High Voltage Capacitor charger

[alec_t]

The vdivider circuit is quite precise, for example, a battery that got 12.18v, with the Vdivider connected i will get close to 122mV.

The voltage divider may well work fine at such a low voltage, but not necessarily at high voltage, due to leakage. How is the divider constructed?

 

[transforman2]

The voltage divider may well work fine at such a low voltage, but not necessarily at high voltage, due to leakage. How is the divider constructed?

 

[alec_t]

What is the manufacturer's specification for the working voltage of the 40Meg resistor? What is the resistor mounted on? What is it housed in? What is its environment (dehumiifed air, tropical, ....)?

 

[Beau Schwabe]

What is the reverse breakdown voltage of the diodes used in your Multiplier?

Also, you can't really place capacitors in series and expect them to charge evenly, so there could be issues there. If the dielectric arcs, the arc acts like a short, causing an avalanche effect to the other capacitor which in turn will arc. <-- (The basic triggering principle of a Marx generator) If the dielectric arcs, now you have a burned (highly resistive path), or a cap that's damaged and/or works but has a high leakage.

 

[transforman2]

What is the manufacturer's specification for the working voltage of the 40Meg resistor? What is the resistor mounted on? What is it housed in? What is its environment (dehumiifed air, tropical, ....)?

I am using x40 of 1MΩ to evenly distribute the voltage.

What is the reverse breakdown voltage of the diodes used in your Multiplier?

Also, you can't really place capacitors in series and expect them to charge evenly, so there could be issues there. If the dielectric arcs, the arc acts like a short, causing an avalanche effect to the other capacitor which in turn will arc. <-- (The basic triggering principle of a Marx generator) If the dielectric arcs, now you have a burned (highly resistive path), or a cap that's damaged and/or works but has a high leakage.

The diodes i am using for the multiplier are UF4007 diodes

I charged them to 2kv, i have 2 in series and each can handle 2kv easily. The internal 1M resistors will distribute the voltage almost evenly right ?

I got one more original cap so i may connect it in paraller and charge to 2kv to see in anything happens. Maybe to 2 caps from ebay are faulty.

 

[Beau Schwabe]

How fast after charging are you trying to discharge the caps and what are you expecting out of them when you discharge them? .... ONE 1uF cap charged to 1kV will drop to about 200V after about 2 seconds with a 1M leakage resistor.

 

[transforman2]

How fast after charging are you trying to discharge the caps and what are you expecting out of them when you discharge them? .... ONE 1uF cap charged to 1kV will drop to about 200V after about 2 seconds with a 1M leakage resistor.

They are actually charging all the time, i just adjust the voltage i want to charge and i stay there. I do not close the charging circuit while i am trying to discharge them.

 

[alec_t]

How are you discharging the caps?
How are you monitoring the discharge?

 

[transforman2]

I tried to discharge them through a thin wire with 5M coaxial cable and a remote relay at 2kv. I do not monitor the discharge somehow, i just expect to something happen to the fine wire, explode or melt or something. I do not even here anything like arcing the relay coil or something, its like dead.

 

[Les Jones]

How thin is thin ? (With respect to the wire you are trying to melt.)
I suggest that you connect an oscilloscope to the output of the potential divider and look at the wave for. A DSO will display this directly if triggered from the same signal that drives the relay. With an analogue scope you may have to make a delay circuit so the scope is triggered just before the relay.

Les.

 

[transforman2]

I do not have an oscilloscope

 

[Beau Schwabe]

You need to scope the discharge curve ... I'm willing to bet, that if the internal leakage is 1Meg (parallel resistor across cap)then the actual series resistance might be high as well.

I don't know what gauge of wire you are using, but in my own tests, you'd be surprised what a 38 gauge wire can take ... It can survive my 750kV Marx generator with several 4nF caps at each stage in series.

 

[transforman2]

Hmmm Interesting. Beau any idea how i can test them ? It is strange that i do not hear anything after i switch the relay. Even if nothing would happened to the wire, an arc should be formed and heard from the relay switching by short circuit and discharging the caps right ?

 

[Beau Schwabe]

... any idea how i can test them ?

... A Scope to measure the discharge curve

... an arc should be formed and heard from the relay switching by short circuit and discharging the caps right ?

... not necessarily if the internal resistance of the caps are high enough

 

[Les Jones]

I remembered that I had a capacitor removed from a microwave oven so I though I would measure it's ESR just for interest. It measured 1.4 ohms. The capacitor is 0.98 uF and 2100 volts AC working. (So it should be OK up to a bit over 2900 volts DC.) If the OP's capacitors had a similar ESR then that should not be a problem. I will try asking the question again. What is the diameter of the wire you are trying to melt ? (Any units will do, mm, inch or wire gauge.)

Les.

 

[transforman2]

I really dunno exacty the diameter, i just put the finest wire i can find. I tried to take a photo but is very fine to get it. It is attached to a 2 pins Header Strip like that attached below. You can see the wire when it shine, it is really fine. A challenge to solder it. It is almost the same wire that some 12v relay coil.

 

[Beau Schwabe]

That's about 38 Gauge wire ... the fusing current for 38 gauge wire is close to 2.6 Amps..... you need to go with thinner wire or a bigger cap that has more energy to "pop" the fuse you made.

Method to determine the wire gauge from image:
1) The pitch of the header is 1/10th of an inch ... the number of pixels from one edge of the connector to the other is roughly 27 pixels
2) the pitch of the wire is roughly 1 pixel (better quality image would reveal a better number ratio)

270 pixels = 1 inch ... so 1 pixel = 0.0037 inches or 0.09 mm .... Chart says 39 gauge (See reference) ... but wire gauges are usually even numbers so I'm going to assume 38 gauge.


Reference:

Fusing Currents versus American Wire Gauge for copper, aluminum, iron, and tin (Cu, Al, Fe, Sn), wire melting currents

Wire size table and fusing currents for design engineers including copper (Cu), Aluminum (Al), Iron (Fe), and Tin (Sn), wire melting amperage

www.powerstream.com

 

[unclejed613]

I am using x40 of 1MΩ to evenly distribute the voltage.

you might want to check a data sheet for the resistors. many resistors have a voltage limit. there's an example in this data sheet. another thing that happens and throws high voltage measurements out of whack is corona discharge, especially from any sharp points on solder connections. i knew somebody that bought a He-Ne laser, and it came with a 10kV power supply and ballast (allowed 10kV to start the tube, then drops to about 2 to 5kV to keep the tube running). he didn't have the proper connectors, so he just twisted wires together. he couldn't figure out why the tube wouldn't start if the twisted connections were out in the open, but everything worked fine if the twisted connections were stuffed into a couple of screw holes in the wood frame he made for the laser. corona discharge from the sharp points on the connections was enough leakage current to create a large voltage drop across the ballast resistor, and so there wasn't enough voltage to fire the tube.

 

[transforman2]

unclejed613 i was thinking to add a switch to voltage divider in order to limit the consumption. Currently i can not test anything because i moved but i will post back when i test few things. I have 2 more caps so i want to remove the old ones and put the other two to test the caps/circuit. I will add a switch to voltage divider to test it too as i said above. I will may remove the coaxial cable too, just for testing it.

 

[gary350]

If there is no time limit to how many seconds it takes for the capacitor to charge this will work. I used this voltage multiplier to charge 30 capacitors in parallel. 1N4007 diodes are 1 amp current needs to be limited to about .9a. I recall 2 stages will double the voltage. 30 stages makes a long spark. Diodes are rated 1000v. Been 15 years since I did this I think my power supply was 600vac