Synchronize a dc pulse through an array of parallel wires

[trini]

Hi everyone,

my question is related to producing a magnetic field from an array of straight parallel wires. In my scenario, imagine the wires as being stacked into a regular array similar to the way a pile of lumber is usually stacked, so that the ends are exposed, eg:

====================
====================
====================
(side)

o o o
o o o
o o o
(front)

I want to send a short pulse (approx 500ns width) through the wires to create a very short duration magnetic field. What would be the best way of ensuring these pulses are in sync? Would I be wrong to simply have a solid conducting base plate attached to each end of the array? eg:

===================|
===================|-------------- (source)
===================|
(single flat plate which I solder all the wires onto)

or should I connect each wire to its own independent source (such as a capacitor) and then try to synchronize the timing with which the sources trigger?

I am a novice when it comes to electrical systems and would appreciate any feedback.

 

[berkeman]

Hi everyone,

my question is related to producing a magnetic field from an array of straight parallel wires. In my scenario, imagine the wires as being stacked into a regular array similar to the way a pile of lumber is usually stacked, so that the ends are exposed, eg:

====================
====================
====================
(side)

o o o
o o o
o o o
(front)

I want to send a short pulse (approx 500ns width) through the wires to create a very short duration magnetic field. What would be the best way of ensuring these pulses are in sync? Would I be wrong to simply have a solid conducting base plate attached to each end of the array? eg:

===================|
===================|-------------- (source)
===================|
(single flat plate which I solder all the wires onto)

or should I connect each wire to its own independent source (such as a capacitor) and then try to synchronize the timing with which the sources trigger?

I am a novice when it comes to electrical systems and would appreciate any feedback.

The return path for the pulses must be taken into account. What volume do you want to create this B-field in?

Do you know what amplitude of current you want to use for the pulse? The characteristic impedance Zo of your transmission line arrangement will then tell you what the pulse voltage must be. If it's not over 5V, you can use traditional logic and line drivers to send synchronized pulses. If you need more like a 24V pulse, you can still drive transistors with the logic gates to generate the multiple synchronized pulses...

 

[zoki85]

What would be the best way of ensuring these pulses are in sync? Would I be wrong to simply have a solid conducting base plate attached to each end of the array? eg:

===================|
===================|-------------- (source)
===================|
(single flat plate which I solder all the wires onto)

Sounds ok

 

[trini]

Hi guys, thank you for the input.

Sorry I didn't post more details, I have since modeled and simulated the device, and have attached some images to aid in visualization.

In the first image I show a cut away of the geometry. In the center is a region to be magnetized. The four columns surrounding it are the linear coils I described. An arrow is shown on one of the coils to indicate the direction of current flow. The 'L' shape where the coils meet are 90deg copper busbar, the faces of which are the base plates mentioned in my first post. On the bottom and top(hidden) are two blocks of soft iron to reduce the resistance of the magnetic circuit. The second diagram shows the field created by the coils.

It is important to remember that in my arrangement the wires of the coil are in parallel as opposed to a regular coil, where they are essentially in series. The 'wires' of the coil are really square copper rods.

I held the total dimensions of the coils constant and performed a parameter sweep playing with wire size and found something interesting (Note, R and L are the total resistance and inductance of the assembly, I_w is the current flowing through each wire, I_t is the total source current required by the source. In all cases, current was varied to produce an H-field of 50,000 A/m (± 100) within the magnetizing region) :

1/4" width wire (44 wires per coil)
R = 0.42 mΩ
L = 1.79 μH
I_w = 762 A
I_t = 762 x 44 = 33528 A1/2" width wire (12 wires per coil)
R = 0.1 mΩ
L = 1.62 μH
I_w = 822 A
I_t = 822 x 12 = 9864 A1" width wire (3 wires per coil)
R = 0.03 mΩ
L = 1.75 μH
I_w = 775 A (Expected to be > 822, we crossed an optimization turning point)
I_t = 775 x 3 = 2325 A

solid copper block (1 wire per coil)
R = 0.009 mΩ
L = 1.83 μH
I_w = 749 A
I_t = 749 x 1 = 749 A

Quite surprisingly (to me anyway, I'm a newbie), the best way to do this task seems to be by using a solid copper piece for the coil.

For my particular application, the most important thing is a short pulse time (<50 μs). It is also bidirectional so I have to use an H bridge. I have found supercapacitors that can easily deliver my current and voltage needs, what kind of circuit can I use to control the pulse time of this capacitor through an H bridge? Does the H bridge control the timing?

 

[zoki85]

Hmm. Pulse times in order of few μs are not problem for fast switching silicon devices of today. But peak currents in order of 10³-10⁴ A might be.

 

[berkeman]

@trini -- Why do you need to use such a short pulse? Are you sure that the alnico material will even acquire a magnetization in such a short amount of time? (or I may be mixing up two different threads in my mind...)

 

[trini]

@zoki85 I have reduced the dimensions of the device and magnetizing region such that I only need a 380 A current, is it now possible?

@berkeman I could theoretically go over that (lets say max 500μs). I can't give too many details as to why but for the purpose of my application I would like to minimize the time as much as possible.

 

[Baluncore]

I do not think you have a synchronisation problem. The current pulse will travel at close to the speed of light so calculate the phase delay on the basis that the pulse will travel one foot per nanosecond.

The problem you do have will be the time it takes the magnetic field to penetrate the material being magnetised by the pulse. Use the formula for skin effect to work out how long your pulse must last.

Another problem you may encounter is physical deformation or a ballistic trajectory resulting from the sudden magnetic pulse. The more conductive the material, the more dangerous the forces and the longer it will take for the magnetic field to penetrate the material because of the induced eddy current field.

To produce a fast pulse of high current I use a Marx generator, not set up as the traditional voltage generator ladder, but as a current loop. Consider a loop of alternate capacitors and spark gaps. Two parallel DC supply rails are used with resistors, (say 10k), to charge all capacitors with the same polarity and high voltage. When it is ready, trigger one spark gap with a voltage spike or a UV flash, all the gaps that can see each other's UV will sympathetically break down and the capacitors will all discharge round the closed loop. The capacitors will need to be able to deliver high current, coaxial aluminium tubes work well. By contorting your loop circuit into your grid of conductors you will have your high current fast pulse.

Beware the physical forces that result from a fast magnetic pulse. You have built a magnetic cannon. Keep well clear during testing, use sandbag walls and make sure it is all bolted down tight.

It would be safest to do the skin effect calculation, then produce a slower pulse that penetrates the material to depth, safely.

 

[zoki85]

@zoki85 I have reduced the dimensions of the device and magnetizing region such that I only need a 380 A current, is it now possible?

Yes. Low voltage igbt chips modules with switching times around 1 μs and rated average current several hundred amps are commercially available. Besides, they can be abused with significantly higher peak currents in intermittent modes if you know how to do it, but I wouldn't recommend that to begginers.

 

[trini]

Hey guys, I'm very sorry but I made a terrible mistake during my simulation and had the turns per coil hardcoded to 10 instead of N (turns per layer * layers) which explains why the inductance was so similar for all of the arrangements. I halved the dimensions of the coil to reduce the requirements, here are the corrected stationary values for the wires(dimensions also halved):

1/8" width wire (48 wires per coil)
R = 4 mΩ
L = 20 μH
I_w = 82 A
I_t = 82 x 48 = 3936 A1/4" width wire (12 wires per coil)
R = 0.25 mΩ
L = 1.2 μH
I_w = 324 A
I_t = 324x 12 = 3888 A1/2" width wire (3 wires per coil)
R = 0.016 mΩ
L = 0.09 μH
I_w = 1234 A
I_t = 1234 x 3 = 3702 A

solid copper block (1 wire per coil)
R = 0.002 mΩ
L = 0.01 μH
I_w = 3556 A
I_t = 3556 x 1 = 3556 A

We can see that there is now only a marginal improvement to total current requirement by using a single block, and the requirement is back into the kiloamp range.

After the stationary test, I changed the simulation to a transient one to model the skin effect and eddy currents. The model assumes a uniform current being applied for all times (which is a valid assumption as long as the rise time of the coil is on the order of nanoseconds). I have attached a gif showing the induced eddy currents as streamlines, as well as the time taken for the field to penetrate the material with the parameters [solid copper block, t_sim = 600 μs, I = 390 A], which only produced an average field of around 7000 A/m within the region at saturation.

It takes roughly 300 μs for the material to saturate. I am still running the simulations for the smaller wires.

@Baluncore that circuit seems promising, I will spend some time trying to learn about it, do you have any resources for me to look at with respect to the current loop version?

@zoki85 I made a mistake sorry :(

 

[Baluncore]

The model assumes a uniform current being applied for all times (which is a valid assumption as long as the rise time of the coil is on the order of nanoseconds).

That is not a valid assumption. The electric field along the surface is close to instant, but the current does not rise fast because eddy currents in the material oppose the changes.

Solid copper conductors are not optimum with current pulses. The current flow and magnetic field travel at close to the speed of light along the surface, but they only diffuses into the copper at a speed of about 2 m/sec, (yes, that is walking speed). A similar thing happens with other material in the magnetic field. You will need to use many parallel thin wires, called “Litz”, or insulated metal shims, (like the laminations of a transformer, only thinner). The electric field travels along the conductive surface in the surface insulation and so can quickly reach the volume of conductor material needed to conduct the current.

What is the material in your generated magnetic field?
Is it a film or a solid block material? How thick is it?

 

[trini]

@Baluncore Yes you are right, the current rise is not instantaneous. I will try to make the simulation voltage driven rather than current driven to model the effects.

Is this conductor saturation time the same as the L/R time constant?

I have a spool of AWG#32 that I can use. Is it easier to get control equipment for high voltage low current switching?

The material is a .5" x .5" x 1.5" block of Alnico 5.