# Which kind of AA battery should I use to power up my ESP-32

Disclaimer : I have few knowledge in electronics

I have an ESP-32 device that use 3.3V to run perfectly. The 3.3V and the Ground of the device is directly connected to an AA battery container with 2 slots.

The device is a homemade sensor that use an ESP-WROOM-32 and some additional component like water detection, humidity, temperature sensors. I don't have the tools to change it. (Datasheet of ESP-WROOM-32)

Until now, I powered up my ESP-32 with 2 Alkaline Battery at 1.5V each. I was able to run it but it didn't last long (Near 20 minutes, not including deep sleep periods).

What kind of battery should I use to power up my ESP-32 and last more than 20 minutes ?

• Maybe C-size? Or D-size? No? – Ale..chenski Aug 29 '18 at 18:41
• My device has already the container with 2 AA Slots. I can't change it. – Gazouu Aug 29 '18 at 18:48
• There is something wrong. AA alkaline batteries typically have capacity of no less than 2000 mAh. Which means delivering 2 A for an entire hour. If you depleted the batteries in 1/3 of an hour, it means that your device is consuming something like 6 A, which is very-very strange. The ESP32 consumes 240 mA typical at highest power, and the AA batteries should last several hours, unless you have some unjustifiably high voltage cut-off. – Ale..chenski Aug 29 '18 at 18:48
• Two AAs, even new, will not deliver 3.3v. 3.2v off load is reasonable when pristine, falling quickly on load or with use. What is your endpoint voltage? To get the quoted life out of them, you need to run them down to 900mV, or 1.8v on the ESP. What's the minimum voltage it will perform to specification at? If it will work at 2.4v, then a couple of NiMH AAs would be better, they have a much flatter voltage curve as they run down. – Neil_UK Aug 29 '18 at 18:54
• are you using alkaline, or "heavy duty" cheapo zinc batteries? – Passerby Aug 29 '18 at 19:32

It looks like the ESP-WROOM-32 module has some problem with accepting wide (dropping) range of typical AA alkaline batteries. Here is a potential solution for this problem, a KENTLI battery. This battery uses a Li-Ion 3.7 V battery inside, and has a DC-DC regulator to 1.5V, thus emulating a regular AA alkaline cell. This combination should produce stable 1.5-V level until the internal battery discharges. They claim total capacity of 2800 mWh (equivalent to 1900 mAh), and it is worth to try if the ESP32 module works down to nominal 1.5 x 2 = 3.0V.

Here are the KENTLI marketing data on batteries,

• Is the DC-DC regulator to 1.5V only present on Kentli battery ? Or if I buy any Li-Ion 3.7V Battery it would work the same ? – Gazouu Aug 29 '18 at 20:06
• I just ordered 4xAA Kentli (+ their charger) from eBay for my own curiosity... – Ale..chenski Aug 29 '18 at 20:22
• @Gazouu You can buy one 14500 Li-ion cell, and use the other AA spot to built a DC-DC switching regulator, from 4.2-3.5 V to 3.3 V. For more information on low-voltage conversion, see this exchange, electronics.stackexchange.com/a/392808/117785 . Or you can get two 14500 cells (making 8.4 to 7V range), and make a DC-DC regulator down to the same 3.3V if you have additional space. But it would require some decent engineering. The KENTLI batteries solve your problem immediately. – Ale..chenski Aug 29 '18 at 20:41
• Interesting replacement battery. A bit too pricey. A NiZn battery would be much more economical and has a great discharge curve for a 3V application. amazon.com/NiZn-2500mWh-Rechargeable-Battery-Camera/dp/… --- Discharge curve: cdn-reichelt.de/documents/datenblatt/D600/ANS_NIZN_4XAA_DB.pdf – Misunderstood Aug 30 '18 at 22:19
• @Misunderstood, I agree, these NiZn cells look very good for the application, thanks. High discharge too. I probably need to order few as well. – Ale..chenski Aug 30 '18 at 22:35

It looks like the esp8266 has an input voltage range of 2.5v to 3.3v, and it is fairly power hungry, especially with the radios and peripherals all turned on. There are a couple of things you can do to extend the runtime of the device, namely:

1. Turn off the peripherals when you aren't using them. From the data sheet, the wifi uses between 50 and 200mA when it's on, which is quite a lot of power. You didn't mention what you're doing with the device, but if for instance you're making a data logging application which uploads some data to a server once a minute, then there's no reason to have the radio powered up all of the time.
2. Similar to point 1, if the micro controller isn't doing anything, there's no reason to keep it on. According to the data sheet, the micro controller uses 15mA by itself, but putting it into a light sleep uses 0.9mA, and putting it into a deep sleep uses $20\mu A$. Again taking the example above, there's not much of a reason to have the microcontroller powered on in between uploads of data, so if you can shut it off, do so.
3. Use a larger battery. Pretty self explanatory, using a larger battery (such as C or D cells) in place of the AA cells will give you more runtime. Additionally, you may want to consider using a lithium battery (with protection!), so when the battery does die you can recharge it instead of throwing it away.
4. Use a voltage regulator. Part of the problem you're having is that fresh alkaline batteries provide 3.0V, and by the time you've discharged them to 2.5V the device shuts off. However, the batteries still have a lot of energy left, which can't be used by your device. To fix this, you can use a buck (step-down) converter, and use a higher voltage battery pack. For example, if you instead use a 6V battery pack (4 cells), and a buck converter that brings the voltage down to 3V for the micro, you can run the cells down to 3V before the microcontroller gives out. Alternatively, you could use a buck/boost converter which can convert the input voltage to a lower or higher voltage. Using this, you may even be able to run your 6V battery down to 1.5V, or run your 3V pack down to 1.5V, giving you more runtime.
• I send data to my server every 15 minutes. And between these data transfer my ESP is in deep sleep mode as you said. For the point 3. / 4. I can't really change the hardware. The device is all in one and already have this AA battery container inside. – Gazouu Aug 29 '18 at 18:46
• If your ESP (including the WiFi!) is in sleep mode for all but a short period every 15 minutes you should get much longer than 20 minutes from two AA's. Check your current consumption during sleep. – Wouter van Ooijen Aug 29 '18 at 18:48
• I don't count the deep sleep time in the 20 minutes. It last for less than a day including the deep sleep time – Gazouu Aug 29 '18 at 18:52
• In that case it is math time. What is are contributions to the total power consumption from the active and sleeping periods? If active, try to reduce the active period (you probably can't do much about the active current). If it is the passive period, try to reduce its power (sleep deeeper). – Wouter van Ooijen Aug 29 '18 at 18:57

Since the ESP32S module is specified at 3.0 - 3.6V operation you absolutely need a battery with a voltage over 1.5V.

The Expressif design guide is worth reading, the implication is that the cpu will work down to 1.8V ….but not necessarily the whole module. For example the Flash is rated at a min 2.7V.

The only choice you have for a single cell is to use two Nickel-zinc (NiZn) batteries with a terminal voltage of 1.65V.

Nickel-zinc (NiZn)

Nickel-zinc is similar to nickel-cadmium in that it uses an alkaline electrolyte and a nickel electrode, but it differs in voltage; NiZn provides 1.65V/cell rather than 1.20V, which NiCd and NiMH deliver. NiZn charges at a constant current to 1.9V/cell and cannot take trickle charge, also known as maintenance charge. The specific energy is 100Wh/kg and can be cycled 200–300 times. NiZn has no heavy toxic materials and can easily be recycled. Some packaging is available in the AA cell format.

In 1901, Thomas Edison was awarded the U.S. patent for a rechargeable nickel–zinc battery system that was installed in rail cars between 1932 and 1948. NiZn suffered from high self-discharge and short cycle life caused by dendrite growth, which often led to an electrical short. Improvements in the electrolyte have reduced this problem, and NiZn is being considered again for commercial uses. Low cost, high power output and good temperature operating range make this chemistry attractive.

Source: Battery University

You must take care when charging, since the terminal voltage can rise above 1.8V. If you fully charge them you will exceed the rating of your module. However you can charge them to a 1.8V terminal voltage without problem.

Discharge Curve

• Not an easy to find battery but you are correct. A NiZn cell has the perfect discharge curve for this application. cdn-reichelt.de/documents/datenblatt/D600/ANS_NIZN_4XAA_DB.pdf – Misunderstood Aug 30 '18 at 22:09
• They appear to be readily available on Amazon. – Jack Creasey Aug 30 '18 at 22:39
• I could not find a name brand supplier. I do not like buying off brand batteries from the Amazon and eBay charlatan battery vendors. But the OP has very limited choices being locked into AA. – Misunderstood Aug 30 '18 at 22:58
• Rayovac make a range, but they are relatively new in the market. – Jack Creasey Aug 30 '18 at 23:07

With an ESP-WROOM-32

If this is the ESP-32 you are using,
you must supply at least 2.7V.

And you need a minimum of 500 mA.

Two alkaline batteries cannot supply more than 2.7V @ 500 mA for very long.

If you run near 500 mA while transmitting the battery capacity is reduced to 500mAh.

So if, while transmitting, you are drawing 500 mA, you will only get about 20 minutes according to the discharge curve (shown below) for a a pair of alkaline AA batteries. Your cutoff is 1.35V (shown in red cross hairs below) which is in the steep part at the start of the discharge curve.

You cannot use alkaline AA batteries to power this device for more than 20 minutes if it is drawing 500 mA and 2.7V is the minimum.