Continuous vs intermittent discharge

Henrik K. Jensen (HKJ) publishes comprehensive discharge measurements for various battery models. All these curves are for continuous discharge at room temperature. However, most people do not use their flashlight continuously in the same setting until battery is empty and at other temperatures too.

Cavers for example use their headlight at a low setting (typical 0,1 to 0,5 A) for walking and climbing all over the day until battery is empty. Nevertheless, sporadically they use the light in boost mode (typical 2 to 4 A) for short time. The curves for continuous discharge do not cover this kind of usage.

To get an idea of effects like depolarisation or internal heating of the cell, I carried out some measurements with two recent samples of LG ICR18650 D1.

I performed three different kinds of measurement (diagram 1). Two of them, continuous 0,2 A (blue) and continuous 3 A (red) are rather redundant, since HKJ already has done them. The third measurement (purple) was intermittent with the following cyclical time scheme: 0,2 A during 1440 s (= 80 mAh) followed by 3 A for 24 s (= 20 mAh). Each cycle discharges 100 mAh. In the average, 80% of charge is extracted at 0,2 A and 20% at 3 A.

Diagram 1 larger image

If there would be neither depolarisation effects nor internal heating, the purple curve would jump between the blue and the red curve exactly.

I performed these first measurements at room temperature (19 to 21°C). Internal heating of the cell can be neglected during continuous discharge at 0,2 A. This is true even for intermittent discharge. However, during continuous discharge at 3 A the cell heats up to more than 40°C (temperature not measured). This explains why the upper part of the purple curve in diagram 1 follows the blue curve, but the lower part differs from the red curve.

Diagram 2 shows the purple curve of one sample zoomed in to a small part of diagram 1.

Diagram 2

The program takes one measurement every 10 mAh. During discharge at 0,2 A this means, it takes a measurement every 180 s. During discharge at 3 A, it takes a measurement every 12 s. Additionally the program takes one measurement immediately after switching the current. Each cycle contains 9 + 3 measurements.

Diagram 3 shows the same small sequence as diagram 2. However, the vertical axis scales in time, not in charge as in diagram 2.

Diagram 3

It is obvious that voltage will sag when current increases due to the internal resistance of the battery. However, the voltage sag is varying over time, mainly due to depolarisation effects. To demonstrate these effects in detail it would be necessary to sample the curvature at a higher sample rate.

If you are interested in something like pulse response (short time behaviour, whatever "short" means) you should take the vertical step height of the curvature, i.e. from immediately before current switching [8. @ 0,2A] to immediately after switching [start 3A] or vice versa from [2. @ 3A] to [start 0,2A]. This kind of "internal resistance" equals about 85 mΩ in diagram 3.

If you are interested in average voltage sag, you rather should take some average voltage before and after the pulse (whatever "after" means) and compare it to the average or a typical value during the pulse. This kind of "internal resistance" equals about 95 mΩ in diagram 3.

Astonishingly, at the beginning of discharge the intermittent curve does not touch the continuous curve. However, at the end of discharge the intermittent curve falls below the continuous one. An explanation to that may be, that depolarisation and heating act contrarily: Heating reduces the internal resistance ("helps") but has an extremely long time constant.

Warning!

This thread will grow over weeks, since I plan to make additional measurements:

• Intermittent measurements at lower temperature, since room temperature does not cover all application (some caves are rather cold).
• Intermittent measurements at room and at lower temperature for other battery models, since there are some good batteries on the market other than LG ICR18650 D1 too.

These future updates just will add new intermittent measurements rather than repetitions of all those detailed explanations. Additionally, I will not measure any blue and red curves anymore, since HKJ is the preferable source for that: I will add purple curves only.

Added: Comparison 20°C vs 5°C

Additional measurements were carried out at lower temperature. The intermitted discharge for one sample of LG ICR18650 D1 was done only. The influence of temperature is dramatically!



Diagramm 4 larger image

The air temperature around the bare cell was 4 to 6°C. This may represent usage in icecaves with temperatures of 0°C or just below.
Reason:

• During caving the battery is hermetically paked inside a "bomb-prove" casing of PVC which also isolates thermically. At typical discharge heating is weak but not zero.
• The battery is mounted at the rear of the helmet. This area is heated by the bearer.

Example of a battery carried at the rear of a helmet

I assume the blue curve to show the worst case scenario for caving - not surely.

Re: Added: Comparison 20°C vs 5°C

This is some nice work :thumbsup: thank you for doing it and sharing the results.

I'm interested to see more of what you post here.

My first reaction was that the first graph is excellent and the discharge goes between the two curves closely enough to not worry too much about this effect. Looking too closely for differences can become an obsession.

When I test lights, I have to run laboratory style 'experiments' and this does involve removing the 'human' aspect of using the light. As you very correctly point out, the way a person uses a light will vary according to their current requirements , so output and therefore loading on the cell(s) will vary. It means that real world runtimes are difficult to predict unless you switch it on to one mode and keep it at a constant temperature.

Watching with interest...

Added: LG ICR18650 E1 at 20°C vs 5°C

Wow, what a difference!

The sample of LG ICR18650 E1 actually I bought from Akkuteile has a violet wrapper, not the known sea green one as in the past. This sample behaves much better than the model 'D1' I described in my last post.

HKJ and many other independent testers concluded that 'E1' has higher capacity than 'D1' to the price of increased internal resistance and more voltage sag. This is gone now.
Additionally the influence of temperature seems to be much lower now than before.



Diagram 5 larger image


Conclusions

• It is necessary to countercheck known measurements from time to time, since technology of a device may change without announcement.
• As far, as can be seen now, actual LG ICR18650 E1 is better than actual 'D1' in capacity and with respect to voltage sag.

Exemplary Evaluation for a light with turbo mode
A light draws a low current of 0,1 to 0,4 A all over the time. It provides a turbo mode with 3 A for short term use. The light needs at least 3,4 V in turbo mode.

• When the battery 'E1' has room temperature (20°C), than turbo mode can be used undimmed until about 2800 mAh are extracted. Beyond this time turbo mode starts to dim.
• When the battery 'E1' has 5°C only, than turbo mode starts to dim as early as 1700 mAh are extracted.

For battery type 'D1' these points are reached at 2500 mAh and 1200 mAh respectively.

Warning: Temperature could have been inaccurate

Sorry!

I encountered problems with the reproducibility of my measurements at low temperature in the range 4 to 6°C.

I use an old refrigerator, filled with 16 bottles of water to stabilize the temperature over time. The battery under test was in the top bay. I measured the temperature with an alcohol thermometer anywhere in the top bay.

Now I realized that the temperature varies inside the top bay spatially. Consequently, I cannot guarantee that I measured both batteries exactly at the same temperature. However, the behaviour of the batteries may change between 6 and 4°C essentially.

At the moment I cannot judge which one of 'D1' and 'E1' works better at low temperature.

Actually, I am working on an improvement of temperature accuracy. I will place the battery under test into a fixture inside a bigger metallic box (which itself will be fixed inside the upper bay of the refrigerator) to decrease the influence of spatial inhomogeneity. Additionally I will use an electronic sensor (LM34) in contact with the battery directly. With such an arrangement, it should be possible even to measure the slight increase of temperature due to low discharge currents I hope. I aim to guarantee an equality of temperature between different batteries of less than 0,5°C (absolute accuracy will be 3°C only).

I will repeat intermittent measurements of LG ICR18650 D1 and E1 at about 20°C and at about 5°C and publish them. With more precisely matching temperatures.

Warning abrogated: Temperature control improved now

The curves derived at precisely matching temperatures have not changed significantly. At the result LG ICR18650 E1 is slightly better than LG ICR18650 D1 at room temperature. At cave temperature it improved significantly.

Now the digital resolution of my temperature measurement setup is 0,033°C. Noise peak-peak is in the order of twice this value. The sensor touches the wrapper of the battery under test. The absolute calibration of the sensor LM34DZ is +/-3°C only. However, the relative accuracy is high enough to recognize even small temperature differences of lees than 0,5°C between separate measurements at least.

I do not aim to make precise definition of what I mean by 'cave temperature'. Nevertheless I aim to make realistic and fair comparison of different battery models. Therefore I have to achieve the same temperature for all models under test. Comparing one cell at 7°C with another one at 4°C will be neither realistic nor fair, since the one at 7°C will have a big advantage.

My measurements at 'cave temperature' start 2 hours after placing the battery into its fixture in the metal box in the refrigerator. At start time the temperature reading is 4,9 to 5,1°C. During the intermittent discharge the temperature reading increases slowly. Each 3A pulse of 24s duration increases the temperature by 0,25 to 0,35°C, but the long period of 24 minutes between those pulses cools it down again nearly completely. Temperature reaches a local maximum about two minutes after the pulse has finished.

The absolute maximum temperature reading of 6,3 to 6,5°C occurs about two minutes after the last 3A pulse of the measurement sequence. The thermal curvature of LG ICR 'D1' and 'E1' are similar and may be typical for any battery in free air (inside a large metal box).

The behaviour at 'room temperature' compares to what I explained above principally.

I repeated all published measurements with the new setup and provide an update to diagram 5 from post #4 here in this post.



Revised Diagram 5 (larger image)

The curves did not change very much. The temperature differences are small enough to approve now: Actual violet 'E1' performs slightly better than actual pink 'D1' at room temperature and remarkably better at cave temperatures.

Here is an additional diagram, showing the same measurement results as diagram 5. In this diagram I showed the voltage at the end of each 3A discharge pulse only, for 'D1', 'E1', at 'room temperature' and at 'cave temperature'. This kind of diagram best suits to visualize the differences between battery models.



Diagramm 6 (larger image)

---

Next step will be to measure behaviour of bare Panasonic NCR 'B', 'BD' and 'BE' and compare them to LG ICR 'E1' with respect to availability of short term 3A turbo mode at 'cave temperatures'.

Re: Warning abrogated: Temperature control improved now

Tobias Bossert said:

Next step will be to measure behaviour of bare Panasonic NCR 'B', 'BD' and 'BE' and compare them to LG ICR 'E1' with respect to availability of short term 3A turbo mode at 'cave temperatures'.

Click to expand...

Interesting tests. Temperature definitely does play a big role in the discharge capability and capacity of cells. It would also be interesting to see a test on a high capacity yet high current cell, with much lower internal resistance such as the Samsung 25R.

Added: Panasonic NCR18650 B

I did the same measurements with the bare NCR18650 B cell from Panasonic as I did before with the cells from LG Chem: Intermittent discharge 1440s 0,2A | 24s 3A; at room temperature and at cave temperature. I show it in two kinds: First in the style of diagram 5 (post #4), so you can see the voltage pulsing synchronously with the discharge current. The disadvantage of this kind is, that the curves hide one another.



Diagram 7 (larger image)

Additionally the same measurement visualized in the style of diagram 6 (post #6)



Diagram 8 (larger image)

The curvature of the discharge diagram of both cells is quite different. The cell from Panasonic shows a voltage falling linear with the charge already extracted. The curvature of the cell from LG Chem starts to get flatter in the middle of the discharge process and then holds the voltage better than the Panasonic cell.

The LG Chem cell has the advantage of a significantly higher open circuit voltage at any time of the discharge process. Nevertheless, the Panasonic cell has the advantage of a lower internal resistance than the cell from LG Chem. Table 1 shows the internal resistances calculated from the measurements:


* see post #1 for explanation of internal resistance
Table 1

Consequently, it depends upon the application, which cell holds up better. At low current levels, the voltage under load of the LG Chem cell is higher, whereas at high current levels the Panasonic cell provides higher voltages under load. At 3A the LG cell still gives higher voltages.

Both cells suffer from low temperatures (cave temperature means cell temperature in the range 4,5 to 6,5°C).

The main interesting quantity for the user of a single cell light with short-term turbo mode is how long undimmed turbo mode will be available. This will depend upon the voltage required to achieve full turbo mode. Table 2 shows the estimated charges, which can be extracted before turbo mode of 3A starts to dim:


* see post #1 for explanation of intermittent discharge
Table 2

The slope of the curves on the right side of diagram 7 are rather flat for the LG Chem cell and rather declined for the Panasonic cell. This means, that the dimming of turbo mode takes place rather slowly with the LG Chem cell and rather fast with the Panasonic cell.

Conclusion:

LG ICR18650 E1 charged to 4,35V is better suited for single cell lights with short-term turbo mode, at room temperature and at cold environment.

---

Next step: I will measure Panasonic NCR18650 BD, since it is known to have a lower internal resistance than type B.

Re: Added: Panasonic NCR18650 B

Interesting comparison. Also take note that most 4.35v cells have a rated average voltage of ~3.75 to 3.78v @ 0.2C discharge. The Panasonic on the other hand has a 3.60v rated voltage @ 0.2C.

Also, LG is coming out with a 3400 and 3500mAh cell very soon!

Re: Added: Panasonic NCR18650 B

jasonck08 said:

Also, LG is coming out with a 3400 and 3500mAh cell very soon!

Click to expand...

Unfortunately this new cell 'MJ1' will be 4,20V only. Nevertheless it will be a verry interesting cell, I guess.

Re: Added: Panasonic NCR18650BD and BE

I measured a grey NCR18650BD and a lime green NCR18650BE from Panasonic now (both are unprotected). Both show curvatures quite similar to the lime green unprotected NCR18650B. Under load BD and BE hold voltage slightly better due to lower internal resistance. At the same time, the capacity of type B is slightly higher than that of both newer types. As a whole, they are three of a kind when it comes to intermittent discharge of 3A.

The following diagrams show five cells: ICR18650E1 and ICR18650D1 from LG Chem (4,35V) as well as NCR18650BD, NCR18650BE and NCR18650B from Panasonic (4,20V). Since it is difficult to hold track of ten crossing curves, I partitioned the diagram into one for room temperature and one for cave temperature.



Diagram 9 (larger image)



Diagram 10 (larger image)

If you prefer to see the voltages during intermittent discharge for the three Panasonic cells, they are here and here. The according curves for the LG cells are published in diagram 5 of post #6.

Table 3 is an updated version of table 1 from post #8.



Table 3

As jasonck08 argued, high current cells have lower internal resistance and perform better for intermittent applications in principal. The internal resistance of all three Panasonic cells is lower than that of the cells from LG Chem. However, the main disadvantage of the Panasonic cells with respect to those from LG Chem (4,35V) remains that the open circuit voltage is significantly lower – especially when most of the charge is already extracted (right side of discharge diagrams).

Which advantage predominates (higher open circuit voltage or lower internal resistance) depends upon the current level.

The only real high current cell I own now is Sony US18650VTC4. This high current cell falls much shorter at 3A intermitted – even though its internal resistance is much lower. Its advantages are much higher currents (not shown here).

Table 4 is an updated version of table 2 from post #8.



Table 4

As you can see from diagram 9 and upper part of table 4, at room temperature LG ICR18650E1 outperforms all other cells for such an intermittent discharge. LG ICR18650D1 is slightly weaker, but also surpasses the Panasonic cells.

The cells from LG Chem suffer from cold environment more than the cells from Panasonic do. Diagram 10 and the lower part of Table 4 show, at cave temperature LG ICR18650E1 lost its leadership in the left part of the diagram – nevertheless it holds it on the right part. LG ICR18650D1 even falls shorter in the right part of the diagram.

Possibly, at even colder environment (below 5°C) also ICR18650E1 may fall shorter than the Panasonic cells do. I have not tested it, but I predict it because they suffer more from lower temperatures.

I have studied all measurements from Henrik K. Jensen, Stephan Knaus and others, which all base on 23°C. I conclude that the actual violet LG ICR18650E1 will be best suited for my project of helmet lamps for alpine caves.

If someone knows any better proposal, post please!

Findings

• Be careful, any well-known battery may be modified without announcement (e.g. LG ICR18650E1 sea green and violet)
• A cell fitting best at standard temperature (23°C) may not fit best at lower temperature automatically (e.g. ICR18650D1)
• Intermittent discharge is harder than continuous discharge, since the cell gets no chance to worm up internally. The lower the environmental temperature, the stronger this effect.

@Tobias Could you please clarify precisely what you mean by depolarization. This term is often overloaded. In particular, do you intend it to include ohmic contributions or not?

The voltage sag under load depends upon duration of pulse, as can be seen in diagram 2.

When the discharge pulse starts, the voltage sags instantaneously. During the pulse it further sags contiuously. After the discharge pulse has terminated the voltage jumps back instantaneously but needs some additional time to climb back to its initial value.

This behaviour is well known from primary cells but generally aplies for all batteries.

You can simulate depolarization by a circuit with two batteries in parallel:
The overall behaviour is an instantaneous voltage sag gouvernd by the impedance of the second battery and a further voltage slope governd by the first battery mainly.

Without depolarization effects the voltage would be nearely constant during a discharge pulse.