Solar harvesting into Lithium Ion Capacitor

A tiny board with AEM10941 solar harvesting IC charging a Lithium Ion Capacitor

Similar projects worth following
Starting from
jaspersikken has 727 orders / 25reviews
Ships from Netherlands
This board was one of the 2021 Earth Day Challenge winners but now for the 2022 Hackaday prize I am resubmitting it to the Planet-Friendly Power challenge because I have significantly improved it. The poor 2.2V/80mA output only supported low power low voltage applications like BLE and LoRa sensors. Now with the new 3.3V/400mA output also NB-IoT, LTE-M and wifi applications can be made batteryless. I am so excited about Lithium Ion Capacitors because it combines many of advantages of supercapacitors and batteries. The AEMLIC V2 is a tiny board with the AEM10941 Solar Harvesting IC from E-peas. It efficiently converts solar panel energy into LIC charge, it even works with indoor light. The 3.3V output is enabled when the LIC has sufficient charge, and it has a low voltage warning that informs the user of impending shutdown when the LIC runs low. It easily integrates in other projects because of the castellated via's, and when soldered onto 0.1' pitch headers it fits in a bread

Solar harvesting into Lithium Ion Capacitor

This board is similar to my Solar Harvesting into Li-ion and Solar Harvesting into Supercapacitors board, but this one is designed to store its energy in a Lithium Ion Capacitor. 

Recently (2021) the price of Lithium Ion Capacitors (LICs) from manufacturer Vinatech dropped in price because they opened new factories. For example this 250F is less than 5USD in qty50 and I even got a quotation for USD1.90 in qty1000. 

The AEM10941 is a highly efficient solar energy harvesting chip from E-peas. I configured it to keep the LIC between 2.5 and 3.8V using R1-4. 

It also has an integrated 80mA LDO (HVOUT) that I configured to 2.2V using R5-6. 

The AEM10941 has a low voltage warning (STATUS1 pin) that informs the user of impending shutdown when the LIC runs low, which is 600ms before the LDO is turned off.

SInce the HVOUT can only supply 80mA, I added a TPS63900 buck boost converter, it is configured to 3.3V and can deliver 400mA. The TPS63900 is enabled by the HVOUT signal. I love the TPS63900 because it is super low power and tiny. The complete schematic looks like this

Then I made PCB layout and ordered bare PCBs from Elecrow

I needed correct the voltage on a pull up resistor and I succesfully completed the hardware design verification. 

Since I have much experience with AEM10941 and tested TPS63900 extensively I immediately ordered 50 PCB assemblies from Elecrow. And they all look good

I created a test rig and if all LEDs light up the complete circuit works. This way I functionally tested the 50 boards.

Then I created a cool 3 minutes video to make everyone excited about Lithium Ion Capacitors and my board.

And then updated my Tindie product page and added stock. 

  • Testing few big Lithium Ion Capacitors

    Jasper Sikken11/07/2022 at 21:02 5 comments

    Recently I bought a 250F and a 750F 3.8V Lithium Ion Capacitor from Aliexpress, it is manufactured by Cda. Are they good?

    250F 3.8V

    750F 3.8

    I compared it to other LICs from well know brands like Vinatech and Eaton. I compared datasheets, measured capacitance and measured leakage. To measure the capacitance I used my own electronic load set at a constant current of 500mA and and measured the time it takes to drop from 3.8V to 2.5V. Capacitance is calculated Q(F)= )=t(s)*0.5A / (3.8-2.5)

    To measure the leakage I charged them up and every day I measured the voltage decay with a bench  multimeter and calculated leakage current using following formula leakagecurrent(A)=capacity(Farad)*(yesterdaysvoltage[V]-todaysvoltage[V])/t[s]

    Cda LIC1840Q3R8757
    specified capacitance (F)250250220750
    dimensions (diameter x length mm)12.5x3516x2016x2518x40
    operating temperature (Celsius)-25 +70-40 +85-15 +70-40 +85
    life cycles20k (10%)30k (30%)500k (30%)30k (30%)
    Current continous A0.750.751.13
    Current peak A 1s51015.330
    ESR AC 1kHz mOhm505010025
    specified leakage current 72h/uA1051223
    price at qty1000$1.90 at Vinatech$1.90 at a Aliexpress seller$8.60 at mousernot requested
    measured capacitance CC 0.5A301F (+20%)220F (-14%) 228F  (+3%)761F (+1%)
    measured leakage 72h/uA3124

    The CDA 250F LIC has 14% lower than specified capacitance, not perfect but not bad either. The leakage of all LICs after 72hrs is very low, less than 5uA, so it becomes insignificant for most energy harvesting application. 

    Conclusion: the 250F and 750F 3.8V LICs from Cda are not bad at all. The specs are good, the measured capacitance is not bad and leakage is good. For the Cda 250F the price is $1.90 in qty1000 same as the Vinatech 250F.

  • Leakage current of lithium ion capacitors vs supercapacitors

    Jasper Sikken03/09/2022 at 21:06 0 comments

    I have compared the leakage current of a few Lithium Ion Capacitors (LIC) to a few supercapacitors (EDLC). I charged them up and then almost every day I measured the voltage decay with a handheld  multimeter and calculated leakage current using following formula

    The chart shows for both LICs and EDLCs that after 72 hours the leakage current is less than 5uA, which is good news for low power IoT devices. Interestingly the high capacity (>200F) LICs have a lower leakage than the 10F supercapacitors. And if you plot the leakage per Farad capacity, then it becomes even more clear that LICs have about 5 times lower leakage than supercapacitors. 

  • AEMLIC V2 has 3.3V/400mA output

    Jasper Sikken01/08/2022 at 20:30 3 comments

    I didn't like so much that the AEMLIC V1 has a 2.2V/80mA output while many wireless development boards need 3.3V and a higher current. Therefor I have designed the new AEMLIC V2. It has an TPS63900 buck boost converter that delivers 3.3V/400mA. It is again based on the trusted AEM10941 (there was a problem with the newer AEM10300).

    When AEM10941 storage voltage drops below 2.5V STATUS1 pin goes high and the host processor is immediately informed of an impending shutdown so it can gracefully end EEPROM FLASH/EEPROM write operations before it's power is shut off. The buck booster enable pin is controlled from the AEM10941 VHV output that goes low 600ms AFTER the host processor is warned. 

    The bare PCB looks like this

    The assembled PCB looks like this

    And after soldering onto headers and adding a 250F Lithium Ion Capacitor and a solar panel it looks like this. 

    I have performed full hardware design verification and found no problem. 

    Then I felt confident enough to order the first batch of 50 from Elecrow. They are currently being produced and I expect that they are available for sale end of March 2022. Elecrow already share an image of the bare PCBs

    Normally the TPS63900 buckboost converter is $0.80USD in qty1000 but the stocks were drying up at the normal distributors and so in January 2022 I have bought 220pcs from ICSOSO for $2.10 each. 1 Month later all stocks were completely dried up and Elecrow could source it for $39 each! So I was happy I bought some.

    I am so excited about this new board because it uses an awesome solar energy harvesting chip, it stores the energy in a truly green storage device and it is can supply high enough current to power most IoT projects, not only BLE or LoRa but also NB-IoT, LTE-M, and wifi.  This week I expect shipping info, which means the board will arrive next week around wednesday. I am so excited!

  • Just received the first batch of 100

    Jasper Sikken04/26/2021 at 19:44 0 comments

    Today I have receiv

    And now I have put it for sale on Tindie. It is still waiting approval. 

  • PCB assembly is in progress

    Jasper Sikken04/15/2021 at 19:33 0 comments

    The Chinese PCB assembler Elecrow has recently shared with me the pictures of the panel and of the first assembled board, for my approval. 

    I see some solder mask issues between the pads of the IC but Elecrow assured with me there is no problem and that there will be no short circuits. 

View all 5 project logs

Enjoy this project?



sup wrote 04/06/2022 at 09:42 point

Meybe create a small tamagotchi , small screen 2 color monochrome, 4 or 5 buttons and meybe irda or wifi for communications.

If everyone can change firmware people make a software ;-)

I will be excited

  Are you sure? yes | no

Nazwa wrote 03/20/2022 at 19:07 point

i need only one thing

is possible runing  a bash on any device for a week?

simple, put solar panel, battery and 24h*7 day working time. I no need browser, sound etc. I need only ssh, bash, awk, sed troff, tee

  Are you sure? yes | no

teraz wrote 03/31/2022 at 15:37 point

for example fuzix?

  Are you sure? yes | no

Nazwa wrote 04/01/2022 at 20:15 point

Yes, fuzix OS is great idea. Small system and very long working device for example month.

  Are you sure? yes | no

Kevin Stuart wrote 09/17/2021 at 21:03 point

Excellent writeup!  (I've already ordered my board from your site.)

If I could ask, why did you set the MPPT to 70%?  With the first solar cell (1V/80mA), I see that Vmp = 1V with Voc = 1.2V, so I would have thought that you'd set MPPT to 75% or 85% (out of the 70, 75, 85, 90% options) since Vmp/Voc = ~83%.

Did you actually measure a different Voc than the stated value?  (I see on the 2nd solar cell (2V/160mA) that they measure Voc = 2.3V but spec Voc = 2V it would appear.)

  Are you sure? yes | no

Jasper Sikken wrote 10/17/2021 at 07:04 point

Hi Kevin, sorry for the late reply. Theoretically 80% is better, but I set mppt to 70% because during a real life test I learned battery charge current was highest at 70%.

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates