipation of the regulator needs to be kept to a minimum.

Incidentally, a Li+ / LiPo can be safely discharged (in my experience) to about 2.7V

• Fair enough but if I used something like the tps61130 I get 90% efficiency and regulated 3.3V output for input voltages in the 2.7V-3.3V range. It still uses an LDO, so heat may still become a problem if I pack things in tightly but don't the benefits more than make up for the increase in quiescent current? – Nate Dec 18 '16 at 18:15
• Even if I chose to use an LDO, how do I find the right LDO? There are so many to choose from. – Nate Dec 18 '16 at 19:03
• The key is to find a solution that works in specific circumstances. This is a guideline. I can put you in touch with FAEs from all the usual suspects who can guide you for a specific project if necessary, – Peter Smith Dec 18 '16 at 19:59
• No that is not necessary. This is a hobby project and I doubt they would invest the time to discuss my particular needs. I appreciate your help. – Nate Dec 18 '16 at 21:30

It is not recommended to drain a lipo below 3.0V.

If you stayed above 3.3V you could just use a linear regulator if you don't mind efficiency too much. Otherwise, if you really want to go as low as 3.0V or be more efficient, you have to use a buck/boost converter. They usually have a very high efficiency of around 90%.

The following pdf has a design example on page 10/11: Uout = 3.3V Uin_min = 2.6V Uin_max = 5.5V Imax = 2A http://www.ti.com/lit/an/slva535a/slva535a.pdf

If you use proper filtering the switched power supply should not be a problem for your bluetooth module (or do you have any other RF circuits on your board?)

• I kinda figured that this was all true. I'll update the voltage range. For the sake of this question though I'm not marking this as answered. I'm looking for a general set of guidelines for finding the right power regulators for the job. – Nate Dec 18 '16 at 16:30
• btw dave jones from the eevblog just published a video on how to measure the efficiency of a dc/dc converter: youtube.com/watch?v=li0XKnpOZyM – Frode Akselsen Dec 19 '16 at 9:58

Yes it is possible for switching power supplies to generate noise, but at low power levels filtering the output is not that hard. I am using switching power supplies in a low-power radio appliation and I am having very good results. With just an RLC filter on the output of the regulator I don't see any RF interference problems.

The regulators maintain the output voltages while only drawing micro amps of quiescent current when the board is put to sleep.

The regulators work with inputs from 1.8V to 5.0V and can automatically switch between step-down and step-up mode making 2.2V and 3.3V used on my board.

The supplies I am using are from Texas instruments. Go to www.ti.com. They have a power supply design tool called web-bench that will design power supply solutions for you within minutes.