Our company, Prime Photonics, has a small, all-optical sensor that we have characterized at the Florida High Field Magnet facility in fields up to 30T with no saturation - the response was incredibly linear.  We developed the sensors for rail-gun applications, and have tested in-house with pulses on the order of several miliseconds.  Our probe heads are roughly 1cm cube.

We haven't sold many of these commercially, so unfortunately, they are on the more expensive side.  Let me know if you'd like to discuss further, and we could see if there is a way to get some of the sensors over to you.  

We are in the process of exploring the radiation resistance of the sensor for possible implementation in beam facilities and have promising preliminary results, but nothing concrete yet.

Not really dealing with this stuff but maybe a set of hall sensors would work?

do Hall sensors saturate?  I haven't worked with them enough to know.

They saturate. Here's a link if you'd like to read a bit.

http://sensing.honeywell.com/index.php?ci_id=47847

(forgot to paste previously)

The sensor is called- B-dot and you can make it yourself in your lab in 10 minutes.

just take a semi-rigid- coax, make a small loop at its open side with the center conductor and connect it to the shield. The response time is ~2L/R where R is the 50 ohm of the coax and L the loop inductance (make a small loop- a few mm diameter to have ~ 10 nH loop) connect the other side to a 50 ohm terminated scope input and you will measure ~d(B)/dt

David Gray: thank you for your interest. We are thinking about make some agreements  with other companies and institutions.

Eyal Kroupp: I tried some similar in the Pulsotron-2 and it was a disaster because dB/dt values changed a lot, but perhaps it is a good option with right filtering. Eyal do you have the equation that I must use?

I am thinking also to use a lohet-II sensor (+-500 gauss) and inject a low current to the coil. By knowing the current flux we could calculate real magnetic field. We can use also the Eyal coil to integrate and compare.

The loop has the problem that response is really very high in high frequency and saturates the scope or a integrator using opamps.

The lohet sensor at low current make an idea of the magnetic field at higher current, but the only way to measure real current without parasitic at high power must be the optical sensor or any other designed to withstand large fields.

1. did you impedance match the scope input to your cable impedance?

2. did you use a semi-rigid cable to avoid RF noise?

3. Remember that your signal is proportional to the loop area. rotating your loop will decrease the magnetic flux through it.

4. you can always use attenuates, but you will always have the 8-10 bits of your scope...

In Pulsotron-2 I used a coaxial 50ohm probe loaded with 50ohms, but noise was about 100 times than the signal, so the 8 bits of the scope and attenuator does not have nothing to do, it is better to use integrated signals.

Perhaps I wil have more luky in the Pulsotron-300k. I am building the coils just now.

I have a similar high frequency requirement, but not so high in Tesla.  My literature research of sensors was done a year ago, and below is my off the top of my head recall.  It might be more applicable to my experimental setup, than yours?  So, I have asked some questions (eg. sources of noise).

The frequency requirement likely eliminates all Hall "IC" sensors, which I found do not go up 1 MHz. 10KHz is common, and I found some over 100KHz.  Even 500 KHz.

A Hall sensor without an embedded IC circuit, just the planar geometry of the Hall sensor, with well shielded leads, might provide more frequency range, above 1 MHz. The planar geometry could be adjusted for less sensitivity, to measure higher field strength. Special geometry shapes, not rectangular solids, would be in order.

There are now IC Hall Sensors with anti-saturation circuitry, that prevents damage at field strengths that exceed the devices rating, to a point.  The devices are designed to put out a constant, max signal, while in saturation.  If it gets damaged, then output typically goes to zero, due to an open circuit (ie melted embedded transistor), but might fail in close circuit (rare).  It got this from reading many data sheets, across a half dozen brands, over 1 year ago.  Their frequency range is not over 1 MHz, at least 1 year ago. 

Consider calling makers' agents, and ask them.  That's a fast method to get knowledge in this area.  They are experts and can direct to you other devices to measure the magnetic field strength.

What are the sources of the noise?

Integrated digital methods may not do well in high noise. Ditto analogue methods.

Are you measuring during your fusion event?  There is likely no way to overcome the massive RFI during this event.  Any EMF antenna design (like a loop) would not just be saturated, but the surge would be damaging to embedded IC micro components (Hall IC sensor). 

Consider mechanical attenuation and then electronic.

That is why many plasma and fusion measurements use loop devices, as they handle some noise types better, and are electronically and physically more durable.

Consider many loops, each at a different angle, to reduce noise in one or more of the loops?

I like the linear optical method.  That's a mechanical means for noise filtering.

Consider measuring outside the coil, for reduced noise, and mathematically adjust the field strength.

In other words, existing commercial applications do not cover the range of measurement for some contemporary research designs involving plasma and fusion.  Do it yourself is in order?

In the Pulsotron-2 there was a very high noise due magnetic breaks and reconections., so to use anything with form of loop close to the high power discharge would generate totally random noise, if attenuator are used the amplitude of some of them decreased but others continue satutrting the oscilloscope. I tried to use a RF meter chip, also diodes connected to capacitors to try to obtain some readable output but it was impossible because rf output was well beyond the 10 gigahertz, out of the better high frequency diodes I found.

I tried also to heat a wire to measure the change of resistance to measure output power, but electromagnetic waves close to the discharge saturates also.

Also anything that is close to the ignition point sufer a sound wave end multimegawatts UV pulses. Pulsotron-2 used to destroy or damage about 80% of new sensors, targets and devices connected to it. Fortunately last test campaign things was better.

Now I want to measure discharges in a smooth manner to measure the magnetic field.

I think that the ics connected to sensor demagnetize them using a square waveform.

Javier, I'd be glad to speak with you about our optical method.  We are immune to any electric field damage.  As I mention, we've successfully characterized pulses from railgun launches in very extreme environments with extremely fast response times.  We can locate any electronic demodulation equipment remotely (~kilometers) from measurement points with virtually no degradation in signal strength.  I honestly think we might be able to work something out for you.

Feel free to message me if you'd like some literature, or if you'd like to set up a telephone call or webex-type discussion.

I reread the question paragraph, and realized you are not trying to measure a coil's magnetic field, but that of fusion discharge events.

If it's the fusion event of interest, I know of no magnetic probe that will ever have the response rate desired, let alone survive the fireball of your Pulsotron.

Why is the magnetic field near the discharge important? 

The rest of this post addresses alternatives to the magnetic field measurement.

I'm looking to measure electron and ion temp with a probe (standard stuff for decades now), and a laser beam refraction/reflection at surfaces of the double layers of the fireball (new method I read about a few months ago).  The laser beam is inviting as it's indestructible compared to the probe.  And the laser frequency response rate is above that of the desired measurement.  So, moving double layers can be 'observed' of single firing!  Once, the double layer trajectory is known, then EM fields can be proposed.

First, I might try a high speed camera at over 1 million frames a second, best would be 10 million/second.  The camera is too expensive to buy or rent for a week.  A week is too short as I estimate over 10,000 firings (3 months) are required to get enough frames to make a single sequence of 5-10 frames of the fireball expansion.  The camera must be located well away from the event, and or behind shielding, mu metal, RFI shielding, Xray and gamma ray shielding.  A long optical fiber cable, likely sapphire, is a better choice to distance the camera, compared to the expensive shielding, which may adequate for the camera housing and circuit boards, but not for the exposed CMOS.

All these methods require multiple runs, into the hundreds to form a 3D picture of the event, and only an 'average' at that (if average exists).

There are published papers on these methods, which is where I learned about them.

>A long optical fiber cable, likely sapphire, is a better choice to distance the camera

Are you going to detonate a B?????

A way to make photos is to illuminate with a laser firing at pulse every 1 microsecond as example and taking a longer exposure. You can use different laser colors every shot. Perhaps you could obtain the photo with less firings. You could discard also bad tests.

> Feel free to message me if you'd like some literature

David and Peter, I have post a message

I read about sapphire fiber in a RG post just last week, and thought to pass it on to you. The expansion of the Pulsotron fusion event, from your web site photos appears to be quite hot, and large.  A quartz window protecting the fiber might survive?

The fusion event in my device is far smaller, and the fiber is not at risk.  I hope.  If I do attempt pictures, then I'll be using regular fiber, and if needed, then behind a quartz window.

While I would like to take pictures, it will not be for many months (years?), until diagnostics warrant such. Trying to take pictures of such a fast event, offers a visual verification (of little value?), where other diagnostic techniques, that are far easier, give consistently more valuable data for the engaged resources.  Of course, the 2D picture provides a 2D map, compared to a 1D linear map from lasers.  So, shape data is acquired faster, and may be of research value or for optimization.

Using different colors to differentiate motion of fireball layer is a good idea.  It will have to be faster than a microsecond.  A nano second more likely.  I could use a far less expensive camera, with a single exposure.  It might work with just LED light, no laser needed, another cost savings.  Thank you.  I'll have to redo the calculations once I find a method to make nanosecond long color flashes (spinning tile mirrored wheel).  It will be interesting, and likely quite easy. Though, the fusion event is filled with white light noise, finding select frequencies to use, requires a spectrum.  Easily done as well, I hope.  As the saturation is high, saturation would be solved with a tiny aperture.  Getting LEDs with the right frequencies is the hard part.  Multiple tunable lasers is not low cost.  Hmm, taking pictures is still a long way off.

While the fusion event fireball velocities and resulting shock waves in small research devices are much faster than any explosive device, they do not scale to macro sizes.  The shock waves quickly die out at atmospheric pressures, within a few inches.  Please do not mention the b word when talking about fusion, as there is enough fear with just the word 'nuclear' as is it.

The quartz windows would withstands but must be cleaned 

It exists leds that withstands lot of amps during short time, and can be operated in clusters. Be careful to use good ones with the same wavelength, because high lumen leds has more variation between them. You can adjust wavelength varying the current. Try to have always the same temperature.

@Javier - Thanks for the info on LED wavelength current dependency, and the cluster idea to overcome white light noise.

Here is a test that I made using 100kg of tnt, the more useful sensors was the acceleration ones, the others did not survived. The laptop  pc that was in a small bunker that collapsed and was  seriously damaged