2
\$\begingroup\$

I have a project which uses microcontrollers and a radio transmitter, both running at 5 V. The circuitry consumes less than 1 mA from the 5 V rail when idle, and 1 A when transmitting data using the radio transmitter. The transmission phases do not happen often, as the device spends most of the time in idle mode, usually transmitting only once every 24 hours for a few seconds (maximum 20 seconds).

Now I'm designing a battery input for the circuitry and I cannot decide whether to use 11 V battery packs (LiPo) with switching regulators, or LiFe batteries instead where I can get a low regulating voltage difference between input and output for the regulator (output of 6 to 7 V) and use a linear regulator with those lower voltage batteries.

Which of these setups gives more efficient results for regulation circuitry in this kind of use?

  • switching regulator with 11 V input, dropping the voltage down to 5 V; or

  • a linear regulator at (mostly) low current and maximum of 2 V voltage drop over it?

One of the main interests in addition to the battery life, is the simplicity of most linear regulators versus switching regulators, which reduces time spent on designing the product and saves space from the board. Switching regulators can also cause some radiation that might interfere with other circuitry and they are also more expensive.

\$\endgroup\$
  • \$\begingroup\$ What microcontroller? One of the attractive features of the AVR series is that they are happy running directly from an unregulated "3.7V" LiPo output. \$\endgroup\$ – chrylis Aug 1 '16 at 19:11
  • \$\begingroup\$ @chrylis We are currently using pic microcontrollers and they would run also with that voltage but the transmitter circuitry requires that 5 volts. \$\endgroup\$ – The amateur programmer Aug 2 '16 at 4:25
3
\$\begingroup\$

Interesting question because I can see the answer going either way, depending on environmental circumstances.

Some aspects of the design which may not be immediately obvious...

1) Switchers are notoriously poor at a tiny fraction of their load - they may consume several mA internally, or they may lose regulation and deliver 7V below some value, say 1% of rated load (10mA in your case) without special care in design.

2) One answer could be a linear regulator during sleep, (even from 11V but there's nothing wrong with a 2S Li-Ion - nominally 7.4V max 8.4V) and the MPU has to wake up a switcher before transmitting. If the linear regulator only supplies a few mA, you can probably find a SOT-23 to do the job, or SOIC-8 at the largest, so I don't believe size is the issue

3) A linear regulator for 1A will need some heatsinking even for 20 seconds ... if there's a convenient chunk of metal, use it. Linear may be more reliable from its simplicity. But what happens if the TX gets stuck "on"? Running the battery flat is one thing, destroying the equipment is another... .

4) I would not, personally, change battery technology simply as a way of tuning supply voltages. If you need lower fire hazard, or greater charge/discharge cycles, or some characteristic of LiFePO4 that's a reason for using them - otherwise stick with commodity batteries for economics and simpler servicing. .

\$\endgroup\$
  • \$\begingroup\$ Yes, I have definitely noticed the the low current draining of power when using buck converters for regulating the input for microcontrollers. They can easily use more power than the microcontroller it self. \$\endgroup\$ – The amateur programmer Aug 1 '16 at 11:16
  • \$\begingroup\$ I think that I'll test out some linear regulator setups and try using the dual regulator also to see the difference between efficiency and if it's big enought difference that it's worth the extra regulator. Otherwise I'm trying to go with the linear regulator to keep things simple. \$\endgroup\$ – The amateur programmer Aug 1 '16 at 11:34
5
\$\begingroup\$

I would estimate that your most efficient solution is to use a low power regulator for your microcontoller and let the microcontroller activate a buck regulator to feed power to the radio transmitter. This will mean that in idle mode only the low power regulator's current consumption is discharging the battery.

Of course this means an extra IO line to "enable" the higher power regulator AND some shortish period of "waiting" whilst the radio becomes "ready" for the data transmission from the microcontroller.

The problem with NOT enabling/disabling the higher power regulator (feeding the radio) is that its quiescent current consumption might be hundreds of micro amps or even low milli amps and this will certainly deplete the battery.

\$\endgroup\$
  • \$\begingroup\$ Using two regulators would increase the design complexity too much and require additional board space as well. In this project we also have some size constraints. \$\endgroup\$ – The amateur programmer Aug 1 '16 at 11:13
  • 2
    \$\begingroup\$ Well, a tiny linear regulator that feeds the micro will be... tiny of course because the micro's current will be so very much smaller that the regulator required for the radio. You asked for "optimum performance", "battery life" and "efficient results"! \$\endgroup\$ – Andy aka Aug 1 '16 at 11:22
  • \$\begingroup\$ Well that might be good design after all, I have already experimented with one TI buck converter which has an "enable" pin which enables the chip when driven low (perfect for using with one of the open collector i/o pin on the controller). \$\endgroup\$ – The amateur programmer Aug 1 '16 at 11:25
  • \$\begingroup\$ You can easily get a SOT-23 regulator for the micro if it only needs a miliamp. \$\endgroup\$ – pjc50 Aug 1 '16 at 12:35
  • \$\begingroup\$ Don't have the rep here to make such a small edit, but "microcontoller" in the first sentence should be "microcontroller" with an extra "r". As far as I'm concerned, feel free to flag this as obsolete after making the edit. \$\endgroup\$ – a CVn Aug 1 '16 at 16:59
2
\$\begingroup\$

Look at the datasheets. This really should have been obvious.

The datasheets of linear regulators will tell you the quiescent current. The datasheets of buck regulators will tell you the quiescent current, and give some guide to likely efficiency. From these, you can figure out the overall efficiency and average power draw from the battery.

You also need to do some basic math. One obvious thing to determine is whether your biggest problem is the occasional but high power RF transmission, or the constant but low power idle current. There are 86,400 seconds in 24 hours. (1 A)(20 s)/(86,400 s) = 230 µA. That's the average current draw due to the radio transmissions. This means the 1 mA idle current dominates the total by over a factor of 4.

There is no substitute for looking at a few plausible alternatives and doing the math to see which one is more optimal. However, my hunch is a buck switcher with good idle characteristics. This would be something that has PWM/PFM switchover capability. Put another way, it doesn't just change the lengths of switching pulses at a fixed frequency, but at low power lengthens the time between pulses too.

\$\endgroup\$
  • \$\begingroup\$ Thinking that "average current" way it would make sense to use the linear regulator setup. I'll have to dig around some of the regulator's datasheets. \$\endgroup\$ – The amateur programmer Aug 1 '16 at 11:12
  • 1
    \$\begingroup\$ Another idea would be a constant-on-time hysteretic buck converter -- these give you the good idle characteristics "for free" along with wickedly fast load transient response \$\endgroup\$ – ThreePhaseEel Aug 1 '16 at 11:41
  • \$\begingroup\$ @Three: Yes, basically any pulse on demand scheme will have good low power characteristics. Your "constant on time" is basically PFM mode. I don't see the need for hysteresis though. The advantage of a PWM/PFM chip is when high power is demanded, it goes to a high switching rate, but then has the capability of varying the pulse width instead. This is better for efficiency and other characteristics at the high power end. There are chips that do all this. All you need to do is hook up the inductor and a few other parts. \$\endgroup\$ – Olin Lathrop Aug 1 '16 at 11:48
0
\$\begingroup\$

With 7V input and 5V output the linear regulator will have to drop 2V, thus achieve an efficiency of about 62%, no matter the load.

Switching regulators have very good efficiency (>90%) at high loads, but will be very bad at lower loads. You’d have to look up the datasheets for real numbers but at a fraction of their maximum load they can easily drop below 50%. Their quiescent currents are often also quite high. There are some good switching regulators out there, you might find one which fits your requirements perfectly.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.