How to generate strong variable-freq. AC magnetic field?

[OldN00b]

I need to generate rather strong EM AC fields over quite a frequency range. What I have in mind is a coil at most the size of an induction heater, but at a much greater field strength. There are various oscillator designs for such a L-C oscillating setup, but none of them are variable-frequency. The difficulty lies in due to the impracticality of having such a large C variable over such a long range.

My coil would ideally just be a single winding with a diameter of 1 to 3 cm (about 1 inch max.). This to be able to maximise the current and hence the field strength.

The goal is to have a high as possible field strength (2 Tesla or higher would be ideal) and to have an as wide as possible frequency range, starting at 200 kHz and at least to 2 MHz but up to 20 MHz would be good too.

I'm looking forward to your suggestions. Please suggest any outlandish idea, as long as it works, because money is no object, this device may cost 50,000 USD in parts.

I am fully aware of the craziness of building such a device, I know full well that the coil for ex. would need water cooling and that a high voltage, high power power source will be required. And that shielding will be required to stay within the RF emission laws. I'm willing to construct my own variable capacitor. Full workshop available. Thank you for your suggestions.

[NiHaoMike]

2T is within the range of what a MRI scanner works at, but static rather than oscillating. MRI scanners do also have oscillating electromagnets to do the scan, maybe one of those drivers modified can achieve a bigger field strength in a smaller volume?

[BrianHG]

2T at 200KHz-2MHz 
This sounds like a super high powered radio transmitter...
Without shielding, you'll kill any AM radio broadcasts in the neighborhood.

[jbb]

Holy #### thats a high T.Hz product.  This thing will be outright dangerous so don't skimp on the shielding & safety.  You can expect significant heating effects for anything conductive or ferromagnetic.  The field gradient (i.e. dT / dt) will probably be directly dangerous to your health too (for example, Peripheral Nerve Stimulation).

I suggest you get someone versed in the art of magnetics to help you model your output coil with 3D FEM.  Effects such as the proximity and skin effect could make this nearly impossible.  High L * dI/dt rates could lead to sufficient electric fields that the coil experiences corona discharge.

I'm guessing here, but I suspect that some kind of resonant LC circuit will be required (a la induction heater) to get enough alternating current.  The trick will getting a widely-variable capacitance value.  For a resonant stage, the basic equation is:
f_r = 1 / (2 pi sqrt(L . C) )

To get a frequency tuning range of 10:1 with a single value inductor, you would need a C range of 100:1, which is quite extreme but probably possible.  Getting a frequency range of 100:1 would require a C range of 10000:1, which is almost certainly impractical.

Some questions which might help you think about the project:

• How do you feel about interchangeable heads?
• Or building a couple of different machines to handle the results?
• Additionally, what kind of tuning do you need to do?
• Do you need continuously variable frequency?
• Or are steps OK? What size steps?
• Do you need to change the frequency while the field is on, or can you turn it off?
• Is it OK if the frequency undershoots/overshoots for a while before settling to the target value?

Something like a Thyristor Switched Capacitor (TSC) and an electronically switched capacitor bank could be adapted to provide the variable capacitance.  (The thyristors will need to be updated to some more modern switch.  Zero Current Switching (ZCS) could be used to help with switching losses.)

[Kleinstein]

I know of one setup, used to make a 0.1 T field at 55 MHz. So somewhat in the direction you are looking for. They used an low loss LC resonator and tube based amplifier to drive it. Already this was quite a crazy thing. One had to be careful with the choice of materials: everything needed to be low dielectric loss (e.g. PTFE) - normal plastics would melt from dielectric loss due to the electric field that comes with the magnetic one.

One might have a chance short time - thus just a single pulse of a high magnetic field. This also contains higher frequency components. Though usually this is more like fractions of a ms and not going to such high frequencies. Even with just a single turn the voltage is too high. Maybe the explosive compression could be that fast - but nothing for an easy experiment and with a lot of paperwork / permits.

[jbb]

Mmmm, high power vacuum tubes. Still the best when your specs go to crazy-town.

50 MHz 0.1 T makes a product of 5 T MHz.
2 MHz 2 T makes a product of 4 T MHz.

So, probably possible, but very difficult. You can look forward to machined PTFE and other such joys. Maybe getting the heating coil plated with hard gold (so it's smooth and stable to manage skin effect) will be required as well as the near-certainty of water cooling.

For the power amplifier, modern silicon carbide
(SiC) or Gallium Nitride (GaN) devices might well suffice rather than exotic vacuum tube kit.

This will need extensive modelling...

[Fungus]

Another wireless phone charger? 

[T3sl4co1l]

PM me if you'd like some consultation on this.

Tim

[Wolfram]

For some rough approximations, to get a 2 Tesla RMS AC field at 2 MHz in a 30 mm single turn air cored coil:

If the coil is wound with 5 mm copper pipe, the current needed to get a 2 tesla field is approximately 8000 amperes. The inductance of the coil will be around 36 nH. This gives the coil an inductive reactance of 0.45 ohms at 2 MHz, so you need about 3.6 kV at 2 MHz to get 8 kA in the coil. Air coils like this typically have a Q factor of a few hundred, so the coil losses will be at least 100 kW, likely more. A suitable capacitor bank for this would be something like 12 parallel Celem CRPI300 100 nF capacitors. The sheer physical size of such a capacitor bank would inevitably add more parasitic inductance, which means that you need more capacitors to compensate it, quickly leading to diminishing returns.

I have no idea if it is even realistic to have 100s kW of dissipation in such a small coil, for sure it needs to be cooled from the outside with water at a high flow rate, but even then it might not be practically doable.

If you reduce your field strength requirement and go for a single frequency, it might be possible to do it within your budget. You could probably buy an induction heater commercially that could do half a tesla at 400 kHz with the budget you have.

[MagicSmoker]

...
My coil would ideally just be a single winding with a diameter of 1 to 3 cm (about 1 inch max.). This to be able to maximise the current and hence the field strength.

The goal is to have a high as possible field strength (2 Tesla or higher would be ideal) and to have an as wide as possible frequency range, starting at 200 kHz and at least to 2 MHz but up to 20 MHz would be good too.

I'm looking forward to your suggestions. Please suggest any outlandish idea, as long as it works, because money is no object, this device may cost 50,000 USD in parts.
...

A 1-3cm diameter coil is going to be too small physically just to handle the level of current that will be necessary here (in the realm of thousands of amp*turns), and if you could somehow wind more than one turn to reduce the current required that would just result in a massive voltage gradient between turns, leading to incessant arcing.

The only practical circuit topology is a parallel resonant tank with a superconducting coil and vacuum dielectric variable capacitors switched in and out of circuit with hydrogen thyratrons (e.g. - Excelitas). With this arrangement you can probably get to 2T at 2MHz, but I simply don't think 2T at 20MHz will be possible except maybe as a pulse-excited, naturally damped oscillation; definitely not CW operation.

[OldN00b]

Thank you all for your valuable suggestions and highly interesting information, and for taking me seriously. I appreciate this brainstorming. I attached a photo of my self-made 2KV capacitor (150 microFarad) to show that I am very seriously working on this project - that Cornell Dubilier cap cost me 5,000 USD in parts and quite a few days of CNC-drilling and soldering. Its internal resistance is 0.05 Ohm. I did not measure its self induction yet. Please note that I am a NOOB. I did not want to buy a Chinese cap bec. my cap is superior in terms of being able to handle much larger and for a much a longer time, ability to self-repair at over-voltage as well as gaining experience with such projects.

I will carefully consider all your comments and suggestions and will base my decisions on them, as well as on further discussion/research/experimentation/calculation/simulation. And I may PM those who offered consultancy.

I will address your points now. Minimum specs in more detail:

- The device MUST be variable-frequency and it must be settable to any arbitrary frequency between 0.5 and 2 MHz (20 MHz was a "nice to have" feature but let's stick to the essentials first) by an analogue dial and the device should be able to operate for at least 10 minutes at a time before cooldown is required. The freq. should be changeable steplessly while the field is on. The field strength should be between 0 and 2T, 2T being nice but 0.5 would already be acceptable. The volume of the field should be at least approx. 1 cubic centrimetre.

- IMPORTANT: It's permissible to have a short duty-cycle. For ex. the device operates for 100 ms and then is off for 900 ms. That would reduce heating by an order of magnitude.

- The budget is 50,000 USD - 100,000 USD.

- Thanks for suggesting tubes. I started to think about that already.

- Thanks for talking about various physics-related issues that are complicating design factors. I tried machining Teflon - indeed difficult. I could freeze it (even in liquid nitrogen) to make that easier. The comment about gold-plating: I actually have several KG of gold at my disposal so I could even make the coil out of solid gold, but isn't silver a better electrical as well as thermal conductor? I will perhaps make the coil out of thick litz, to help with the higher frequencies. A disadvantage is that you can't run water through litz.

- I have been thinking of constructing a variable-induction coil but I think that's less practical than a variable C.

- This is a private project with perhaps commercial potential in the far future.

- I am aware of potential adverse health effects and will strive to provide adequate shielding with mu-metal (crazy expensive stuff...)  etc.

- I will resist the urge to try how quickly I can wirelessly charge my smartphone with it. If I break down and can't resist that urge anymore, I will try it first with my wife's phone and get back to you with the results.

- I am aware of potential harmful radio interference and will work to avoid getting into trouble with the law. I have RF probes for my spectrum analyzer and successfully done FCC pre-compliance testing on one of my products so I should be able to detect such stray radiation.

- The attached cap I made was for the purpose of generating a multi-T AC field for just a few microsecs. Using a spark gap as a switch and a MOT-based PSU.

- Interchangable heads is an interesting idea but there still needs to be a very significant freq. range, so perhaps not practical.

- Different machines: Too expensive/impractical. I rather spent even 100 grand on one machine and have it weigh a tonne.

- It is NOT OK if the frequency undershoots/overshoots significantly for a while before settling to the target. The freq. should be stable to a reasonable degree but can vary due to thermal issues etc. There are no requirements as to maintaining a set freq. stably. Any significant deviations of the freq, even for one microsecond, are NOT permissible.

- The field should look like a sine wave and not have harmonics or other freq. components.

- The coil, absent heroic measures, will come apart due to the enormous forces generated inside of it, so it will have to be heavily reinforced by composite materials.

- Any fellow aspies noticing that those bolts are ugly-long: I replaced them with shorter ones :-)

[Wolfram]

A 1-3cm diameter coil is going to be too small physically just to handle the level of current that will be necessary here (in the realm of thousands of amp*turns), and if you could somehow wind more than one turn to reduce the current required that would just result in a massive voltage gradient between turns, leading to incessant arcing.

The only practical circuit topology is a parallel resonant tank with a superconducting coil and vacuum dielectric variable capacitors switched in and out of circuit with hydrogen thyratrons (e.g. - Excelitas). With this arrangement you can probably get to 2T at 2MHz, but I simply don't think 2T at 20MHz will be possible except maybe as a pulse-excited, naturally damped oscillation; definitely not CW operation.

The problem here is, for the given coil size, you need a few hundred nanofarads of capacitance to get a resonant frequency of 2 MHz. This would be very difficult to do with a vacuum capacitor, without also adding more stray inductance than the work coil itself.

Otherwise, your suggestion is probably the closest it is possible to get to the target performance, but even this is a (multi) million dollar implementation, far from the 50 000 suggested parts budget in the original post.

Also, I just noticed that I was off by an order of magnitude in my previous calculation, the losses in a copper coil would be in the megawatt range for 2 T at 2 MHz, a further nail in the coffin for any non-superconducting coil.

[The Soulman]

the losses in a copper coil would be in the megawatt range for 2 T at 2 MHz, a further nail in the coffin for any non-superconducting coil.

Someone did mention liquid nitrogen...
I'm curious, what is the application for this device? 

[OldN00b]

I prefer not to disclose that.

[The Soulman]

I prefer not to disclose that.

Fair enough, we'll file it under "personal pleasure". 

[OldN00b]

Since the feasibility of such a device seems to be low, let's concentrate on the duty cycle.

I said previously that a 10% duty-cycle would be acceptable, but in fact, a duty cycle of 1% or 0.1% or even 0.01% will also be acceptable.

Meaning, every 10 seconds a pulse of 1 ms with that variable frequency.

Would that make the device more feasible? Certainly in terms of mechanical stress and cooling, as well as the power needed for continuous operation.

[Wolfram]

That makes a huge difference. If it is pulsed at a low duty cycle, it does not neccesarily have to be that difficult. A 0.1 % duty cycle at 2 T and 2 MHz brings down the coil losses to the single digit kW level, roughly approximated, which can be managed with internal water cooling at a modest flow rate.

For the pulsing, what is the minimum acceptable pulse length? Is it acceptable if the pulse amplitude changes during the pulse? If so, it could be as simple as dumping a capacitor into the coil at regular intervals.

The very loosely defined requirements and the secrecy does not give me a good feeling about the project, but in any case it is an interesting thought exercise.

[daqq]

Interesting requirements

Maybe you could make a high voltage push pull DAC out of Behlke switches, with a low pass filter on the output?

[OldN00b]

Thanks Wolfram. The requirements, I'm constraining them to the bare minimum, based on your feedback.

I realize that I do not need more than say 100 periods of the "signal" per experimental iteration taking say 10 secs at minimum, and since it's for ex. a 1 MHz signal, 100 periods would be a mere 0.1 ms.

Dumping a capacitor is exactly what I want to do, but then you get 7 periods (due to the decay) and of course it's fixed-frequency. I'd like the amplitude to be roughly constant over those 100 periods.

Basically, what I need is to have a variable-frequency oscillator with ridiculously high power. 100 periods generated at 1.21 MHz, ten seconds later 100 periods generated at 1.22 MHz, etc.

It would be NICE to have more features, as I initially mentioned up to 20 MHz, continuous operation etc. but none of it is a hard requirement, as I am forced, by your comments, to examine more harshly what is absolutely required for an initial setup.

The problem remains: How to make this variable freq...

[OldN00b]

About the secrecy: It's a very interesting application that I wish to remain secret for the time being, but later I will vlog about it and you guys will be the first to know.

[OldN00b]

As to any type of solid-state switching: I examined that option but I ended up with the need for Thyristor switches costing 15 grand. They are very fickle (break easily) and the manufacturer said they could be used to trigger the explosive lens around a nuclear bomb, so they refused to sell it to my company, not even with the proper forms filled out (no re-export to Iran etc.) So if it comes to dumping a cap into a coil, a (triggered) spark gap seems the way to go. I'll construct those myself.

[daqq]

The problem remains: How to make this variable freq...

Note that I'm not experienced in the art of high powered oscillators, so what I say might be rubbish.

Since you have a fixed coil, then you can only change the capacitance in an LC resonator. Variable caps do exist, even for very high powers. For example see:
http://www.comet-pct.com/products/vacuum-capacitors/variable-capacitors


The naive approach would be to set your capacitance in your LC circuit, then dump your capacitor bank (different from the capacitor in the resonator) into the LC resonator and let it resonate. Sort of like a tesla coil with a spark gap. See: http://www.electron-tubes.cz/en/spark-gap for a local supplier

So if it comes to dumping a cap into a coil, a (triggered) spark gap seems the way to go. I'll construct those myself.

Well, thyratrons are still manufactured :-) http://www.electron-tubes.cz/en/51-tr-50

[OldN00b]

I think that a motorized variable cap with the specs I need does not exist and a custom-made one may cost 1 million dollars... The one here:

http://www.comet-pct.com/products/vacuum-capacitors/motorized-capacitors

...is for a very low capacitance.

And the switches are not my problem - a triggered spark gap works fine (when sound-insulated and cooled - in fact I'd design my own, since a regular spark gap does not last longer than a few hours of operation in my case).

I start to think that my best option would be to construct a motorized variable capacitor. When you build it yourself, 50 grand goes a long way.

[OldN00b]

What I could do is have many fixed-value caps that I can put in parallel, using thick copper cables and manual switches (easier than plugs) that can handle many dozens of KA and then one variable cap of the same range as one of those fixed caps. Then make some kind of Colpitts-type oscillator, some kind of circuit used in induction cooking to generate the field. But the issue then becomes semiconductors not liking to conduct 10 KA to 50 KA. And them either not existing or being very, very expensive and constantly breaking with these kind of experiments. And such an elaborate setup will have a significant self induction and resistance.

[daqq]

What kind of an inductance are you hoping to use? As in how many uH?