1
\$\begingroup\$

I have noticed that a lower-power device that I am testing is running out of power much sooner than it should. The device should last up to 2 years, however, it is running out of battery at the 5-month point. This is considerably sooner than expected.

My question is: when measuring current (mA) in series using a multimeter, does a multimeter affect the battery life of the tested device? Maybe the burden voltage or something else is affecting the battery life?

For a full explanation of my situation, see below:

I have a BLE device that uses a nRF52832 controller and is powered by a CR1220 coin cell battery. This device's default state is to sleep, however, once a button on the device is pressed it wakes, broadcasts a value to an app, and then returns to sleep.

I'm measuring the current in serial using a mains-powered UNI-T UT803 multimeter, with the below setup.

             +-------+
  +----------+Battery+-----+
  |          +-------+     |
  |                        |
  |                        |
+-+----+              +----+--------------+
|Device+--------------+                   |
+------+              | Multimeter (mA)   |
                      |                   |
                      +-------------------+

I'm measuring current in series to simply allow me to tell when the device is awake and when it returns to sleep, as the device itself has no lights to allow me to tell its state. So if I see 2.5mA on the multimeter, I know the device is awake, if I see 0mA, I know it's asleep.

I've also more accurately measured the current in the device's states from borrowing a friend's uCurrent GOLD adapter, so I have been able to estimate the device's lifespan. Which gave the following:

  • The CR1220 battery has a manufacturer-stated capacity of 35mAh
  • Sleep state current reading: 1.5uA or 0.0015mA
  • Wake state current: 2.5mA
  • mAh used per second when awake: 0.000682 mAh

For an example of what these values mean (using this tool for convenience):

If the device is always awake:
35mAh capacity / 2.5mA wake current = approximately 14 hours of battery life.

If the device is always asleep:
35mAh capacity / 0.0015mA sleep current = approximately 2.5 years of battery life.

If the device wakes for 5 seconds:
35mAh capacity / 2.5mA wake current = 50,400 seconds of battery life
So, 35mAh capacity / 50,400 seconds = 0.000682mAh used per second awake
So, 5 seconds of the device being awake = 5 * 0.000682mAh = 0.0034mAh used
So, every 5 seconds of the device being awake uses 2.266 hours of battery life
but, given the 2.5 years of battery life when the device is sleeping, this is acceptable.

Even if I were to assume that the actual battery capacity was approximately only 80% of what is expected as it may drop off when at a low charge, that would still provide 28mAh or 2.1 years of battery life when the device is sleeping.

The tests have been carried out inside and at room temperature. So all I can think is either:

  1. The multimeter is affecting battery life.
  2. The brand-new battery happened to have 25% charge when the test was started.
  3. I've made a significant error on the current tests or associated maths.
  4. The firmware is waking the device when it isn't being watched/pressed, which seems incredibly unlikely.

UPDATE

As recommended, I've tested the device with a Nordic Power Profiler. This did show significant spikes in current. For example, when pairing, the device used an average of 1.72mA current, but with spikes of up to 19.5mA, albeit for fractions of a second. The below graph shows how many seconds the current was in each mA range over a 40-second pairing period.

Graph of current usage when pairing device

Unfortunately, this still doesn't explain why my device ran out of battery at the 5-month point when it should have been at least 1 year or more.

For the above-graphed test, the total mAh used was 0.030247569mAh.

This was calculated the same as before, for example for the "Between 2mA and <2.5mA" range, using 2.5mA as the worst-case figure:

35mAh capacity / 2.5mA current = 50,400 seconds of battery life
So, 35mAh capacity / 50,400 seconds = 0.000694mAh used per second awake
29.4066 seconds measured at 2.5mA
So 0.000694mAh * 29.4066 seconds = 0.02042125mAh used

This was done for every range measured, giving the value 0.030247569mAh used to pair the device.

Pairing the device is the most power-intensive operation that can be done. If I repeated the pairing process until the battery was fully depleted it's an incredibly large number, so 35mAh / 0.030247569mAh per pair = 1157 successful pairs. Even at 80% battery capacity (28mAh) that is still over 900 successful pairs. The test I carried out was over roughly 150 days with a pair carried out once every 7 days, so the battery should not have depleted as it did.

I'm continuing to test, but ultimately, I'm still unsure what caused the battery in the test device to deplete far sooner than expected. Any suggestions of other things to test are welcome.

UPDATE 2

As suggested, I have reviewed the battery's self discharge, which is <1% per year. This should not noticeably affect the device's expected battery capacity, especially within the first year of use.

From looking at the below battery specification I can also see a "Standard Discharge Current" and "Max Cont. Discharge Current", which I thought may impact battery capacity.

Renate CR1220 battery specification

The device uses 2.5mA for roughly 29 seconds when pairing, which is higher than the Max Continuous Discharge Current of 1mA. However, I’m not sure if it makes a difference having a 29-second burst above 1mA, followed by 12 to 24 hours of sleep state at less than 0.5uA.

The Standard Discharge Current is 0.1mA and for the majority of the device’s life, it is sleeping at around 0.5uA. However, we do have up to 35-second bursts where the average is 2.5mA with spikes between 10mA and 20mA for 0.2 of a second.

The below graph shows the CR1220 battery’s capacity at various loads.

CR1220 capacity at various loads

The battery is 3V, and assuming the temperature is 23 degrees:

  • A 1kohm load, so 3mA current, provides 30mAh battery capacity (I = V/R, So 3v/1kohm = 0.003A or 3mA)
  • A 3kohm load, so 1mA current, provides 36mAh battery capacity (I = V/R, So 3v/3kohm = 0.001A or 1mA)
  • Anything above 3kohm / below 1mA does not negatively impact the battery life

With the Max Cont. Discharge Current being 1mA and this device using an average of 2.5mA for 35 seconds every 12 or more hours over its multi-year lifetime, this may drain more battery life than expected. However, I'm not sure how to calculate the impact to understand if it's significantly different to standard battery drain calculations.

Furthermore, would the short bursts between 10mA and 20mA that occur for 0.2 of a second every 12 or more hours dramatically affect battery life beyond standard calculations due to the Max Cont. Discharge Current of 1mA?

Even when considering the above, my battery life calculations so far have been calculated using a figure of a 35mAh CR1220 at 80% capacity, so 28mAh. This reduced figure of 28mAh is lower than the 30mAh graphed at a continuous 3mA load, so should already take a reduced battery capacity into account.

\$\endgroup\$
10
  • \$\begingroup\$ It's worth noting that I understand that the multimeter burden voltage might prevent the device from working if I tried to measure a low current, but that isn't the case here. I am running long term series current tests with the multimeter on mA and the device performs as expected. \$\endgroup\$
    – Calco
    Commented Nov 8, 2023 at 16:04
  • 3
    \$\begingroup\$ The BLE broadcast might consume a significant energy peak you don't see on the multimeter. \$\endgroup\$
    – Jens
    Commented Nov 8, 2023 at 16:09
  • 2
    \$\begingroup\$ what is the battery life if the device continuously transmits? \$\endgroup\$
    – jsotola
    Commented Nov 8, 2023 at 16:10
  • \$\begingroup\$ Jens - Do you mean that for a split second it could be using significantly more than 2.5mA? \$\endgroup\$
    – Calco
    Commented Nov 8, 2023 at 16:15
  • 2
    \$\begingroup\$ What's the self-discharge rate of the battery? \$\endgroup\$
    – Hearth
    Commented Nov 15, 2023 at 16:24

1 Answer 1

1
\$\begingroup\$

Your calculations seem correct, and the multimeter (if the resistance is low enough) should not have any effect on the performance.

However, the 2.5mA doesn't seem right! Your multi-meter could be filtering the sudden changes in measurement. Use a power profiler to see the fast changes and calculate the power.

I've tested the nRF52840 before. At 0dBm and 1Mbps, I could see at TX, it was taking more than 8mA. This link presents almost the same values.

If you are using less than 0dBm TX power, it does not change much as the CPU will be on and consumes a big part of the power. Even in RX mode, when the CPU is on, it will take around 5mA.

\$\endgroup\$
6
  • \$\begingroup\$ Ok, so whilst my multimeter is showing a result of 2.5mA, it could actually be split-second jumping up to as much as 8mA and then back down to 2.5mA? If so, that would certainly have a dramatic effect on the device's lifetime! \$\endgroup\$
    – Calco
    Commented Nov 8, 2023 at 16:33
  • 1
    \$\begingroup\$ Kind of, yeah. Let's assume the duration of the TX could be for example (not a real value) 25ms with 8mA current. Then, assume the time constant of the multi-meter filter is 80ms. It will not reach 8mA at all and would show something about 2.5mA (these are rough numbers, I didn't calculate the correct values). \$\endgroup\$
    – Saadat
    Commented Nov 8, 2023 at 16:51
  • \$\begingroup\$ I've ordered a power profiler to more accurately measure the peaks in current usage. I'll update this question with the results of the testing shortly. \$\endgroup\$
    – Calco
    Commented Nov 9, 2023 at 10:16
  • \$\begingroup\$ I've edited my answer with the results of power profiler testing, but unfortunately, that hasn't provided any answers. It has given me a much more accurate current reading for my projections though, so I'm thankful for the suggestion! \$\endgroup\$
    – Calco
    Commented Nov 15, 2023 at 16:21
  • 1
    \$\begingroup\$ Glad it helped. If the data shows low power consumption, the other thing to look into, as you mentioned in the question, is the battery itself. It could be a bad batch or a faulty battery. Other than that, I have no idea. \$\endgroup\$
    – Saadat
    Commented Nov 15, 2023 at 16:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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