Trying to build a capacitor charging circuit

[FelixP]

Hi, I'm trying to build a capacitor charger that can charge a capacitor to a higher voltage than the battery voltage I have. When reading up on charging capacitors, the simplest method seems to be wiring something similar to the capacitor charger for a camera flash. Unfortunately, a camera flash will charge a capacitor to a much higher voltage than I am looking for.

Anyway, the capacitor charger doesn't need to be based on a camera flash, it's just an example. Here's what I'm looking to do:

The goal of the capacitor charger is to charge a 10000uf 50v capacitor to 18-25 volts as quickly as possible and with a fairly compact setup of components. The capacitor is then going to be discharged manually through 4 short pieces of kanthal resistance wire, heating them incredibly fast in a tiny fraction of a second. I'm planning to make an initial test by using one piece of kanthal wire, a 2000uf capacitor, and several 9 volt alkaline batteries first (18v at 1amp draw using 4 batteries, two series pairs wired in parallel), to make sure that the measurements I calculated are correct.

The best means to use this in a compact space is to use a LiMn battery. These are made to be used for "sub ohm" electronic cigarettes, and can provide 20 amps of current continuously, and 35 amps in short bursts (I am planning to only draw 20 amps at a time, so as not to drain the battery(s) as quickly). LiMn batteries provide 4.2 v when fully charged, and 3.6 v through most of their charge. They should not be used in series, hence why I need a means to bring the voltage up (also to save battery space since I will need only 1 or 2 at most).

I recognize this is a lot of current, however I believe that by keeping well below 50 volts, human skin resistance should keep me from any fatal accidents. By multiplying the battery's voltage by 5 or 6 times I should have the correct voltage required to charge the capacitor for my needs, as well as keep well below its safe working voltage.

The thing that's hardest for me to understand is transformers, what their properties are, and how to modify their properties to change the behavior of a circuit.

 

[Averagesupernova]

Look into marx generators. They typically use spark gaps but you could replace the gap with switch contacts that trigger at the same time.
-
Have a look here: https://en.wikipedia.org/wiki/Marx_generator

 

[Baluncore]

Welcome to PF.
Convert the DC to a square wave then use a diode capacitor ladder multiplier.
https://en.wikipedia.org/wiki/Cockcroft–Walton_generator

 

[Jeff Rosenbury]

How hot is hot? Are you planning on melting the wire?

How many cycles of the capacitor do you need? Just the 4? Or are the wires in parallel? And what sort of charging time were you thinking was fast? Under a second? Under a millisecond? Under a minute?

What method you should use depends in large part on what you are trying to accomplish.

 

[FelixP]

How hot is hot? Are you planning on melting the wire?

How many cycles of the capacitor do you need? Just the 4? Or are the wires in parallel? And what sort of charging time were you thinking was fast? Under a second? Under a millisecond? Under a minute?

What method you should use depends in large part on what you are trying to accomplish.

Wires will be in parallel.
Charging time should be under .2 seconds.
By my calculations, each wire will heat up by about 448 degrees C within 1 millisecond when discharged (assuming resistance is low enough) which (I think) is more than enough for what I need.

Look into marx generators. They typically use spark gaps but you could replace the gap with switch contacts that trigger at the same time.
-
Have a look here: https://en.wikipedia.org/wiki/Marx_generator

The only issue with the Marx generator is that it would require 5 or 6 10000 uf capacitors, which is rather bulky, and kind of costly (10 of these capacitors is 80 bucks). It's definitely a simple, and probably efficient solution, but it's not portable or cost effective enough.

Welcome to PF.
Convert the DC to a square wave then use a diode capacitor ladder multiplier.
https://en.wikipedia.org/wiki/Cockcroft–Walton_generator

The questions I have about this are:
1. how do I convert the DC battery to an AC or pulse? I assume with a transformer, but I don't know how to use one.
2. for the capacitor I'm charging, where do I connect the terminals? There's only 1 "DC out" point on the generator.

 

[Baluncore]

1. how do I convert the DC battery to an AC or pulse? I assume with a transformer, but I don't know how to use one.

1. No transformer. Just two mosfets, one N-chan the other P-chan, driven by a 555 timer.

2. for the capacitor I'm charging, where do I connect the terminals? There's only 1 "DC out" point on the generator.

2. One side of the cockcroft-walton generator is ground, the other is fed the square wave. DC output is between ground and the series column of capacitors not connected to the square wave generator. That column can be the output capacitor that you are charging.

See also; https://en.wikipedia.org/wiki/Voltage_multiplier

 

[FelixP]

I just had someone explain to me that if you have a transformer with a higher number of wrappings around the secondary coil, then the voltage will be increased by the proportion of coils. So would that mean if I ran the battery voltage through a transformer where the secondary coil had 5-6 times more wrappings, and the secondary coil were connected to the capacitor, that it would provide the voltage I'm looking for?

Or would this be less efficient?

Also, if so, would I have to make it myself, or is there one out there that provides this ratio already?

 

[Jeff Rosenbury]

Transformers work with AC voltage only. Of course since your circuit will charge in 200 mS, that's AC enough. But that will require switching, so some sort of switched solution is a must.

Transformers are somewhat heavy. DC-DC switchers are typically preferred. There are some nice LED power supply circuits which provide voltages and currents around what you seem to need largely off the shelf. (You will still need to run the numbers and build controls.) Transformers are also doable. Which I would go with depends on cost points. If you value your time, I would try the LED. If this is a learning project and you want to save money, try the transformer. For large scale production, DC-DC switchers will likely be cheapest.

 

[FelixP]

Maybe my google-fu is lacking, but for DC-DC switchers, there seems to be two types: Ones that have the right voltage, but they are limited to 5 Amps of input current at most, and ones that require 12 volts input current but can support high current. The LED power supply circuits I saw needed way higher voltage.

For a transformer, how bulky would it be? They don't seem to be that big. Nonetheless, if I wanted to use the transformer, I'd need to switch the DC current to AC, so, I guess the question is how to do that (also how do I determine the max amperage of a transformer).

I looked up transformer cores, and if I only need to use 1, I could probably use a 60mm diameter toroid core, assuming that's the right shape for this application. I don't know if I would need a different core shape.

 

[DrZoidberg]

So you want to charge the thing in less than 0.2 s and discharge it in 1 ms. How often do you want to do this per day? Maybe 9 V batteries would be a good solution, then you don't need any transformer at all. Also instead of storing the energy in a capacitor it is also possible to store it in a toroid core. Meaning your transformer and your energy storage device would be the same component. A 60mm toroid may be able to store several Joules of energy, enough for your purpose.

 

[Jeff Rosenbury]

Capacitors are measured in Farads. 1 Farad will hold 1 Coulomb at 1 Volt. 1Amp is 1 Coulomb/Second.

.01F×30V=.3C; .3C/.2s=1.5C/s=1.5A.

So to charge a 0.01 F (10,000 µF or 10mF) capacitor to 30V requires 0.3 Coulombs. To do it in 0.2 S requires 1.5Amps at 30V, for a constant current source. I saw an LED driver that output 1.2 amps but had 4 channels, so could charge several, slightly smaller capacitors. I'm not sure about all the details though.

If you go the transformer route, you could wind your own transformer on a nail bent in a circle (or other closed loop iron). As I said, this is more of a learning project. It won't be that efficient.

Most DC-DC switching supply chips have an internal switch that limits their current draw. But most of the time this switch can be used to switch a larger switch for more current.

There are lots of switchers out there. I've never found a switcher I'm in love with. The ones I've used were finicky leaving poor board yields (i.e. many supplies needed to be hand tuned to work.) Perhaps someone else can suggest one? If not, here's one that might work. It might need an exterior PNP switch since there's an internal current sense resistor.

 

[Baluncore]

You are new to electronics design. Switching converter design is beyond many with experience, so you should consider buying a module.

Start by searching DX.com for ' boost converter '. You will find things like this …
http://www.dx.com/p/dc-3-30v-to-dc-4-35v-adjustable-boost-converter-charger-module-148492#

 

[FelixP]

I've been using this site to calculate capacitor charging:
http://mustcalculate.com/electronic....php?vfrom=0&vto=29.99&vs=30&c=10000u&time=.2

And according to it it requires 12 amps to charge a 10000 uf capacitor to 30v in .2 seconds.

What properties of a transformer dictates its efficiency?

 

[Averagesupernova]

The only issue with the Marx generator is that it would require 5 or 6 10000 uf capacitors, which is rather bulky, and kind of costly (10 of these capacitors is 80 bucks). It's definitely a simple, and probably efficient solution, but it's not portable or cost effective enough.

I think the capacitor issue is unavoidable since you need a lot of current very quickly. You WILL be dumping money into capacitors in order to achieve this. You will need to look into the ESL and ESR of the capacitor(s) in order to be able to get the power into the wire. You are basically driving a short circuit and your source impedance needs to be much better than a short circuit. Charging the capacitors is secondary. Whatever energy you dump out of the capacitor you need to dump back into it in about 200 mS according to your spec.
-
Putting the capacitors in series as in a marx generator multiplies the ESR and ESL by the number of caps as would any voltage multiplier scheme utilizing Cockcroft–Walton whether they are used to charge the main capacitor or used as the main capacitor. You haven't mentioned the duty cycle and this is very important in determining how to best charge. I am assuming since you want it charged in 200 mS your duty cycle is pretty high. It would be a lot easier to charge over several seconds.

 

[Jeff Rosenbury]

I've been using this site to calculate capacitor charging:
http://mustcalculate.com/electronic....php?vfrom=0&vto=29.99&vs=30&c=10000u&time=.2

And according to it it requires 12 amps to charge a 10000 uf capacitor to 30v in .2 seconds.

What properties of a transformer dictates its efficiency?

That website is for a voltage source with the current limited by a resistance. A constant current source is different. Try using a 30V supply, but only charging to 20V. With a 100Ω resistor I got 300mA. (And 9W through the resistor -- ouch!)

Most off the shelf power transformers use special alloys and construction that limit the formation of eddy currents. There are some other potential problems (hysteresis, air gaps), but that's the big one.