Supercapacitor PSU how to charge quickly?

[neazoi]

Hello,
I would like to experiment with supercapacitors to make a small PSU.

I have two 400F 2.7V supercapacitors
I also have a 45VA 1.42V-0V-1.42V transformer

Could you suggest me the fastest way to charge each of the two supercapacitors?

I was thinking a regulator would not be a good choice, since it will limit the initial charging current.
I was also thinking of a power resistor to limit the current, but when the capacitor is half charged, then it will take ages to charge more unless I replace this resistor with a lower Ohms one.
What if I connect the capacitor directly to the transformer (through a diode or bridge). Will it draw as much current as the transformer can supply, or will the transformer be damaged from the large initial current drawn by the supercaps? (assuming the rectifying diodes amperage is rated more than that of the transformer)

 

[betwixt]

Possibly a low value NTC thermistor in series with the capacitors would work. One that limited the inrush current for a few seconds then dropped in value to keep the current high.

Brian.

 

[neazoi]

Possibly a low value NTC thermistor in series with the capacitors would work. One that limited the inrush current for a few seconds then dropped in value to keep the current high.
Brian.

This thermistor should be able to handle a large current (say 20A), for the capacitor to charge quickly.

According to the calculator https://www.supercircuits.com/resources/tools/volts-watts-amps-converter My transformer can deliver about 20A at 1.42V (assuming that the VA label refers to the low voltage wiring and not the 220V one) and about 10A at 2.84V (if I ignore the center tap...?)

Do I need to obtain a thermistor to limit the primary (220V thermistor type) current, or is it better to put the thermistor after the secondary (low voltage thermistor type)

Have you got any clue what NTC thermistor to look for? Something like this maybe? **broken link removed**

 

[betwixt]

As a rough guide, consider VA as being Watts and the transformer as being 100% efficient. It isn't of course but it's near enough for a calculation. I = W/V so the 1.42 should be able to deliver around 45/1.42 = 31.7 Amps.
There is very little technical detail in the link you found but assuming the 1 Ohm is at standard 25C, it would try to drop 31.7V at switch-on which obviously isn't possible. It would limit the current far too much.

It is more sensible to limit the current in the transformer primary where it is at a more reasonable level. Under full load the transformer would draw around 45/220 = 0.2A. Knowing what to limit the current to is the problem and some guess work is needed. I would suggest an NTC thermistor with a cold resistance of around 50 Ohms and a hot resistance of 10 Ohms or less would work. Without exact transformer specifications, especially in overload condition, it is almost impossible to be more precise.

Brian.

 

[neazoi]

I would suggest an NTC thermistor with a cold resistance of around 50 Ohms and a hot resistance of 10 Ohms or less would work.

Brian.

Do you refer to 250V, so it can be user at the primary right?

 

[betwixt]

Yes, the very high current in the secondary would force you to use very low value resistors, maybe 0.01 Ohms or less and I doubt you could find a suitable thermistor.

I would suggest you use the whole secondary to give 2.84V RMS, this is around 4V peak and a bridge rectifier will drop at least 1.2V from that bringing your capacitor voltage down to 4 - 1.2 = 2.8V. In reality the voltage drop in the rectifiers will be a little over 0.6V per diode so you will get slightly less than that but still close to the capacitor maximum rating.

The choice of thermistor is difficult, when the capacitor is fully charged the transformer current will be almost zero, in fact only enough to overcome the core losses (a few mA). A fully discharged capacitor will appear almost as a short circuit and completely overload the transformer making the current very high. Ideally what you want is one with a value that limits the primary power to it's manaufacturers rating, in other words 45VA while starting to charge and drops to near zero as the current decreases. Unfortunately, transformer manufacturers do not specify such things as primary current with the output shorted so some guesswork and experimentation is needed.

There is a 'dirty' trick you could use, at least to see what happens, that is to wire a 100W incandescent light bulb in series with the primary. That is, if you can find one these days! Hopefully is will light up when charging the capacitor and dim as the current drops.

Brian.

 

[neazoi]

Thank you Brian, I will try these points and let you know.

Any quick guess about the Ohms value of the thermistor at the primary?

Finally, what if a thermistor is not used at all, will the transformer get damaged, or it will self-limit the current up to the point that it cannot supply more?

 

[betwixt]

A guess would be one measuring 50 Ohms at 25C but I could be wrong.

Whether the transformer gets damaged is unpredictable, for a while it's secondary would be almost shorted out so it would be as overloaded as it's possible to get. Most would buzz and get hot but how long they could hold out for is really down to their design. The most likely scenario would be the primary winding passing enough current to 'fuse' the wire and make it open-circuit. You should also consider that the current into the capacitors might have to be limited for their own protection. 400F is a huge value to charge up!

Brian.

 

[PA4TIM]

I would use a resistor to limmit current and when the cap is charged enough and current is come in the range your supply can deliver you bridge the resistor with a big relay. Use a comparator with a potentiometer and a reference voltage to measure the cap voltage at what you want the relay to switch over. I would use a good clampmeter or other way to measure current to measure if your transformer is in the dangerzone. If so you can turn the potmeter and switch the resistor back in.

If you have 5 caps in series you could use a car battery to charge them.

Be carefull to use supercaps as a power supply, there is no current restriction. There are youtube videos where a guy uses 6 supercaps in series to replace his car battery and he starts his car on them. He even lets the car stay for a weekend, play the radio for half an hour and other tests. I think his purpose was to have a battery replacement to start more easy in cold winters.

 

[neazoi]

Yes this is what I did initially. With a 5A max bench psu I limited the current usign a 5R 5W resistor, then as the voltage on the cap increased, I added more 5R 5W resistors in parallel. It took about 20-30mins to completely charge the 400F. I know how to do it usign a comparator but charging time will never be as low as with a combination of a suitable thermistor and transformer. That is because the thermistor has an infinate state of "current limiting values" which depend on the current drawn by the load, at any time (there may be some thermal hysterisis but let's ignore that). The comparator on the other side has only two states, the one below threshold and the one above. It simply cannot monitor currentdrawn and limit the current accordingly, at any time.

- - - Updated - - -

A guess would be one measuring 50 Ohms at 25C but I could be wrong.

Whether the transformer gets damaged is unpredictable, for a while it's secondary would be almost shorted out so it would be as overloaded as it's possible to get. Most would buzz and get hot but how long they could hold out for is really down to their design. The most likely scenario would be the primary winding passing enough current to 'fuse' the wire and make it open-circuit. You should also consider that the current into the capacitors might have to be limited for their own protection. 400F is a huge value to charge up!

Brian.

Datasheets show two thermistors at the primary, one at the phase and another at the neutral. Do I need two of them really? I thought one would suffice.

 

[betwixt]

It will. If one is in phase and the other neutral they are in series anyway so you could replace them with a single component. I would guess the datasheet is trying to double their sales!

Brian.

 

[PA4TIM]

I do not think a termistor will work. It starts high, the current heats it and resistance drops, but current will drop as voltage rises and the termistor then cools off and increases the resistance so it will take for ever to charge. I reform many many caps and allthough they are not supercaps they behave the same and I monitor voltage. I see the voltage rise and current dropping. I set the voltage at the max voltage the cap can have and feed it through 8 resistors in series. If a certain voltage level is reached current drops to zero and I switch that resistor out the chain so now 7 resistors do the job etc. I look for leakage so I can not speed up this way but it is easy to change it to set the levels for keeping the current rate around maximum


The comparator is to monitor the voltage, not the current. And you have to monitor it only a few times to check if you have chosen the right time and resistor values. If you use a few quad comparators you van do it in 8 or more steps. You can make a "digital" MOV this way that alow you to source the max current your transformer can deliver
.
And if that takes 20 minutes a comparator will indeed not speed this up, you then need a more powerfull supply but probably still limmit the first inrush.

 

[betwixt]

PA4TIM may be right but I don't think it will take forever to charge. It will certainly take longer but that is intentional when limiting the rate of charge. The thermistor will heat up during the initial charging surge but as the charge current drops, so does the need to take so much primary current in the transformer. So if the resistance increases, or at least reaches equilibrium point, the voltage to the primary does not necessarily prevent full charge being reached. Consider that at full charge, the secondary current will be almost zero so the primary current will also be very low.

Another disadvantage of the comparator method is the value of resistors and rating of switches you have to use. We are talking in excess of 30A initial charging current but at very low voltage so the resistor values would only be a few milliohms. It would be relatively complicated to monitor capacitor voltage and switch resistors into the primary side of the transformer where the values are easier to handle.

Brian.

 

[neddie]

What is the ESR of the Supercaps?
Some of theses caps have ESR's in the 10 of ohms range , as they are not meant for high current discharges , but for memory retention purposes.
If this is the case , then trying to limit inrush current is pointless :0)

 

[betwixt]

I think the ones Neazoi has are the ones that have super low ESR, only one or two milliohms. They are physically bigger and can deliver hundreds of Amps. Most of the memory retention ones are much lower in value, typically < 1F and usually rated at 5 or 5.5V. These are 400F, 2.7V ones.

Brian.

 

[neddie]

Cool.
How about some sort of BIG constant current source?
Supply voltage may be a problem there. :0(

 

[betwixt]

Not possible I'm afraid. Although the idea is OK, the capacitors are to be charged to full voltage (2.7V) and the transformer secondary is only 1.24V RMS from center tap to each end. Taking into account the drop in the rectifier there is only just enough voltage available and no overhead available for a regulator or any kind. A circuit providing constant RMS current in the primary might work but it starts to get complicated very quickly.

Brian.

 

[PA4TIM]

I have tried it in a sim using DC of 2.7V. If he makes the connections through thick copper bars and ESR , copper, connections etc are no more as 10 mOhm then it will draw 250A inrusch current and drops to 50A after 15 seconds for two x 400F parallel.
100 mOhm, will take around 500 seconds and starts at 30A and drops to 15A in 50 seconds, and 100 mOhm is not very much unless you use short soldid copper bars and very low resistance connections. And you have connections from transformer to diodes, to bars and to caps.

Your transformer is limited to around 10A if I understand you well, so if your total resistance is 200 mOhm the inrusch starts at 12A and drops to 4A in 200 seconds. The transformer will probably survive, but a complete charge will take you more as 500 seconds.
To be much quicker you need an other transformer.


I do not think you can get under 10 minutes with this transformer.

 

[neazoi]

What is the ESR of the Supercaps?
Some of theses caps have ESR's in the 10 of ohms range , as they are not meant for high current discharges , but for memory retention purposes.
If this is the case , then trying to limit inrush current is pointless :0)

the specs are here. **broken link removed**

- - - Updated - - -

To be much quicker you need an other transformer.
I do not think you can get under 10 minutes with this transformer.

Yes, it is not that big, 10 minutes is good enough, since I am thinking of charge the one 400F cap and use the other one alternatively.
More powerful transformer means more heavier construction, more expensive rectifiers and thermistors. Unless a huge current is drawn by the load, 10 mins charging time seems fair enough. After all my limitation is this transformer from an argon ion laser I had

Ok, I have some SG160 2.5R 15A thermistors which I can use on the secondary, and some SCK103 10R 3A ones, which I could use on the primary. I am not sure if the primary thermistor is suitable, or a more powerful type needed, so I would like to try the secondary ones. Another point is that the resistance of the thermistor before the supercap, may limit the voltage a little, so that the capacitor is charged in a more safe level (since on post #6 it seems charging voltage at the limits of the supercap.
After this test, I will probably finally order a suitable primary thermistor. How many amps should it be and how many amps should the fuse be?

 

[betwixt]

Looking at the photographs, I would be worried about passing more than about 10 Amps through them, the rivet tags are not suitable for very high current. The ones rated for high current have post and screw terminal connections.

A charged capacitor draws no more current than it's internal leakage so they will not prevent the voltage from reaching it's maximum, it just takes a little longer to get there.

I would suggest three or four of the SCK103 thermistors in the primary side the limit the current.

Brian.