Why is my super-capacitor self-discharging so fast?

Question

I'm experimenting with a night light powered by a mini solar panel charging a super-capacitor instead of battery. (I know, I know. It's not recommended for this usage but I want to try it anyway). It works but the capacitor discharges too fast even before it is actually used. I tested the super-capacitor alone to see what is happening.

I have a 5.5V 4F super-capacitor very similar to the one in the image.

I charged it using a 18650 battery through a 5V step up converter for a couple of minutes. Then I disconnected it from power and connected it to a multimeter. I turn it on and measure value during the day.

This is what I've got:

When I disconnect power, it drops immediately to about 4.5V, then rapidly discharges at the beginning. I read many articles about discharging. It is hard to find info about this but according to what I learned it should drop about 10-20% per day. In this case it drops 1.6V (36% of charge, 29% of total capacity) in 8 hours.

I expected self-discharge but I don't understand why it is so fast. It does not help to add four parallel supercapacitors for a total of 16F. The behaviour is similar. I need about 2.6V to run the circuit but it discharges even before it should do anything.

---

• Is this discharge normal?
• Is it possible that the capacitor is low-quality with high leakage?
• Do I understand this topic correctly?
• Did I miss any important info about super-capacitors?
• Can you explain why this fast discharge is happening?

---

Update

Thanks for comments.

---

@Russell McMahon Exact product link is no longer available but it is probably this one Aliexpress GOHJMY V type 5.5V . I couldn't find datasheet for this, seems unbranded to me.

I found some branded expensive ones but they also have about 1000-1500 hours life, so about 5-8 years at 30°C. Is that too short?

---

@Gos I do what you suggest. Currently trying to "wake up" the cap by pumping 5V for hours. I am not sure if that's gonna work for long but let's try it.

---

@Tim Williams Thanks for exhausting answer. I like this kind of replies. I had to read it several times to get the idea. I think I get it mostly.

How do I understand

What I found about similar V-type coin cells they use to have purpose as "backup power supply" similar what you said. I understand that as these super-capacitors are exposed to power persistently during circuit operation and it's purpose it provide power when energy drops out. So they can have very-slow charging, pumping to every corner of capacitor under high impedance and then when needed, slow discharging provides long backup power supply. (Is that correct?)

About my cap

I tried to check ESR of the capacitor. I am not mistaken, it has about 350-450 Ω. (calculated by applying power and measure voltage with current). Pretty high I guess.

Searching for solution

I was looking for low-ESR types, it seems they have it about 0.1-0.6 Ω. I guess you were mentioning these "fast" ones right? Usually have fast charging and that I think is what I need. There is not enough sun during the day, I don't have eternity to charge it. I also found some "hybrid" types that should have high energy density similar to battery but durability of the capacitor. It seems these fast caps are not cheap but sounds like what I want so maybe I give it a shot.

Conclusion

I can say overall project was successful, cap nicely charge during the day and discharge during the night.

My project is not finished yet but I can close this one with solution. It seems capacitors I used are not suitable for application like I have. I replaced 4x 4F caps with 1x 100F (SECH C22S-3R0-0100) that has very low ESR (13 mΩ). I would interpret that as - able to charge and discharge much faster (don't accidentally bridge terminals, it will discharge in a second ... trust me 😑). Capacitor's voltage during day/night cycle ranges from about 2.2V to 2.65V. These are summer days, sun is strong but I think I can increase load for brighter led.

Details about solution. What changed:

• Size. It's huge, not easy to fit under mini solar panel. It is similar to stacked 4x the old caps but 6x more capacity
• Cap's voltage. Maximum voltage is now 3V. I need to use 3.3V step-up with diode to safely charge from panel. I also made protection using TL431 to be sure it never exceed 3V for safety
• Output voltage. Discharging circuit drains capacitor using another 3.3V step-up. It increases and stabilises voltage of linearly dropping cap's power and effectively drains all power

Thank you @Tim Williams it helps me a lot.

Answer 1

You need to charge it for more than a few minutes.

The "self-discharge" curve you read, is also the charging curve (e.g. current flow for step voltage change).

These components have a diffusion characteristic, meaning the impedance varies as \$Z \propto \sqrt{\frac{1}{f}}\$, or similarly over time (hence the curve being fast at first, gradually slowing down; but not following a hyperbolic or exponential curve exactly).

In other capacitor types, this effect is small, and bears the name "dielectric absorption". An aluminum electrolytic capacitor, suddenly discharged, might "recover" a few percent charge in some hours' time, but not usually anything dangerous. In the double-layer capacitor, the dielectric itself (the ionic double layer) moves, migrating through myriad microscopic pores in an activated charcoal electrode; the ESR through any given pore can be almost arbitrarily high, so it takes a long time to charge up the most distant surfaces of the capacitor. Meanwhile the nearby face, and the larger and shallow pores, charge up quickly, so you might deliver the bulk of current flow in a short time period, but charging (or discharging) current keeps on going for a long time after. Slow and steady wins this race!

Typically, minimum leakage current is achieved after "soaking" for several days at given voltage. Only then is leakage sensible in the usual way, for example (i.e. increasing exponentially as rated voltage is approached).

Since impedance changes so oddly with frequency, the capacitance also changes: at short durations (seconds to minutes), you may find it looks like a capacitor a fraction of the nominal value, with as much ESR (i.e., Xc = R, the characteristic of a pure diffusion (Warburg) element). The rated capacitance is only reached at the given test frequency. Ultimate capacitance, over the longest time scales (days to weeks), may even be higher than rated -- they just don't care to test the thing for so long!

Another way to put that is, the amount of capacitance, the active surface area, varies with frequency; the high ESR through narrow pores effectively blocks further area from being active at a given frequency. Given more time, more surface is active, and the measured capacitance rises at low frequency.

This also means that a fast discharge is not very practical: the capacity is greatly reduced. How fast, depends on efficiency/capacity requirements, and the type. Low-ESR variants are available (e.g. Maxwell BOOSTCAP, having useful capacitance in the seconds to minutes range), but very slow types also exist.

Beware that, capacitors in this coin-cell style, may only be suitable for CMOS memory backup applications -- they typically have quite high ESR, in the kΩ. Slow types. Even for something like a solar night light, you may want a faster kind; check the datasheet to confirm.

Answer 2

Just a guess, Try checking the polarity of each capacitor. Even if 1 capacitor's polarity is wrong, it will discharge fast.

Edited - Try changing +ve and -ve terminal and see in which case the capacitor behave normally instead of discharging fast, if the problem was actually polarity.