WB-8 Coming

[MSimon]

Good idea. Except that you would need some kind of custom manufacturing setup of considerable complexity to wind and impregnate the coils. Even square plan coils are going to (most likely) require several iterations to get it right. That is the beauty of MRI coils. They are already in production.

Not so difficult. Winding the two types of magnets (squares and triangles) is easy. Going with bow-legged squares and triangles vice straight legged would be mostly a wash in the winding. Soldering on the casing would be the hardest part, and that shouldn't be too difficult for an accomplished solderer.

If you want to do LN cooled, we probably would be best served by going with the modified MPG winding method discussed in other posts. That would start getting complex, but I think it would still be doable. Again, the casing would be the hardest part.

I believe a little more study of the difficulties is in order. The quench problem is significant. And it has a lot to do with how the coils are made.

Of course if there are "unlimited" funds - no problem. Budget $5 million or $20 million for a custom winding facility and various tests to work out the bugs. Vs a million or two for off the shelf coils and a few hundred thousand for supports.

[MSimon]

Kiteman,

I think our differences are that you want to do the very best.

I want to do the best given likely budget limitations.

I'm all for blue sky and flying unicorns. Show me the money.

[KitemanSA]

Kiteman,

I think our differences are that you want to do the very best.

I want to do the best given likely budget limitations.

I'm all for blue sky and flying unicorns. Show me the money.

I am not sure you do know what I want to do. Your comments about blue sky and unicorns suggest you have religiously kept you eyes and ears covered and your mind closed.

I want to try out other magnet shapes to obtain sphericity/transport data. I did not originate this idea, DrB did. I just happen to think it is a good one. The exact unit(s) to build should probably be based on best available simulations. Building and testing other magnet shapes in Cu at small scale should not be difficult. With some thought (an I have given this some) it MAY actually turn out that winding a Cu tube LN cooled magnet set would be as easy as winding a wire-wound uncooled magnet. That is for someone more expert on magnet winding to deside, but I have some suggestions for them if they wish to listen.

Regarding quenching, AFAIK it is a property of SC magnets. I have not proposed doing SC magnets except to state that once it is decided to build a full scale unit, the magnet shapes should probably be a minor issue. I am not even sure LN cooled Cu is needed to get the data at the small scales I want to see.

Once the magnet form factors wrt transport are better understood, EMC2 can deside to either go for a 100th power unit as you suggest or go full scale. Going 100th power just for scaling might be most efficiently done using COTS SC magnets. If so, then that is what should be built. Indeed, once the data are in hand it may prove most economical to do a simple full scale toroid ala WB7. Maybe not. But in any case, good data on other magnet shapes is important for good design of a full scale unit.

[MSimon]

I did not originate this idea, DrB did. I just happen to think it is a good one.

Show me the money.

There are a lot of good things to try if the resources are there. If not choices must be made.

We do know from the ITER experiences that bigger and longer are different.

Or as Joe used to say. Quantity has its own quality.

I think bigger machines and longer operation should be attempted as soon as possible even if it means skipping some things that may prove long term useful.

[KitemanSA]

I did not originate this idea, DrB did. I just happen to think it is a good one.

Show me the money.
There are a lot of good things to try if the resources are there. If not choices must be made.
We do know from the ITER experiences that bigger and longer are different.
Or as Joe used to say. Quantity has its own quality.

Hmm. I was under the impression that you considered Joe's philosophy to be defunct. Now you quote him? Tsk tsk.

I think bigger machines and longer operation should be attempted as soon as possible even if it means skipping some things that may prove long term useful.

The you and I agree to disagree on this. I do hope you get the $ to do your 100th power unit. I hope EMC2 is getting the $ to do the other magnet shapes study. I wil continue to work on other avenues to do the small scale tests. If done by slav... unh student labor, such a unit may actually cost nearly nothing to build.

[MSimon]

I did not originate this idea, DrB did. I just happen to think it is a good one.

Show me the money.
There are a lot of good things to try if the resources are there. If not choices must be made.
We do know from the ITER experiences that bigger and longer are different.
Or as Joe used to say. Quantity has its own quality.

Hmm. I was under the impression that you considered Joe's philosophy to be defunct. Now you quote him? Tsk tsk. :wink:

I think bigger machines and longer operation should be attempted as soon as possible even if it means skipping some things that may prove long term useful.

The you and I agree to disagree on this. I do hope you get the $ to do your 100th power unit. I hope EMC2 is getting the $ to do the other magnet shapes study. I wil continue to work on other avenues to do the small scale tests. If done by slav... unh student labor, such a unit may actually cost nearly nothing to build.

Mao's total philosophy is defunct as well and yet as the writers of the American Constitution understood: "Political power does come from the barrel of a gun".

Something true spoken by an evil one does not make it false.

If done by slav... unh student labor, such a unit may actually cost nearly nothing to build.

I guess you and Joe are not too far apart after all.

My theory is to hire the very best I can find at every level. The cost of mistakes avoided generally pays for the added labor expense.

===

IMO funding for further experiments are better attracted by a Q=1 machine rather than 5X improvements in a Q=.00001 machine.

Jump to light speed.

[Aero]

The Wright brothers could have waited for a more powerful engine, or metal construction technology, either of which would have allowed an improved flying machine. But they took the attitude that if its good enough to fly, lets fly it. Flight technology is much further along today than it would have been if they had waited for a better understanding of the technology.

To the point of the BFR; If it flies, fly it, and improve it later. Just MHO.

[ravingdave]

Good idea. Except that you would need some kind of custom manufacturing setup of considerable complexity to wind and impregnate the coils. Even square plan coils are going to (most likely) require several iterations to get it right. That is the beauty of MRI coils. They are already in production.

Not so difficult. Winding the two types of magnets (squares and triangles) is easy. Going with bow-legged squares and triangles vice straight legged would be mostly a wash in the winding. Soldering on the casing would be the hardest part, and that shouldn't be too difficult for an accomplished solderer.

If you want to do LN cooled, we probably would be best served by going with the modified MPG winding method discussed in other posts. That would start getting complex, but I think it would still be doable. Again, the casing would be the hardest part.

I believe a little more study of the difficulties is in order. The quench problem is significant. And it has a lot to do with how the coils are made.

Of course if there are "unlimited" funds - no problem. Budget $5 million or $20 million for a custom winding facility and various tests to work out the bugs. Vs a million or two for off the shelf coils and a few hundred thousand for supports.

I have been wanting to bring this up, and since your message touches on it, now seems as good a time as any.

Back when I was looking at how difficult it would be for an amateur to build an experimental reactor, I was trying to learn how they "charge" a superconducting magnet. I found various sources that reiterated the idea of heating a portion of the magnet then applying a voltage across it. As it turns out, the people who manufacture these things, also manufacture machines (power supplies) that charge the superconductors. Apparently SuperConductors REALLY REALLY REALLY don't like changes in their magnetic field, and for this reason, the current must be built up very slowly. The machines for charging a superconducting magnet are designed to automatically ramp up the current at rates which are settable by the operator.

The part I found most interesting, and likewise most serious, is that superconductors REALLY REALLY REALLY don't like changes in their magnetic field. Ramp the field up too quickly, and the Magnet "quenches". Likewise, if you disturb the field too much, the magnet "quenches."

Apparently charging a superconducting magnet is a tricky business.

In this same pdf file, it mentioned that the problems of charging more than one superconductor in near proximity become compounded immensely! If the fields are capable of interacting, (an absolute given in a polywell) then any rapid increase (or decrease) in the field of one will quench them both. And this is just with TWO superconducting magnets ! Imagine the hassle with SIX !

In any case, it looks like someone is going to have to design a machine that can successfully charge six superconducting magnets at the same time, all in close proximity to each other. Sure, it's likely dooable, but it's just another layer of difficulty to deal with. By the way, if you charge up a superconducting magnet high enough, and then "quench" it, you are discharging all that energy into a no longer superconducting conductor. You can very easily "Fry" the magnet.

Another added difficulty is how to ramp up the wiffleball and alpha flow without creating a rapidbly changing magnetic field capable of quenching the magnets. (Remeber, if you quench any one of them, it will probably result in a cascade quenching of all the rest ! )


Can you imagine what would happen if you were operating a 100 MW polywell and the magnetic confinement suddenly quenched ? Tom Ligon's description of what happened with PXL something or other is probably a good indicator.


Something to think about. When I get more time, i'll try to find that Magnet charging operators manual I had found previously.


David

[Barry Kirk]

I just have to say it....

"Captain, the engines are shutdown cold, we can't restart them for 30 minutes... There is no way to change the laws of physics..."

Lt. Commander Montgomery Scott, USS Enterprise


or something like that... I'd have to look up the exact quote.

[MSimon]

ravingdave,

I posted a link in the last week or two about charging up a SC magnet for a satellite experiment. It was most informative.

http://www.scientificmagnetics.co.uk/pd ... ion-02.pdf

viewtopic.php?p=19866#19866

One of the reasons I favor MgB is that it is conductive even when superconductivity fails. I'm also encouraged that YBCO SCs are used in high field AC motors which are by nature magnetic situations with large field variations at sub-millisecond scales.

I estimate that a lot of how to fix all this is dependent on air gaps. Or vacuum gaps in the case of Polywell.

What concerns me about square form coils is the quench potential of having forces concentrated at the hinges (corners). If you can't fix that then the gain advantage can't be practically realized vs the higher fields of SCs.

As is usual in most engineering - the devil is in the details. And for a "high cost" experimental set up I'd prefer to stick with objects that have dealt with the details in production quantities as much as possible. As a young engineer I was entranced by possibilities. Most of which I was unable to realize due to the difficulties. Now I live, eat, breathe, and will probably die with KISS.

[MSimon]

Re: magnet frying. It is normally handled by using a constant current power supply with a diode and an energy absorbing resistor in series. Most of the energy is absorbed in the resistor. The power ratio being coil resistance (non-superconducting) divided by energy absorbing resistor value - and if you want to get exact (probably not necessary) including diode losses. And you also need piping and relief valves capable of handling coolant vaporization. There are in fact ASME rules for dealing with this.

No doubt all this is incorporated in MRI magnet systems. A very good reason for sticking with known technology for now.

IIRC the SC magnets are ramped up at a 5 mT per minute rate. Which means about 17 hours of charging for a 5 T magnet. Faster rates are probably possible once the magnets are broken in.

[KitemanSA]

Good idea. Except that you would need some kind of custom manufacturing setup of considerable complexity to wind and impregnate the coils. Even square plan coils are going to (most likely) require several iterations to get it right. That is the beauty of MRI coils. They are already in production.

Not so difficult. Winding the two types of magnets (squares and triangles) is easy. Going with bow-legged squares and triangles vice straight legged would be mostly a wash in the winding. Soldering on the casing would be the hardest part, and that shouldn't be too difficult for an accomplished solderer.

If you want to do LN cooled, we probably would be best served by going with the modified MPG winding method discussed in other posts. That would start getting complex, but I think it would still be doable. Again, the casing would be the hardest part.

I have re-thought this last part and now believe there is an easier way to wrap a LN cooled, Copper, bow-legged, square plan-form Polywell. Boy that's a mouthful! It should actually be VERY simple to do.

[ravingdave]

ravingdave,

I posted a link in the last week or two about charging up a SC magnet for a satellite experiment. It was most informative.

http://www.scientificmagnetics.co.uk/pd ... ion-02.pdf

viewtopic.php?p=19866#19866

One of the reasons I favor MgB is that it is conductive even when superconductivity fails. I'm also encouraged that YBCO SCs are used in high field AC motors which are by nature magnetic situations with large field variations at sub-millisecond scales.

I estimate that a lot of how to fix all this is dependent on air gaps. Or vacuum gaps in the case of Polywell.

What concerns me about square form coils is the quench potential of having forces concentrated at the hinges (corners). If you can't fix that then the gain advantage can't be practically realized vs the higher fields of SCs.

As is usual in most engineering - the devil is in the details. And for a "high cost" experimental set up I'd prefer to stick with objects that have dealt with the details in production quantities as much as possible. As a young engineer I was entranced by possibilities. Most of which I was unable to realize due to the difficulties. Now I live, eat, breathe, and will probably die with KISS.

I read that when you posted it. It is not the same file I ran across, but I think it may have come from the same company. Scientific Magnetics sounds familiar. What I had read was more like an operators manual on using their device to charge up a superconductor.

Interestingly enough, I recall reading that 5 volts was the nominal charging voltage, and I was suprised to discover that you could create a five volt voltage drop across a piece of heated superconductor. I was thinking that even in it's non-conducting state, a superconductor would still have a very small resistance, but once you started pumping 2500 watts into it, (500 amps at 5 volts) it would quickly heat up.

For some reason I had always thought they would use millivolts, and I wondered what kind of power supply would produce millivolts at hundreds of amps, all the while being especially clean power. The rectifier diodes would have been using more power than the coil !


In any case, it appears superconductors are a lot more finicky than I had first realized, and i'm not sure how many people who read this website are really aware of just how finicky and temperamental they are !



David

[MSimon]

Assume the battery is a constant current variable supply.

Edit: The resistor is the quench resistor. BTW the coil inductance will be well into the Henry range as far as inductance goes. The supply will probably be controlled by a monotonic 14 to 16 bit D to A with some kind of integrator on the output to smooth the steps. The integrator will probably have a time constant on the order of tens of seconds.

The power supply blocking diode should have a PIV in the hundreds of volts to maximize the amount of energy absorbed by the quench resistor. The quench circuit diode only needs a few tens of volts of PIV but I would probably use the same diode as the power supply (if it is external to the supply) to minimize inventory.

David,

The 5 volts of the supply is mostly to handle external losses. Even then the total circuit resistance should be under .01 ohms (5 V @ 500 A).

[ravingdave]

Assume the battery is a constant current variable supply.

Edit: The resistor is the quench resistor. BTW the coil inductance will be well into the Henry range as far as inductance goes. The supply will probably be controlled by a monotonic 14 to 16 bit A to D with some kind of integrator on the output to smooth the steps. The integrator will probably have a time constant on the order of tens of seconds.

If I am following you correctly, I think you are suggesting that the power supply has some sort of series regulator using a D to A converter to control it's output ?

I am thinking that a switching power supply would work better. Just use a PWM feedback control loop in conjunction with a D to A converter controlled by a microprocessor. (With LARGE output capacitors.)
With the right code (and hardware) you could get better than 65,536 steps. In fact you could get your resolution down to whatever degree you wanted it.

The power supply blocking diode should have a PIV in the hundreds of volts to maximize the amount of energy absorbed by the quench resistor. The quench circuit diode only needs a few tens of volts of PIV but I would probably use the same diode as the power supply (if it is external to the supply) to minimize inventory.

I'm not sure I understand exactly. My understanding of a controlled quench is that you slowly heat a segment of the coil, and that the resistance will slowly leak off all the energy in the form of heat until the coil is completely quenched.

In the case of an uncontrolled quench, some part of the coil goes resistive first, the energy starts to get converted to heat causing a rapid decrease in the magnetic field which then causes the entire superconductor to go resistive, which then converts ALL the remaining energy into heat.

David,

The 5 volts of the supply is mostly to handle external losses. Even then the total circuit resistance should be under .01 ohms (5 V @ 500 A).

That's probably the case. .01 ohms must represent the Total resistance of the entire circuit. The heated superconducting segment resistance has to be much lower, or you would be dissipating 2500 watts into that segment. I find it hard to believe that a superconducting segment can be cooled
down to a superconducting condition while it is dissipating 2500 watts !


It is probably all the external wiring (and series resistor) that is actually dissipating most of that heat.

Most likely, the segment resistance (while non conducting) is far less.

This begs the question. What IS the resistance of a heated segment of superconductor ?

Hmmm.....


David