0
\$\begingroup\$

To give a background on this, I have been running tests to determine the actual lifespan of a 3 cell pack of some Samsung 18650s and Tesla 2170s at just above freezing. The Samsungs are INR18650-35E and the Teslas were pulled from a 2018 Model 3. Voltage was checked on all cells before being wired together with a BMS connected to them. This one.

These cells were placed in a cold environment, around 35F and then connected to the LED strip and left to run in front of a timelapse.

In between the cells and the LED strip was a watt meter/power analyzer, I bought it from Amazon here.

wiring diagram

The Voltage dropped from a full charge to 9.0V as expected and then cut out. However after harvesting the data, I noticed something odd. The life of the Samsung 18650s in my test lasted about 7.5 hrs. The amp and watt pull slow decreased over the course of the first 4.5 until they hit 0 (although the LEDs didn't turn off), and then rebounded back up to to their initial pulls at the start of the test and then slowly receded until the BMS cut them off.

Any ideas why this would happen? The same thing happened with the Tesla 2170s too, just after over half their capacity was used.

\$\endgroup\$
11
  • \$\begingroup\$ How are you measuring capacity/capacity remaining? Just the voltage? \$\endgroup\$
    – Ron Beyer
    Commented Apr 4, 2019 at 20:24
  • 1
    \$\begingroup\$ Brian - (a) You said: "The amp and watt pull slow decreased over the course of the first 4.5 until they hit 0 (although the LEDs didn't turn off)" That seems impossible - if the LEDs are still "on", then current (as you called it "amp") must be > 0. (b) IMHO adding photos of your test setup into your question, would help readers to better understand it. \$\endgroup\$
    – SamGibson ♦
    Commented Apr 4, 2019 at 20:28
  • \$\begingroup\$ @SamGibson - Right, that's what I was thinking. That doesn't make any sense and was thinking it had to be the cheap Amazon anazlyer. Let a sketch up of a what I had, since I already dismantled it. \$\endgroup\$
    – Brian
    Commented Apr 4, 2019 at 20:33
  • 1
    \$\begingroup\$ Can you show exact data in a graph? There are reasons why batteries have memory capacitance with high ESR that cause batteries to bounce back when cut-off is reached and then cut-in is reached continue as before. This memory capacity can be as much as 50% but is not used as much as the lower ESR capacitance in fast discharge tests. ESR is also much higher on the bulk capacitance at low temps. I use an equivalent circuit with 2 RC values and a threshold voltage but in reality, there are many ESR*C shunt values in a cell \$\endgroup\$
    – D.A.S.
    Commented Apr 4, 2019 at 20:59
  • 1
    \$\begingroup\$ The values are on the order of 10kF and 50 mOhms with Vt around 2.5V but in the case of Maxwell Ultracaps they use a similar model to determine both C's combined based on slow and fast charge rates. I think they both have a double-electric charge layer effect. Also for this reason ESR changes with low frequency due to ESR1*C1=T1 and T2 effects \$\endgroup\$
    – D.A.S.
    Commented Apr 4, 2019 at 21:25

1 Answer 1

Reset to default
0
\$\begingroup\$

When you are performing test and/or other research activities, it is customary to qualify and calibrate all your instruments with known good sources and loads. Please do this with your power meter.

Get/rent a good benchtop power supply with logging capabilities (to record voltage and current), and get a known good load, either electronic load or a good resistor, and take readings of your power meter over a good range of loads and supply voltages, and check if they match the external (calibrated) numbers.

Only after that you may start researching behavior of highly non-linear sources as Li-Ion cells and highly non-linear loads as LEDs. For research of time-variable sources and loads, it is highly advisable to have meters with logging capabilities, so you will be able to correctly get integrals over time.

\$\endgroup\$

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.