Building a capacitor bank capable of pulsing 16000 A DC

[trini]

thanks Bob, i know I've been asking you for a lot of readouts, but i ran some numbers and now think that 3 should be enough to get where i need to get to, but would like a readout just to be certain.

total:

C = 0.036 F
R = 0.00533 ohm
V = 400 V

I promise this is the last time i'll ask as I already know that 10 would work and now am just looking for the minimum i would require,hope i have not been too much of an inconvenience, thanks again for all the help.

 

[Bob S]

trini-
OK. 3 caps in parallel. 16,000 amps peak.
Bob S

 

[gary350]

What are you trying to do, what is the objective?

I have several capacitor banks. I charge them up by pluging them into the wall 120 volts AC. My largest capacitor bank has current limiting so it will not pull more than 15 amp and trip the circuit braker while it is charging. This capacitor bank will charge to 54,000. Amps at 5000 Volts DC = 270,000,000. watts. When it discharges it sounds like a high power rifle going off it makes a very loud explosion. It will vaporize 6 ft long by 1/8" diameter solid steel wire. A friend gave me a used 20KW flash lamp from the airport we flashed it one time. It vaporized the whole flash lamp. I found a few pieces of melted glass about the size of a pin head.

I have another capacitor bank a little smaller. I discharge it threw 3 turns of #10 copper wire wound into a coil around an aluminum can. The wire vaporizes and the magnetic field crushes the can into a ball. It will also crush a quarter down to the size of a nickel.

I have some smaller capacitor banks too. My smallest cap bank is 250 VDC, 2 good size power supply filter caps in parallel. I use this to super charge magnets. A 1/2" diameter x 1/2" long magnet will lift 1 lbs. I discharge the cap bank into 1000 turns of #24 wire the magnet will now the magnet will lift over 300 lbs for 2 seconds. The super strong magnet field has a half life of about 2 seconds so after 10 seconds it will lift only 10 lbs. After another 10 seconds it will lift 2 lbs for about the next 24 hours. After a week the magnet strength of the magnet is back to normal about 1 lb.

 

[trini]

gary, the field will be used to saturate a rare Earth magnetic powder, i need a magnetizing field of 1.6 MAm to do so.

Bob, i found out these capacitors are indeed unipolar, what would be the effects of this on my circuit, also, i found some unipolar protection devices which work up to 45 kA, would the use of one of these mean that i can go ahead and use the unipolar capacitor?

EDIT:

I'm putting this question here as it relates to the same project, as you guys were mentioning, the skin effect will come into play, so my best bet is to find and bend tubing into my coil. In the event that i cannot bend this tubing accurately enough though, i have sourced #70 copper wire, which is made up of 18 strands of copper wire each about 3 mm thick, and the total diameter of this wire is about 1.2 cm. Would a cable made up of smaller wires(note they are not twisted around inside, just held together in a bundle by the insulation) produce the same magnetic field as a single wire of the same size, and would it also have the same tolerances to current?

 

[Bob S]

Yes, you can use reverse protection diodes or your unipolar protwection device to shunt any reverse current through the capacitors. Your magnetization current would then be the first half cycle.

You can buy Litz wire from several mfgrs. It would be nice to find a stock cable that is on the shelf. Google " Litz wire". The Litz wire is geneally listed as for example AWG 08-32 which as I recall means 8 Ga copper wire made up using 32 Ga Formvar insulated solid copper wire. The DC charactistics of Litz wire is the same as the quoted Gauge solid copper. Magic numbers for the numbers of conductors is 7, 19, etc.
3 mm dia wire is probably 9 Ga. Is it insulated? Re 70 Ga wire: Do you mean 7 "ought" wire (meaning 7 zeros Ga) wire instead of 70 Ga wire?

 

[trini]

alright guys a few weeks of gathering supplies and working out kinks and I'm nearly ready to try this, the last thing left to do is wrap my wire around my frame. Bob I looked into the litz wire, and it seems to me to be extremely similar to amp wire, although less twisted. do you think that i could use amp wire, which is more available to me, even though the strands themselves are twisted around more inside? more specifically, do you think that the twists in the wire will significantly affect the final output field?

 

[Bob S]

trini-
In Litz wire, all the indvidual strands are insulated from one-another. Amp-wire is certainly better than solid wire, and bends better, but not as good as Litz wire in terms of eddy current limitations. A problem I have had with large bundles of Litz wire in high-frequency ferrite magnet coils is making good solder contact will all zillions of strands. Use Amp-wire. Keep us posted on your progress.
Bob S

 

[trini]

Alright I'll go ahead and use the amp wire. I've been looking for good diodes to use with this system, but can't seem to find any. Does anyone have a suggestion (note: operating voltage = 400V, Imax= 16,000 A). Also, would it be practical to create some sort of wheatstone bridge to rectify the system, perhaps using lower rated diodes?

 

[Bob S]

trini-
Please see my thumbnail in post #7 under the thread "Pulse Width of Magnetic Field":

https://www.physicsforums.com/showthread.php?t=329842

When you use a series diode to get a single half cycle current pulse, the capacitor gets charged to a high voltage with reversed polarity. unless the Q of your circuit is very low. So unipolar caps should be used with caution in a resonant circuit.
Bob S

 

[trini]

Bob,

is there any sort of fuse I can use which will allow the first half wave to pass, then blow on the reverse and physically discharge the coil, thus meaning the only pulse i see is my first pulse, because I want the direction of the resultant field in the magnet to be as strong as possible in one direction. Once the field changes direction, it's going to flip over some of the domains in my magnet, which is counter productive. Basically, what simple, reliable solution is there if I cannot source the right diodes.

 

[Bob S]

trini-
I have looked through high current diodes, and have found ones with adequate peak 1-cycle surge current and adequate peak reverse voltage, but they are very expensive (~$600.00). I do not advise paralleling diodes in this situation, because one will invariably conduct most of the current. I Don't have any other suggestions right now.
Bob S

 

[trini]

Hmm well the diode is necessary in any event. What is the difference between a thyristor and a diode in this application? would a suitable thyristor be more economical than a generic diode here?

 

[Bob S]

Your suggestion of a triac (an ac SCR) led me to a search of SCRs. An SCR is actually better than a triac in your case, because they conduct in only one direction, and can hold off lots of volts in the other. They will not conduct after the capacitor voltage changes sign. Here is one:
http://www.pwrx.com/pwrx/docs/t9g0--10.pdf
It is about the size of an ice hockey puck, and costs ~ $160. It can handle a single 16,000 A pulse, and hold off up to 2000 V. You can use it for your switch. You will need to ac couple to gate.

 

[mheslep]

trini-
I have looked through high current diodes, and have found ones with adequate peak 1-cycle surge current and adequate peak reverse voltage, but they are very expensive (~$600.00). I do not advise paralleling diodes in this situation, because one will invariably conduct most of the current. I Don't have any other suggestions right now.
Bob S

Power MOSFETs could be used and controlled w/ current sensing feedback so that they're forced to share the load equally. This is essentially what's done in high power switched mode power supplies.

 

[vk6kro]

The ringing effects may not occur in this case.

They are caused by the coil returning power to the capacitor, but in this case, the power is being removed from the coil to magnetize some rare Earth material and to supply eddy current losses.
Also, the resonant frequency will be pulled down by the magnetic properties of the rare Earth mixture being inside the coil.

So, the damping effect may be much more severe than predicted by BobS's wonderful graphs.

 

[trini]

Guys, I was looking over my calculations this morning, and think i may have misinterpreted the relative permeability factor in my original equations. My original eq'ns considered the permeability and not the relative permeability. The following shows my revised calculation of the required current:

∫H.dl = κ [(N/ι)(I) + dφ/dt] {Re: dφ/dt = -ε0 LI}{κ = μ/μ0}
= κ [(N/ι)(I) - ε0 LI] {Re: L = (N2/ ι) A }
= κ I [(N/ι) - ε0 (N^2/ ι) A ]
Where,
∫H.dl = magnetizing field
K = average relative permeability
N = number of turns in solenoid
ι = length of solenoid
I = current
A = average cross sectional area of surface being penetrated by solenoid

Given:
Bsat = 0.925 T
M = Md(6000)/4∏ = 0.05307 T
=> Hreq = (Bsat / μ0) - M
= (0.925 / 1.25663706 x 10-6 ) – 0.05307
= 736,092 Am^-1
Also;
N = 6
ι = 0.09 m
A = 0.036483 m^2


Now my magnet is to be aligned diametrically, so the cross section of my solenoid will appear as in the attached file:

The pink area denotes the cross section of a bag which will be filled with iron powder, the relative permittivity of which is 700.

The stainless steel tube is non magnetizable, so I have ignored it for this calculation.

The blue area denotes the B powder, the relative permittivity of which is 939,014.

The ratio of the areas of pink : blue = 4.29 : 1
=> average permeability of the magnetic path, K = [(4.29)(700) + 939,014] / 5.29 = 178,075


So,
∫H.dl = κ I [(N/ι) - ε0 (N^2/ ι) A ]
736,092 = I(178,075){(6/0.09) – [ε0 (6^2/0.09)(0.036483)]}
= 11,871,666 I
I = 0.062004 A = 62 mA

this seems very small to me, though the physics does check out. the only question i have regarding this is my interpretation of the relative permeability, as i am not sure if i should just use 700 which is the permeability of just the steel(in which case my required current is about 12 A, which is more understandable). Also with these new current values, i can probably just use a car battery to apply a steady DC current while i heat the powder to activate its thermosetting resin.

 

[Bob S]

Trini-
Here is the central B field for an air-core solenoid of length L and radius r from Smythe "Static and Dynamic Electricity" 3rd Ed., page 297 eq(4). Note that the radial size reduces the central field for a fixed length L

B_z = u₀NI/sqrt(4r² + L²)

[Edit: To include the iron powder, multiply Smythe's formula by the effective relative permeability of the iron powder. It is something like 700, NOT 178, 075. Very few things have permeabilities over 10,000.]

The solenoid inductance calculator used by NASA is available for download at:
http://www.openchannelsoftware.com/projects/Solenoid_Inductance_Calculator/

The analytic equation for a single layer solenoid (thin solenoidal current sheet) is derived by Smythe "Static and Dynamic Electricity" 3rd Ed., page 340

L = pi u₀ a²n²[(z²+ a²)^1/2 - a]

where z= length, a = radius, and n = # of turns.

There are several on-line calculators. One used by ham radio operators is:

http://hamwaves.com/antennas/inductance.html

 

[trini]

Unfortunately I can't use these calculators as my coil is cuboidal in shape rather than cylindrical(to allow for the shape and orientation of my magnet), however my calculations are based off of first principle, so I do not doubt them.

The question here lies in my interpretation of the relative permeability to be employed. I suppose i could always allow for a lower permeability and just use 700, because using a higher current can only serve to create a stronger more uniform field in my rare Earth powder.

 

[Mike_In_Plano]

Few thoughts

1. You'll have to drive the Neo into saturation, so it's perm = 1
2. When considering rectifiers for pulse applications, try evaluating them by I^2t. They'll have a rating for peak current, usually for a 8.3 ms half sine wave. Integrate over this to get the I^2t (fusing rating). Then calculate the integral of your pulse's I^2t to see if the parts will survive.
i.e. Average I=100amp, peak I =2500amp. i(t)=2500 sin(120 pi t) (0 - 8.3ms)
. Integration of i(t)^2 x t (0 - 8.3ms) gives 108.5 a^2 s

3. Multiple rectifiers are fine - use a length of lead in series with each to form a ballast resistor. If the lead drops 2v at peak, the rectifiers will track fairly close.
4. In place of Litze wire, try parallel windings side by side (filer). Or you can look into purchasing some flattened wire.
5. Be very careful - In the kiloamps, Lorentz forces can cause wires to explode outwards.

Mike

 

[trini]

mike, the MSDS listed the permeability at my density(7.6 g cm^-3) as 1.18. if i am not mistaken, as the iron powder is essentially a magnetic path, i am concerned with the relative permeability, correct?

all my previous calculations assumed an air core, but this is now a steel core. because the powder will be within a field 700 times stronger than an equivalent air field. the magnet assumes saturation when it is placed in a field which relative to itself is above a certain value, Bsat(in this case= 0.925 T= 736,090 A/m). what are your thoughts on my deduction?

 

[Bob S]

Hi trini-
The iron powder dominates the field properties inside the coil, and B_longitudinal is continuous, so the iron powder will determine the B field in the sample. Do you have a curve like my annealed-iron thumbnail for the powder?
Bob S

 

[trini]

i have a similar graph but not one including the relative permeability, though i can say it is effectively 939,014 at my density. I think it is safe to assume that a 15A current will more than drive the powder material to this permeability in this core. My scanner is down so i can't upload the graph, but is it agreed that this is the best way to go for my project?

EDIT:

totally missed the point of your question, no i don't have one of those graphs for the iron powder i'll be using, i still have to locate it.

 

[Bob S]

i have a similar graph but not one including the relative permeability, though i can say it is effectively 939,014 at my density. I think it is safe to assume that a 15A current will more than drive the powder material to this permeability in this core. My scanner is down so i can't upload the graph, but is it agreed that this is the best way to go for my project?

trini-
Nothing has a permeability that high. Mu-metal is like 50,000. I suspect that you should use a number more like 700. Your formula for H should look something like my formula from Smythe, and you should specifically be correcting for the width of the coil if it is more than say 25% of the length of the coil.
Bob S

[Edit] For rectangular coil of width w, this should be pretty good:

H_z = NI/sqrt(w² + L²)

B_z = u u₀NI/sqrt(w² + L²)

where w = width of coil, and L is length.

 

[trini]

bob,
I was referring to my rare Earth powder, I may be wrong but since my neo powder has a u = 1.18, then my relative permeability = 1.18/u0 = 939,014. Regardless, even using 700 as my relative permeability, I only need 12 A, so that's why I think a 15 A DC source running for my heating time(1 minute to heat powder to 195 C using IR heating) will overcome the eddy current losses generated. Also, the width of my coil is negligible compared to its length(its a long magnet, 50 cm x 1 cm thick).

http://www.lessemf.com/278.html

that site shows a similarly composed composition with a relative perm of 1,000,000.

 

[Mike_In_Plano]

I worked on a project where GE was building brushless motors for us, and they had a terrible time with an experimental Neo design because the magnetizer was so hard to build. As I recall, they were playing around with numbers like 20kG or so. At that point, a lot of magnetic materials start saturating.

So, you get a goodly many webers through the core, until it saturates. Beyond that, your perm=1, and and additional magnetization is due to pushing a lot of current through the coils.

In the end, we settled for large ceramic magnets - much cheaper.

 

[Mike_In_Plano]

Hmmm, I don't think it's fair to expect any magnetic powder to approach the permeability of a strip of metglas. It (metglass) is incredibly homogeneous. Most varieties don't even have anything you can define as magnetic domains (though some have oriented "micro crystals" which are introduced to optimize the B-H loop).

Powder, on the other hand, is the accumulation of a multitude of diverse particles. Within any given particle,you can make a valid claim that it has a tremendous perm, but the irregular packing between the particles will make the whole appear far less permeable.

 

[vk6kro]

I'd like to see the powder above and below the coil as well as down the middle as in your diagram.
The complete path includes the area outside the coil and the less air gap there the better.

Also, it should taper towards the magnet rather than tend to bypass it. You really want to concentrate the field in the area of the magnet.

 

[trini]

I updated my design module, making this a more typical magnetic circuit:

 

[trini]

Ok so I've been meddling around with my magnetic circuit design, and right now my biggest problem is having enough ampere turns in the space. in my current design(see above attachment), i can fit 6 turns of gauge 0 amp wires per side.this still requires a huge current to fully utilise the size of wire. instead, i have devised a plan to maximise space and minimise required current.

I will construct a series of magnetic flux 'cells'(shown below) which can then be any required length to allow for any desirable amount of turns. the cells will be made out of annealed iron rods, and will 'impale' themselves into the powder core, which I can make into any shape I desire. Each cell will consist of 4 parallel magnetic flux sources, so the total flux into the powder core would be 4 times each of the sources' flux. I will distribute these cells evenly throughout the length of the powder core to ensure uniform field distribution.

I would be foolish to assume there would be uniform permeability throughout the powder core, so my main goal should just be to drive the core to saturation( about 1.2 T). My magnetic material has an Msat of 0.925 T, so I think all I have to do is create a field of that or higher in my air gap to saturate my material, unless Msat means something else entirely.

 

[Bob S]

Hi trini-
I like your coil configuration in post #58 better than #59, because there will be less flux leakage, which will literally defeat some good dipole magnet designs, which are made with higher relative permeabilities.
Your #58 design is called an H-magnet design, and the powdered iron funnels the flux around a bend into the sample, where the H-field (but not necessarily the B field) is concentrated. Do you plan to slide the sample in/out or do you plan on removing ferrite to get the sample in/out?
In any case, the powdered iron/ferrite will saturate before you get to 1.4 Tesla, and there are few alternatives to adding brute-force amp turns.

The best possible design might be a cylindrical version of an H magnet (similar to a gapped pot core), with a gap in the center for the sample, and the sample surrounded by the coil. This would get the coil close to the sample, but it would have to be disassembled (like pot cores) to get the sample out.
You can minimize the size (max the Ga. #) of the wire in your coil by picking a wire size, and then integrating the current-squared times wire resistance over time:
E = ∫I²R dt (energy dissipated in wire during single pulse)
and choose a wire gauge such that the temperature rise in one pulse raises the copper temperature say 25 to 50 degrees.
Remember that the powdered iron raises both the magnetic field intensity and the inductance of the coil.
Bob S.