I have purchased three sets of supercapacitors over the last few years. The first two pairs of Supercaps were 350F and 400F units that I bought directly from Digikey. I believe that they were priced somewhere around $12 per supercap (here's a link to the Digikey webpage for the 400F unit). The last two pair of 100F supercaps I bought from eBay, for about $3 per pair. I thought that the eBay stuff was a great deal...until today.

I was attempting to test the 100F supercaps in a dual supercap UPS configuration and write some new code for the UPS in general. I was getting squirrelly results. The UPS would shutdown properly, but after a few minutes it would attempt to turn on the booster, which would cause the Raspberry Pi to begin its boot sequence, but there wasn't enough juice remaining in the supercaps and the output voltage collapsed after a few seconds. This process would repeat every few minutes for what seemed like hours.

Bad Supercaps

I noticed that there was a significant difference between the two supercap voltages: 1.5V and 1.1V. At the end of each false startup attempt the larger voltage would decrease by less than 100mV, but the smaller voltage dropped to about 0.8V but then climbed slowly back until the total of the stack reached 2.65V and the LTC4041 would fire up the booster again. This behavior (behaviour?) was not supposed to happen.

I extracted the two 100F supercaps from the breadboard and charged each of them with a power supply to 2.25V with a current limit of 0.5A. I let the charge current drop below 10mA and then I discharged each capacitor with a 10Ω resistor to 1.5V. Here are the results SCAPB (the bad one):

SCAPB dropped to 1.5V in about 2 minutes. Note the large step at the beginning of the discharge and recovery cycles. The voltage eventually recovered past 2.0V after a 20 minute wait.

SCAPA faired better:

SCAPA has a much smaller step at the beginning of the discharge, takes longer to reach 1.5V, and recovers by only 165mV (it was still increasing after 22 minutes). But even this performance is pretty bad -- two supercaps with this behavior would still creep up enough to cause the LTC4041 to enable the booster when it should not.

So I dug one of the 400F supercaps out of a junk box and repeated the experiment -- this time using a 2Ω load for discharge:

This is what the behavior should look like. There is a sharp drop when the >1A load is applied -- attributed to the ESR. There is another sharp rise when the load is removed -- ESR again. But there is no significant ballooning of the voltage afterward. 

I'm tempted to spend a few $$ to obtain some legitimate supercaps in lower Farads to see if there is a difference between Digikey and eBay. I need a 100F supercap ( a real one...not from eBay) to power my heating system. I also would like to see how a couple of 20F supercapacitors from Vishay (Digikey) compare to the eBay stuff. Stay tuned.