Thread: Non linear capacitor scaling

http://www.mouser.com/search/Refine....38&Ns=P_SField

I was looking through their capacitor list and noticed something strange. Smaller capacitors often times are written to be small in dimension even to those tens of times higher in capacitance value.

Look at the details to this 50 farad capacitor: http://www.mouser.com/Search/Product...A1N59peg%3d%3d

18X40mm??? How is this possible? When you type even 1 farad capacitor into google images you see this: http://images.google.com/images?clie...-8&sa=N&tab=wi

Besides that I need a 15,000v capacitor at around 1.3micro farads. Where can I find this? I am actually considering putting together a capacitor; is this very difficult? From what I read all I would need is a plexiglass container, a glass sheet dialectic, two pieces of aluminum and connecting tabs.

Cold Fusion,

This will be a hard ask; 15 kV at 1.3 uF represents a very serious component!

The theoretical capacitance for a parallel plate capacitor is;

C = e * A / d

where
e = permittivity (8.85 * 10^-12 C^2 N^-1 m^-2 for a vacuum)
A = area of each metal plate in m^2
d = separation of plates in m

If you add a dielectric e.g. glass then it will increase e by its relative permittivity er

e' = er * e

er for glass is about 4.

(Ceramic capacitors use different mixtures to change er from 50 or so (NPO types) to 15,000 (YUV etc). High permittivity types show large variation with temperature and DC voltage (up to 1000%) and NPO types in contrast are extremely stable.

The withstanding voltage for a vacuum is "infinite" so d can be made small, but er=1. Air breaks down at about 1,000 Volts / mm. Glass is better so let's assume 1 mm glass dielectric with er=4.

The plate area A will be

A = C * d / (er * e)

set C = 1.3 * 10^-6, d = 0.001 m, er=4, e = 8.85 * 10^-12

=> A = 36.7 m^2

i.e. about 6 meters by 6 meters.

In a lab this is OK I guess. (I hope I got the numbers right!)

You can stack the panels in parallel of course




It "might" be possible to use several layers of plastic glad wrap and aluminum foil also.

A note of caution though - 15 kV at 1.3 uF has a very high likelihood of causing death if you discharge it through your body!

How did you get 10^-6? Wouldn't it be 10^-4? or am I wrong in assuming this is equivalent to 10x10^-4?


I actually received 3.61x10^-5 for the area......and a considerably lower number when I put in the 10^-6 and 10^-12.

A capacitor for my purpose is not entirely necessary..... as long as grounding either end can store enough electrons at 15,000v and 5ma for 30 seconds; what is the grounding limit of a piece of aluminum? How much does its size matter?

Thanks for the info.

Perhaps I got things wrong, so let's recalculate;
From

where











i.e. if each plate is square

Well same answer?

As for charge storage you can get an approximate "feel" for this using the charge storage equation for a capacitor




i.e.


If the current drain was constant, the capacitance needed for a discharge of given and would be

(note - typo corrected)

i.e.



If you need the voltage drop to be, say 10 times less over the period of 30 seconds, then C would need to be ten times larger, i.e. . This is a seriously large capacitor at 15 kV!

Yeah, you are correct, I thought for some reason that the base unit was in micro farads.

That is awful.....36 square meters.

When you have a capacitor and a resistor connected in series, will the capacitor see the source voltage, or the voltage after the voltage drop from the resistor?

I researched these super capacitors; it turns out that they actually do possess their rated values. They are achieved through new exotic methods and are believed by many to replace batteries in the future.

Just look at this baby! http://search.digikey.com/scripts/Dk...me=495-2066-ND

How did you get 10uf? Your equation looks a little off. Is the 30 above the 15,000?

Originally Posted by Cold Fusion

When you have a capacitor and a resistor connected in series, will the capacitor see the source voltage, or the voltage after the voltage drop from the resistor?

Remember the voltage divider equation for two resistors? When you have two resistors in series, the voltage across each is



Well, you can use the same voltage divider equation except you have to convert the capacitance into an impedance. The impedance of a capacitor is



Then the voltage across the capacitor is given by



Notice that for d-c (frequency f=0)the impedance of the capacitor is infinite, so all the voltage appears across the capacitor. the higher the frequency, the more voltage will appear across the resistor and less across the capacitor.

Yes, Harold14370 is correct but I wonder if Cold Fusion's question was aimed slightly differently,

When you have a capacitor and a resistor connected in series, will the capacitor see the source voltage, or the voltage after the voltage drop from the resistor?

If I am guessing correct, the resistor across the capacitor (equally in series) will "see" an ever reducing voltage over time as the capacitor discharges. The current drawn will fall off exponentially with time. The equation I presented (typo corrected Cold Fusion ) assumes a constant current and for short time periods this approximation is valid. It gives a useful "feel" for how horrible the capacitor dimensions need to be though!

Yes, super capacitors are intriguing and readily available for many Farads, often low voltage. They can certainly replace batteries and offer a potential 100% charge to discharge efficiency.

From memory lead acid car batteries are about 70% efficient (charge in, energy out), sealed "gell cells" are about 95% and NICADS are somewhere in between. Chemical process have losses from heat and in open cells, evaporation also.

The only minor hassle is that the super-capacitor voltage falls continuously with time. However simple switchmode regulators are everywhere these days so regulating a changing voltage up and/or down to a constant level is fairly trivial. I suspect we will see more applications as time goes on, beyond simple computer memory backup supplies to powering laptops or even electric cars.

Notice that for d-c (frequency f=0)the impedance of the capacitor is infinite, so all the voltage appears across the capacitor

Damn....and I thought I could use the two resistor trick to vastly lower the voltage across it. With infinite impedance that would be utterly useless. 36 square meters.......that would not only be very large, even with stacking, but the glass would likely cost allot. I guess I will have to think of something else more reasonable than that.

And capacitors (as far as I know) cannot explode like LI-ion's when faced with large impact forces. I might just have to buy a few and report back how useful they are

Out of curiosity, why do you need 15 kV Cold Fusion? I guess you would need to rectify it first to get DC suitable for a capacitor (unless you are using static charge from a Van Der Graph generator).

Old Black & White TV tubes have an internal capacitor based on an outer (graphite) coating and an inner conductive film. This is charged to about 15 kV from a EHT transformer and a vacuum diode. You could probably get some use from this.

Is it necessary for the apparatus to retain its charge for a long time if this charge is continuously replenished?

Can't say :wink: Though a few side uses will be with Tesla coils and such.

I have found a work around....so a capacitor is not necessary now. Though If I did use one, I would need it to take a DC current for a little while. I'm assuming you were under the impression that I was going to send AC at it, which would in a sense continuously replenish it.

I will definitely have future uses for a capacitor like this....

Another thing; whenever I wire a polar capacitor the correct way, with the negative terminal from the battery connected to the negative end of the capacitor, the capacitor heats up considerably and stars to make a boiling sound. This is obviously very very bad, so I wire it the other way, with the positive wire connected to the negative end. When I do this is works perfectly. What is with this? Are my bridge rectifiers and battery terminals all labeled wrong, or is it the capacitors I tend to buy?

When rectifying AC to DC, diodes and filtering/smoothing capacitors are used. At least for the applications that I have dealt with, 10,000uf capacitors have been used; often in synergy with 20,000 and smaller 1500 and 5uf capacitors. Though this has all been done at voltages under 100v. For my application I may need to smooth the signal out more than what diodes can accomplish. As we have discussed, even 10uf at 15,000 volts is preposterous, but that was for a very specific application separate of smoothing; now all I need to do is smooth the signal-will small values at high voltage (~.02uf, 10-15,000v) work to accomplish this? What about values in picofarads?

After rectification I usually see a fair voltage gain, between 20% and 40% usually. Will the same voltage gain occur when you rectify Hv? I went to a website that claimed the voltage is doubled....

Hi Cold Fusion

For my application I may need to smooth the signal out more than what diodes can accomplish

The diodes don't provide any "smoothing" instead they just provide a 1 way direction for current flow, hence "rectification". If the source is AC, then the current out from a diode is pulsating DC. The capacitors is then added to "smooth" these pulsations into a quasi DC output.

The amount of capacitance depends on the expected current drawn, the acceptable "AC ripple voltage" and the frequency of the AC input. Low voltage, high current requires large capacitor values. High voltage, low current requires less, however the energy = C*V requirement is much the same.

It may be better to cascade LC sections to reduce ripple - for example adding a series inductor after the first C and then adding another C to ground after it will form a "pi" low pass filter and reduce ripple (i.e. smooth the DC further". If the current drawn is low, then a resistor can be used instead of an inductor - large inductances are as hard to fabricate as large capacitors when it comes to HV.

Finally, you may see a higher DC voltage compared to RMS AC voltage. If the source is sinusoidal then this peak rectified voltage will be about times higher than RMS AC. However this depends on the wave shape etc. You don't get anything for nothing and drawing current will not allow better VI at DC than at AC.

Yeah, the full wave rectified voltage would have exactly the same rms value as the a-c input, so any difference your voltmeter reads is probably because it is multiplying the peak voltage by 0.707 when you have it set to read a-c.

I have settled on a 10kv transformer; my application will draw current between .1miliamps and 20miliamps, though I may only need to smooth the signal for amperage's under 1 miliamp. With .1miliamps, how much capacitance would be necessary? It will be at 60hz.

the full wave rectified voltage would have exactly the same rms value as the a-c input, so any difference your voltmeter reads is probably because it is multiplying the peak voltage by 0.707 when you have it set to read a-c.

Why does it multiply by .707?

If the source is sinusoidal then this peak rectified voltage will be about \sqrt{2} times higher than RMS AC

It seems like you guys do not agree???

Originally Posted by Cold Fusion

Why does it multiply by .707?

Because 1 / sqrt(2) = 0.707106781.

Unless you have a true rms voltmeter, which you probably don't, it measures the peak voltage, but is calibrated to read out in rms voltage. The root mean square of a sinusoidal voltage is the peak voltage divided by the square root of 2. So the a-c voltmeter is calibrated to display a number that is about .707 of the peak. Reasonably accurate for the measurement of sine waves but can be off if there is any distortion.

It seems like you guys do not agree???

Not so. We do agree.

Originally Posted by Cold Fusion

And capacitors (as far as I know) cannot explode

Wanna bet?

One of my friends almost lost an eye when a capacitor shot boiling electrolyte into his face.
And it was a much smaller capacitor than the monsters you are talking about.

Take cover,
Leszek.

Read Harold14370 carefully. Both answers are correct. A meter set to read RMS AC on its scale will report an RMS value that is lower than its peak AC value, assuming a sinusoidal waveform. Equally, the peak AC value will be times higher than the RMS value. I suggest you consult any electronics book at your library Cold Fusion. The concept is fairly basic.

So it measures the lower value only to save money on the necessary processing power to average the 1/square root 2 and sqr 2?

Are rectifier diodes rated at their voltage for AC or DC?

Is this enough to rectify 10,000 volts 60hz AC? http://cgi.ebay.com/15000V-45AMP-Hig...QQcmdZViewItem

Don't worry, I always wear impact resistant glasses while dealing with these things. (of course there is still my face to worry about, which is why I come here)

Cold Fusion - diodes just avalanche when you exceed their reverse blocking voltage. They may survive such an insult and if the VI is tempoary they won't melt.

DC values are most often used - e.g. a 1N4007 diode will withstand 1000 V and this rating will be exceeded if you apply to it with a storage capacitor placed after. The peak voltage seen by the diode will be a vector sum

Wait.....aren't diodes used most often to rectify AC? If they are, then they should be rated for AC to begin with.

Originally Posted by Cold Fusion

Wait.....aren't diodes used most often to rectify AC? If they are, then they should be rated for AC to begin with.

Anybody designing a rectifier circuit will need to understand what peak reverse voltage will be applied to the diode at a given a-c voltage.

Hi Cold Fusion. The capacitance needed depends on how much DC "ripple" voltage you can tolerate.

If we assume 60 Hz and a full wave rectifier then the current pulses will be at 120 Hz - in other words about 10 ms between each output.

Then



So if amps, seconds, then using a 1 uF capacitor will result in an approximate ripple voltage of



This would be small compared to 15 kV. It seems that you could make a smaller capacitor within physical constraints.

I always repost on the danger in HV. Even a small capacitor charged at 15 kV can kill if you get your fingers across it. Please be careful and have an assistant nearby if things go wrong.

Shouldn't the peak reverse voltage be the same as the peak forward voltage? Sorry, I never learned about this in school, only the basics; no AC, semiconductor, diode, or capacitor instruction.

Originally Posted by Cold Fusion

Shouldn't the peak reverse voltage be the same as the peak forward voltage?

If voltage is applied to a diode in the "forward" direction, the diode behaves almost like a piece of wire - current flows almost freely, and there is very little voltage across the diode. Typically, between 0.7 and 1.2V, which is less than nothing compared to your 15kV.

If, against all odds and common sense, you want to create a forward voltage of 15kV on a diode by getting some overwhelmingly big power supply, your diode (and some of your tabletop) will become a cloud of high-energy plasma.

I could explain how a rectifier works, but really, this has been explained hundreds of times in sources which are readily available. Please spare me the trouble and do some homework. For starters, you could look here.

When you have at least tried to read it, come back with more specific questions.

I misunderstood the terminology; I understand how a rectifier works, but do not understand some of the terminology and mechanics behind it. Part of the problem is that I do not know what I need to know to make some of my projects work. I recently attempted to build an amplifier, but destroyed a few of the components due to a lack of knowledge behind a few things. I researched everything immensely beforehand and even signed up to an audio forum to make sure I had everything correct, but even that was not enough. This is why my questions may seem somewhat redundant and non-specific-I am trying to make 100% sure that I have everything right. My amplifier IC merely blew itself in two, but I'm afraid that what I am dealing with could do much worse.