Somewhere around 10 Kelvin and below, a silicon pn junction stops conducting in the forward direction. We spec the diodes so that at 4.5 Kelvin, it will block (no current ) at three to eight volts. Our magnets are all in series, each has a diode across it. When a magnet quenches (stops superconducting), it will dissipate internally due to IR drop and 7000 amps current. Because the string is so large, it takes 30 seconds to remove all the stored magnetic energy (because the LdI/dt voltage would be too high if we ramp down too fast). During those 30 seconds, the diode across the quenched magnet will see a voltage over it's threshold and will turn on and drop to 1.4 volts VF. That will divert all the string current around the quenched magnet, protecting it from it's mates while it converts all of it's own magnetic energy to heat. Each magnet is designed to be able to absorb it's own energy as heat (self protecting).What voltage drop do you expect from the chilling out semiconductors? Seems it is gonna be tough for them to keep their cool. Ever look at germanium?
It in essence acts like an scr, turning on when needed. But because it is in the helium, there is no need for a set of wires to room temperature where an scr would work properly. The warm to cold transitions put heat into the liquid helium, to the tune of many tens of watts per lead and it takes a kilowatt of refrigeration power for every watt getting to the helium. With over 800 magnets, 3200 watts of heat getting into the cold would require 3.2 megawatts just to recover the losses.
The newer magnets using niobium tin have significantly higher energies stored, so we have to install heaters throughout the magnet to quickly snap the entire magnet to a non superconducting state for uniform dissipation as well as fire external scr's to extract as much energy as possible out of the magnet. We also have to detect a quench in the single digit millisecond timeframe or it's too late, magnet is melted. Now, think ITER's central solenoid with it's 1.5 gigajoules, 40 thousand amps, with it's single turn secondary plasma as it initiates it's million amp current snap on, and think about detecting a few millivolts of IR drop in a 20 kilovolt signal on a 40 kilo amp conductor.
If you think I enjoy playing with these kind of things... You would be correct...
Jn
Last edited:
I can see why you live in fear
Nah, not anymore.
Given the high cost of failure, they understand and support the need to thoroughly analyze, develop, test all the needed systems.
Engineering is all about breaking a complex problem into smaller more manageable subsets, and solving those.
Jn
Ok so you haven't tried germanium diodes. Now if you put the coils in parallel... (Obviously you need non thermal conducting connection wires with low resistance.)
And I thought engineering was about taking vaguely defined problems and narrowing them down to precise issues that then can often be solved by a lesser skilled person or soon to be machine.
Now if you put the coils in parallel... (Obviously you need non thermal conducting connection wires with low resistance.)And I thought engineering was about taking vaguely defined problems and narrowing them down to precise issues that then can often be solved by a lesser skilled person or soon to be machine.
Thats it!! you've solved the problem...
800 magnets in parallel at 7000 amps apiece..spaced 2.4 miles linear.. I would buy stock in a copper wire company.We are actually saying the same thing..
I've got the "lesser skilled person" covered...That's the cover sheet of my personnell file..."lesser skilled"..
jn
Thats it!! you've solved the problem...
jn
Actually sounds like a maxima-minima problem as to what the optimum magnet distribution is via a series parallel configuration to minimize issues such as damage vs power loss.
Of course if you really wanted to protect things you could try some burblebees or some such!
Next we can talk about changing to room temperature superconductors...
You know all the beyond the state of the art stuff.
All the magnets have to have current tracking at the 22 to 24 bit accuracy level with respect to each other.
jn
To be fair to John, he did allow for me doing other things.You coil winder ignorant
A nine year old post? whoa.
jn
You don’t have to dig that deep in the past. A search using the key words “coil winder”, reveals 4 posts by John Curl addressing you in JC III thread alone.
4th Dec 2018, 28th Dec 2018, 27th Aug 2019, 3rd Sep 2019
George
Agreed. However, buried deep in the discussion is the fact that no matter how careful you are at winding an inductor, number of turns on a specific form can't get really close duplication. The air coils we had vendors make required build tolerances of roughly 4% by inductance, 2% by resistance. Once put into an iron dominated system, match was closer.
When all is working, dead silence.
When bad things happen, seismologists in neighboring countries (outside Geneva comes to mind) call you up and ask..."did you feel that?".
For the 22/24 bit DAC system 25 years ago, they actually used a 16 bit and an 8 bit DAC weighted properly in a summing node. They used a high falutin 10 digit dvm (Dual slope IIRC) to create a lookup table for the two DAC combo, as the 16 bit unit was certainly not 24 bit accurate. The table was used to correct the summation so that it could be about 22 bit monotonic. I assume a good S/H was involved to keep transition errors out.
jn
Do changes in air pressure, temp, humidity make noticable differences in the inductance of an air coil?
While trying to find out I came across some new info on the subject, from wikipedia:
"On 20 May 2019, a revision to the SI system went into effect, making the vacuum permeability no longer a constant but rather a value that needs to be determined experimentally;[2] 4π × 1.00000000082(20)×10−7 H⋅m−1 is a recently measured value in the new system. It is proportional to the dimensionless fine-structure constant with no other dependencies.[3][4]"
Whats going on? How much does it vary?
"On 20 May 2019, a revision to the SI system went into effect, making the vacuum permeability no longer a constant but rather a value that needs to be determined experimentally;[2] 4π × 1.00000000082(20)×10−7 H⋅m−1 is a recently measured value in the new system. It is proportional to the dimensionless fine-structure constant with no other dependencies.[3][4]"
Whats going on? How much does it vary?
If they ever get ITER to work properly I can see CERN phoning up and saying 'I'll take 3'.
Dumb question, when things go non-linear in this way, how quickly does the beam dump have to kick in? I assume the loss of a single magnet in the ring will ruin the beam pattern but it will still hold together enough to dump, whereas if a couple go off line at once in a line then bad things may start to occur?
- Status
- Not open for further replies.