INTRO: (PS, if you're not interested, skip to the last paragraph)
I'm building a battery pack for my university electric racecar team. We plan to use 18650 cells, the Samsung 30Q which are only rated for 15A continuous. However, Mooch over at e-cigarette-forum tested them and shows that they are capable of 20A continuous and 25A peaks -
However, in his testing, he pushes them to 50A pulses for 5 seconds and they still work fine. Due to weight and size constraints, we're forced to use the batteries in 8 parallel circuits. According to Samsung's datasheet, that's only 120A continuous, but going by Mooch that's 160A continuous with a 200A peak. We might, however, draw up to 270-300A in very short bursts of less than 5 seconds, that's a burst from the battery of about 37A. Our problem is that we are required by the rules to fuse the batteries at the rated max current of the cell which isn't mentioned in the datasheet, and results posted on a forum don't count as scientific evidence of max possible current. Moreover, we have a custom cooling solution which should help increase our peak currents even higher than the tests mentioned above. Our only option then is to test the batteries ourselves and show sufficient proof of the max current, which brings me to the problem I'm having. A single cell, going by the aforementioned tests should be able to output 50A+ and thus a pack of 8 of them can easily exceed 400A given proper cooling. Why do we need to test the whole pack of 8 you ask? If we blow the fuse on one cell, say if its defective, or even just badly welded on, the remaining 7 cells will have to supply more current than before. We need to ensure that it won't cause a chain-reaction of fusing all the cells and break the circuit.

To test the batteries, we need some sort of current limiter and load. All the circuits I found here and elsewhere are rated for around 50-100A. How would I go about building a battery testing rig at that can limit up to currents of 350A? Could I do a multiple parallel BJTs in a current mirror circuit configuration? Are there any caveats to that?
Also, what safety precautions do I need to take considering I'm pushing these cells to their limits and to failure?
I appreciate all the help and suggestions I can get

I realized there were some additional calculations I missed. The lowest allowable battery voltage(not cell EMF) is 2.5V. At 40A, that's an external load of \$62.5\,m\Omega\$. To reduce power dissipation of the current limiting circuit, I can add a power resistor of that resistance in series, limiting the power dissipation of the current limiter to \$20*20*62.5*10^{-3}= 25\,W\$ (Max power dissipation when loads are matched).

When testing 8 cells in parallel the external impedance can drop to
\$62.5/8=~7.8\,m\Omega\$ and power dissipation through the limiter would be about 800W

  • 3
    \$\begingroup\$ My company builds exactly these things. Save yourself the trouble of testing them yourself and hookup with your electrochemistry department or a battery researcher. You don't want to invent this wheel. (It's not that I doubt your abilities, but I don't want you to chew up your time or budget building a battery tester) \$\endgroup\$
    – pgvoorhees
    Dec 8, 2017 at 23:18
  • 1
    \$\begingroup\$ The name of the circuit you are looking for is called an "active-load". How you go about building one of these is an entire product worth of information. You're definitely barking up the right tree with a current mirror, but the transistors have to be reasonably closely matched. We had to build another device to characterize the transistors in our active loads so they would share the current correctly. \$\endgroup\$
    – pgvoorhees
    Dec 8, 2017 at 23:28
  • 1
    \$\begingroup\$ Excuse me, but are you going to connect in parallel to a single load several high current batteries which are separate before you connect them together? Are you going to do it without any preceding tests? \$\endgroup\$
    – user136077
    Dec 8, 2017 at 23:33
  • 1
    \$\begingroup\$ It's both. I really don't have relevant experience. But i would go to a current mode boost cenverter with power resistor as load. And i would regulate it by input current rather than output voltage. \$\endgroup\$
    – user76844
    Dec 8, 2017 at 23:36
  • 1
    \$\begingroup\$ Mooch at e-cigarette must be a big authority in testing of 18650 batteries. How about this document (from Samsung SDI), eu.nkon.nl/sk/k/30q.pdf , which shows testing at 22A? The cell temperature will go over water boiling point however. There is no real limit, just cell thermal dissipation ability from the DC internal imedance. \$\endgroup\$ Dec 9, 2017 at 0:05

1 Answer 1


8p78S cells =624

How to choose charge limit and discharge limit?

  • The speed by which a battery can safely be fast-charged is governed by only by temperature (with a 10'C rise) but datasheet defines nominal values
  • Hence diode thermal sensing is essential and thermal resistance to inner cell core worst case cell temp difference to sensor must be specified.
    • This can be predicted by measuring the cell ESR initial, but it rises rapidly with low SOC and slowly with each use cycle.
    • Capacity and ESR improves with higher temp but aging is faster and risk is higher if temp duration is extended >5min.
    • Exposure to humidity and heat will shorten life;
    • 10% swelling from internal gassing from self-heating indicates cycle EOL or worse, pressure induced cell rupture can cause high risk fires.
    • gaps are critical
  • cell capacity is a complex function of Voltage, Current, Temperature and Coulomb counting.

Samsung Specs: INR18650-25R Version 1.0 Mar. 2014

Nominal 0.2C capacity: 2,500mAh ( if used as follows);
... (implies 4 hr rate @ 625mA)
Charge: 1.25A CC, CV 4.20V, 125mA cut-off (10% CC),
Discharge: 0.2C, 2.5V discharge cut-off,
Rapid charge: 4A CC, CV 4.20 ± 0.05 V, 100mA cut-off (2.5% CC)
Charge Time: Standard: 180min , Rapid: 60min (at 25℃)
Max. continuous discharge: 20 A (@25℃), 60% at 250 cycle
Weight: 45.0g max
Size: (max) 65.00 mm H x 18.40 mm D
Temp Charge : 0 to 50℃ (recommended recharge release < 45℃)
Temp. Discharge: -20 to 75℃ (recommended re-discharge release < 60℃)
Temp. Storage * : 1.5 year -30~25℃, 3 months -30~45℃, 1 month -30~60℃ * only if kept at 50% SoC then > 90% recovery

  • There are double layer electric properties ( multiple RC equivalent internal circuits) which give rise to a short term memory effect to surge load and voltage rise when released.
  • Each ESR C value function of total capacity and the reason why capacity drops going from discharge rates of 1C to C/50 .
  • but specs only apply to 0.2C discharge rate

If used as Samsung specifies. at 0.2C rate, I calculate C (equivalent) = 500 Farads @ 3.6Vavg


simulate this circuit – Schematic created using CircuitLab

    • Current sensing per string is easy Ultra-fast charging should only be applied at a moderate temperature. The charger should limit the time the cell stays at elevated voltage.

Cannot be ignored

Anyone playing with large LiPo pack designs, ought to know Samsung's cardinal Rules.

  • Misusing or mishandling lithium ion batteries can pose a SERIOUS RISK of - personal injury or property damage
  • Usage of batteries is AT YOUR OWN RISK!
  • ONLY use with proper protection circuitry
  • DO NOT short circuit intentionally or unintentionally
  • KEEP AWAY from metal/conductive objects to prevent short circuiting
  • DO NOT use if PVC wrapper or terminal insulator is damaged or torn
  • DO NOT use if battery is damaged in any way
  • DO NOT over-charge or charge above the maximum voltage rating
  • DO NOT over-discharge or exceed the continuous discharge rating
  • DO NOT modify, disassemble, puncture, cut, crush, or incinerate
  • DO NOT expose to liquids or high temperatures
  • DO NOT solder onto battery, spot weld only
  • DO NOT use force to install or install in reverse/backwards
  • ONLY use within manufacturer’s specification
  • KEEP AWAY from pets and children
  • ALWAYS charge in or on a fire-proof surface and never leave batteries charging unattended
  • ONLY use a smart charger designed for this specific type of battery
  • DO NOT mix and match brands and models, old and new, used and unused batteries
  • STOP immediately if while charging/storing/using the battery it emits an unusual smell, feels hot, changes color or shape, or appears abnormal in any way
  • It is your responsibility to determine that your charger or device is functioning properly
  • If exposed to battery electrolyte, flush with water immediately and/or immediately contact a physician or emergency services
  • DO NOT throw away in trash; contact your local jurisdiction for proper recycling or disposal
  • \$\begingroup\$ Wow, thank you! We're setting charge limit at 300V, the max allowable voltage. Discharge limit is simply 2.5*78=~200V. We're required by rules to measure temperature less than 10mm from the negative terminal of the cell anyway, so that's not a problem. And we can use the thermistor to ensure a constant temperature of 40-60C, the ideal temperature for running the battery. We'll look into the rc effects of having such a large number of batteries for sure. I'm not sure I completely understand the 'current sensing per string' line. Does that mean we would monitor the current of every cell? \$\endgroup\$
    – c10yas
    Dec 9, 2017 at 10:55
  • \$\begingroup\$ What happens if you charge up 78 capacitors in series each 500 Farads each with a tolerance of 10%. What is the worst case deviation in final charge voltage. Same thing for ESR except 50% tolerance, Now imagine the 500F is only 200F but you have 300F in series with 100mohm like 2 of the 4 caps in my model. Now exercise that with your charge/discharge profile. and see the difference in string currents and pack voltages and compute worst case T rise per cell. guessing R jc for 18650 is 10'C/W from ESR losses depending on your package thermal Rca add more ... \$\endgroup\$ Dec 9, 2017 at 15:47
  • \$\begingroup\$ 200V and 300V is not what I would choose. \$\endgroup\$ Dec 9, 2017 at 15:49
  • \$\begingroup\$ Now imagine Ceq2 is 300F like above but RC=T time constant is in minutes ie ESR>1 like the response you get from pulses. While Ceq2 is 200F , and your motor load DCR is now less than 1 , what is the battery cell efficiency <50%? yet towards maximum power transfer. Your cell model will vary. Do a Falstad simulation. \$\endgroup\$ Dec 9, 2017 at 16:05
  • \$\begingroup\$ why wouldn't you choose 200V to 300V? \$\endgroup\$
    – c10yas
    Dec 9, 2017 at 23:33

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.