http://opalkelly.com/

Author Topic: Low power LiPo project - LDO vs buck converter  (Read 258 times)

0 Members and 1 Guest are viewing this topic.

Offline drogus

  • Contributor
  • Posts: 9
  • Country: pl
Low power LiPo project - LDO vs buck converter
« on: July 18, 2017, 02:07:26 AM »
Hi,

I'm working on a bike alarm project and I would like to run it on a LiPo and possibly not have to recharge it for at least a few months. I'm planning to use Atmega32u4 (for simplicity of USB programming if I have to do updates) and an accelerometer with an interrupt feature. That way the MCU would be mostly sleeping and only activated when someone moves the bike. That's why I'm assuming current consumption of about 10-20µA (about 5-10 for Atmega32u4 and another 5-10 for an accelerometer).

I was planning to power it directly from a battery to avoid losses, but it seems that all accelerometers that fit my needs need about 2-3.6V range, which doesn't fit the battery. That led me to idea of dropping the voltage down to 3V (as LiPo is dead at 3V anyway). I'm considering 2 options:

1. A buck converter, for example based on TPS62736 (it has a sub µ quiescent current and a decent efficiency at light loads)
2. An LDO, for example MCP1700

According to my calculations, an LDO would actually have lower losses (because for most of the time it would need to only drop from ~3.6V to 3.0V and quiescent currnet is really low), but I'm not quite certain about one thing. The datasheet says: "The minimum VIN must meet two conditions: Vin >= 2.3V and Vin >= (Vr + 3.0%) + Vdropout.". I'm not sure what happens when the Vin drops below the specified level, will it just stop working? If that's true, than it seems like a good option, typical Vdropout is 178mV for V>2.5V and even maximum value (350mV) is not too scary - I would be loosing only about 20% of battery, worst case.

TPS62736 on the other hand could work across the full battery range and still gives a good efficiency, but it will quite possibly have a higher ripple on the output.

Now the question is - knowing these tradeoffs, what would you choose for this project? Or maybe there's yet another way that I'm missing?
« Last Edit: July 18, 2017, 02:43:02 AM by drogus »
 

Offline Jeroen3

  • Super Contributor
  • ***
  • Posts: 2256
  • Country: nl
  • Embedded Engineer
    • jeroen3.nl
Re: Low power LiPo project - LDO vs buck converter
« Reply #1 on: July 18, 2017, 02:54:59 AM »
Do not forget the environment: Winter cold -15* to baking in the summer sun +60C.

*I think, does Poland get colder?
 

Offline drogus

  • Contributor
  • Posts: 9
  • Country: pl
Re: Low power LiPo project - LDO vs buck converter
« Reply #2 on: July 18, 2017, 03:06:46 AM »
Do not forget the environment: Winter cold -15* to baking in the summer sun +60C.

*I think, does Poland get colder?

Good point! I live in Berlin now, so it doesn't get as cold as it can get in Poland, but it can get quite cold. As far as I can see on the discharge curves for LiPos -20*C would cut the battery life in half. That said, it still should be fine if I can get lower than 50µA during sleep. I was planning to use 2Ah battery, so even if the bike alarm goes off a couple times (the 110dB buzzer that I found is ~2W) and even if Atmega needs to wake up from time to time (I plan to wake it up on motion and then be active for some time to check if the alarm should go off), it should easily give me months of battery life.
 

Offline drogus

  • Contributor
  • Posts: 9
  • Country: pl
Re: Low power LiPo project - LDO vs buck converter
« Reply #3 on: July 18, 2017, 07:36:40 AM »
I had a bit more time to read on LDOs and I learned that different types behave differently with input voltage lower than output voltage + dropout voltage. I found this answer on stackexchange, which mentions that the output voltage in  MOSFET based LDOs will start dropping when the input voltage is too low. I haven't found a corresponding chart in MCP1700's datasheet, but I went through the MCP1703's datasheet and I found this graph:



I think that it would work similarly for MCP1700, because they're both based on MOSFETS and they seem very similar.

In that case I'm leaning towards the linear regulator now, for the following reasons:

- hand soldering QFN will probably be a pain in the ass and TPS62736 is only in QFN package (and I can't find any buck converters other than TPS62736 that seem good for the task). I want to try it some day, but maybe not on this project
- my project will work down to 2.7V (I'll be running Atmega32u4 at 8MHz), so I should be fine for the entire battery voltage range (or almost entire range, but I will probably want to recharge much sooner than any problems would occur anyway, to leave power for the time when the alarm needs to turn on)
- no need for extra components
- MCP1700 should be more efficient than TPS62736 in this case
- MCP1700 is 6 times cheaper than TPS62736, not even counting the extra inductor and bigger capacitors
 

Offline drogus

  • Contributor
  • Posts: 9
  • Country: pl
Re: Low power LiPo project - LDO vs buck converter
« Reply #4 on: July 18, 2017, 04:40:14 PM »
One more thing that I found is that there're Arduino inspired boards using MCP1700 on the market, for example Moteino. They're using it to step down to 3.3V, but in my project I don't really care if it's 3.3V or 3V and I also don't mind lower voltages (or at least I think I don't mind ;) ). If I need a reference steady voltage I was thinking about using a reference voltage IC or the MCP1700 itself, but with a lower output voltage (so use  the 3V version for supplying voltage to most of the components and use something like 2.5V or 1.8V for a reference).
 

Offline MrAl

  • Frequent Contributor
  • **
  • Posts: 308
Re: Low power LiPo project - LDO vs buck converter
« Reply #5 on: July 18, 2017, 08:28:44 PM »
Hello,

Yes LDO regulators can be more efficient than bucks but that is when the input output differential voltage is low.

For example, if you wanted 10v out and you had 11v in, an LDO that could work with 1v differential would provide approximately 90 percent efficiency, which is pretty good considering that a lot of bucks are only 80 percent.

On the other hand, if you had 12v in and 10v out then the efficiency would drop to roughly 80 percent, and if you had 13v in and 10v out it would drop down to roughly 70 percent.

You can do more exact calculations if you wish, but you see that it is very dependent on the input output differential voltage, and of course the particular LDO has to be able to work with a low differential voltage.
 

Online Gaurav

  • Regular Contributor
  • *
  • Posts: 120
  • Country: de
Re: Low power LiPo project - LDO vs buck converter
« Reply #6 on: July 18, 2017, 11:53:12 PM »
Hi,

1. A buck converter, for example based on TPS62736 (it has a sub µ quiescent current and a decent efficiency at light loads)
2. An LDO, for example MCP1700

Now the question is - knowing these tradeoffs, what would you choose for this project? Or maybe there's yet another way that I'm missing?

just to add another part to your list TPS709 available in various voltages have 1-?A  Quiescent current  , i have tested this regulator in my outdoor temprature transmitter project .
MCU core is sleeping most of the time. Timer activate MCU every minute , RF module send RF packet and everything goes back to sleep wating for next timer interrupt. i have charged prototype battery once and it is still running on first charge. since march.

i have used MSP430 FRAM MCU , they are quite low power.

 i have capture few quite nice plots of current consumption using Oscillosopea and current sens register .

http://www.circuitvalley.com/2017/03/wireless-outdoor-tempreature-sensor-msp430-NRF24F01-DS18S20.html
 

Offline drogus

  • Contributor
  • Posts: 9
  • Country: pl
Re: Low power LiPo project - LDO vs buck converter
« Reply #7 on: July 19, 2017, 03:06:07 AM »
Thanks for more info MrAl!

just to add another part to your list TPS709 available in various voltages have 1-?A  Quiescent current  , i have tested this regulator in my outdoor temprature transmitter project .

Thanks, it looks interesting. What output voltage are you using? The datasheet defines typical dropout voltage at 295mV and maximum dropout voltage at 650mV for a 3.3V version, so it seems that you may get voltages lower than 3.3V when battery drops lower than 3.6V. Have you checked how it behaves in practice? Because I think I'll be in a similar situation.

i have used MSP430 FRAM MCU , they are quite low power.

I've came across this MCU some time ago and wanted to give it a try as it looks very promising for low power applications, but I don't want too many new things in this project, so I can hopefully finish it soon.

i have capture few quite nice plots of current consumption using Oscillosopea and current sens register .

http://www.circuitvalley.com/2017/03/wireless-outdoor-tempreature-sensor-msp430-NRF24F01-DS18S20.html

Very interesting post, thank you!
 

Offline drogus

  • Contributor
  • Posts: 9
  • Country: pl
Re: Low power LiPo project - LDO vs buck converter
« Reply #8 on: Today at 04:00:55 AM »
Just in case anyone is reading this and looking for answers for similar questions, there's one more thing that I missed in my research. The dropout voltage depends on load current. For example this is a chart for MCP1700:



So for low power purposes the dropout voltage can be pretty much ignored, which is a very good news for me as now I'm certain that using 3.0V version of MCP1700 I will get the entire LiPo range.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf