State of charge (SoC) decreases dramatically around the half way through?

Dear all,

I used two LiPo batteries (Samsung ICR18650-26F 2600mAh) for powering Electron and attached sensors (BME280 and TSL2561). Sensor datasets were upload to ThingSpeak every 5 mins. The firmware used is 0.6 rc1.

The SoC and voltage figure (see below) shows that device consumed more in the half way through of its life and it used power more conservative when SoC is is below 10%. By now, SoC is 1.51% and the device is still running and reporting data. :worried:

Any ideas why this happened? Suggestions for improvement?

Thanks.


The fuel gauge is not that accurate but it is better than nothing.

There is not much you can do to change this behavior without using a different fuel gauge chip which would be a shame.

You can log the max charge voltage and then the voltage right before it shuts off and then write your own code to provide a more accurate SOC reading based on battery voltage.

How did you attach the battery to the electron? What’s the connection like?

From the fuel gauge datasheet:

The MAX17043/MAX17044 are ultra-compact, low-cost,
host-side fuel-gauge systems for lithium-ion (Li+) batteries
in handheld and portable equipment

I don’t think it can handle a Life-po4 battery.

The fuel gauge will not work with LiFEPo4 correctly.

Based on his discharge curve he is not using a LiFePo4 battery but the Samsung Li-Ion battery which should be fine.

The fuel gauge should be fine with the Samsung battery as the data sheet says, but it’s just not as accurate as we would like it to be. If you want super accurate, you would need to use a TI Impedance Track Fuel Gauge.

My bad, but it now depends on how it is wired. If it’s some long wires and so on, it affects the reading.

I join batteries in parallel and attach the battery pack to the electron using a 2 pin JST connector.

The length of the JST wire is ~20 cm.

This is quite normal for any kind of battery. They drop more in the beginning and at the end of their cycle.

See this image. Just make sure you don’t discharge your LiPo too much, otherwise you might damage the cell and this could cause an explosion. Seriously, be careful with a LiPo. They are one of the most dangerous batteries and should always be used with a chip that controls charge, discharge and depth of discharge.

Actually, it drops more in the middle of the cycle.

I am going to see what is the lowest reading of SoC and Voltage in this project. Just wait and see. By now, it reports Voltage at 3.3V and SoC at 1.07%. In average, it drops 0.03% very 5 mins.

I got a LiPo battery (link) with a chip on it and will use this for next project.

Any comments?

Here are some more battery protection PCBs to prevent the cells from over discharge and short circuits which could prevent a fire.

Just log the batteries voltage after it has sat for a few hours after a full charge and then measure the voltage when it’s empty and then just code your SOC level display using the MAP function. So map the 3.7v - 3v battery range to 100% to 0% SOC and you should be able to get a more accurate reading for your setup.

That is basically what the chip is doing anyways just not as accurately as we would like.

1 Like

I would advise against getting the lowest reading. If you get too low, the cell is already damaged. Look up the averages of a normal LiPo cell’s voltage for safe usage and stick to those values.

1 Like

The discharge cut off voltage for your Samsung cells is 2.75 volts. Do not discharge below this unless you want to risk damaging the battery.

1 Like

The battery is still running (see attached figures). Thanks for all the tips. I keep my eyes open not letting the discharge under 2.75 volts. Just curious about mapping the SoC and Voltage… These two measurements are not in a linear relationship. It seems that mapping function may not be working. Maybe setting a threshold 5% above the cut off voltage an ideal solution?

You will have to play with it to get it just right for your setup but it shouldn’t take to much work to get it close to where you want it.

1 Like

Pretty much this. There is an expected behavior from every LiPo cell, but still each cell is a bit different. Experiment with how the voltage drops over time to ensure that you can draw the required amperes within a specific amount of time.

Again: safety first. Don’t push your battery, better safe than sorry. LiPos are incredibly dangerous without a regulating chip. I’ve researched it myself for a product I’m developing and in the end I decided to just use lead acid batteries. Read up on LiPos first, would be my advice.

1 Like

Dear all:

That is the last updated ThingSpeak dataset received from the Electron. Please see the attached figures.

Here is some questions from this experiment:

(1) The onboard power management stops the Electron from working when the Volt drops to ~3.00. However, the Electron can still operate even the SoC drops to 0. In this experiment, the Electron sent out 35 readings of SoC at 0 (5 mins uploading interval) before the last reading uploaded. I will use other batteries with a regulating chip for the following projects. For the safety issue, I would like to know any built-in functions to shutdown entire system when Volt reaches at 3.00 as well as sending out a system shutdown notification?

(2) I am lucky with these Samsung LiPo batteries in this experiment. They did not exposure and cause any fires. :grin: I would like to gauge the health of these batteries! Can you please provide some methods that I can understand how much damage has been caused in this experiment?

(3) I check with the Particle list in the CLI. It shows this Electron is still on line, although it never uploads data to ThingSpeak. Do I miss anything here?

Thanks for your comments and helps in this experiment. Any comments?

If the PMIC is cutting power from the battery off to the Electon at 3v then that is good since letting the battery drain to 2.5v will cause it's life cycle span to decrease. The biggest thing that will affect the life cycle span of the battery is how deeply you dischage it on average. Normally these batteries will not last more than 300-500 cycles when discharging them to empty on a regular basis.

Fires usually only happen when the battery is over charged or short circuited or punctured / damaged. The PMIC is cutting the discharge off a 3v which is good, and its not allowing charging above approx 4v from my testing so that is good also since it prevents over charging which can be dangerous.

You can buy batteries just like what you have with protection cirucits built into them in the future.

The Electron will show online in the CLI or online build IDE even though it's offline, it's always been this way so you can't trust it 100%.

1 Like