Re: OT Question for Iggy

I guess I'm constitutionally incapable of a true cliff-notes length answer,
but this is my short version:

It's a 6" horizontal room temperature bore 3 tesla persistent mode
superconducting magnet with a homogenous cylindrical volume of about 2"
diameter by 4" long with a uniformity of 10 ppm (parts per million). There
are actually 10 superconducting shims that can improve this to 1 ppm but for
this experiment that's not needed. It's the magnet for a Fourier transform
mass spectrometer. That's the what for, the what is that it has about nine
miles of superconducting wire on the order of 0.1 mm diameter wound into a
solenoidal coil about 30" long, ID about 8", and a complete guess on the OD
is12". This is in a dewar so it can be submerged in liquid helium at 4.2
kelvin, surrounded by a radiation shield thermally connected to the top of
the He dewar boiloff tubes so it stays at about 20 K, which is then
surrounded by a liquid nitrogen dewar at 77 K which is wrapped loosely with
20-50 layers of superinsulation (very thin aluminized mylar for thermal
radiation shielding) which is then surrounded by the outer cryostat housing
at room temperature. Down the bore there are three tubes within tubes - the
smallest is at room temperature, the next is mounted at the ends to the
nitrogen dewar, and the next is mounted at the ends to that 20 K radiation
shield. The gaps between tubes and between the 20 K tube and the helium
dewar are about 1/4" (on the radius) so the thermal gradient there is pretty
steep and that's mostly what drives the helium boiloff rate. Takes about 30
liters of liquid He to keep the coil submerged, and the dewar holds another
30-40 L above that for a hold time of about 2 months. I forget the nitrogen
capacity but a 160 L dewar provides a refill every 3-4 days and lasts 2
weeks. Go look at ebay item 310164162456 for a vertical bore version from
the same company, Oxford Instruments (this magnet has higher field, smaller
bore, and about the same homogeneity after shimming the superconducting
shims). Outer cryostat diameter is about 36-40". New back in the early
1980's this was a $50,000 or so magnet.

For whatever reason Oxford favored small diameter wire and lots of turns,
which has the effect of making the inductance of the coil very large. I
think this magnet is about 30 henries and it reaches full field at 36 amps
or so. This inductance limits the rate of change of current because you
want to keep the voltage as low as possible. The superconducting wire in
this case is single filament niobium titanium embedded in copper. Start
with a tube of niobium, put a titanium rod down the middle, and draw it down
to some mm diameter (I could have the metals reversed). Put that in a
copper tube and draw that down to under 1 mm diameter and maybe 1-5 miles
per spool. Put that in a furnace for days and form a cylindrical shell of
the superconducting niobium-titanium alloy, which is used because it works
and is flexible :-). Okay, it can be used up to maybe 5 tesla in this form
at current densities of 10^5 amp/cm^2. To go higher do all this but make
multiple filaments which gets you to 7-11 tesla depending on bore size. To
go higher make a smaller coil wound with similar wire but niobium-tin and
fire that coil after winding because that alloy can't be bent without
breaking and put that coil inside a larger coil made from the first wire (to
minimize the amount of expensive wire) to get to 15-18 tesla and $2-4
million per magnet at the upper end for a room temperature bore of 6".
Okay, back to the single filament wire - wind that on an aluminum bobbin,
bringing the ends out each time a spool runs out. After winding, splice the
ends by crimping and maybe soldering with pure lead and support these on
stalks so they stick up like a porcupine. This gets them to a low enough
magnetic field that they stay superconducting - the critical current density
and magnetic field of the joints is markedly inferior to the wire.
Somewhere along the way vacuum impregnate the coil with epoxy so the wire
cannot move under the hoop stress induced by the magnetic field. Take the
very ends of the solenoid and connect a special piece of superconducting
wire which has a heater wrapped around it and support this somewhere towards
the top of helium dewar so when you heat up the heater you don't boil off
all the liquid. Wire some power resistors in parallel with the switch, put
the coil in the dewar, weld everything together, pull a vacuum on the dewar,
cool everything down, and connect a power supply across the coil. Other
manufacturers chose bigger wire and fewer turns, like ebay item 120465624954
made by Magnex that runs at 296 amps and can be charged in 5 minutes but now
all the wires from power supply to coil have to be that much bigger and will
carry that much more heat into the dewar, boiling off that much more helium
during charging. In the end I think they all work and none is obviously
better :-).

Okay, back to charging our magnet. First connect everything and set the
power supply to zero volts and zero amps. Then set the voltage to maybe 2-3
volts and slowly raise the current limit so that over maybe 10-30 seconds
you get to full current of 36 amps. At this point the current flows from
power supply to the junction at one end of the coil, through the
superconducting switch which is still cold at this point and takes the
current because the coil inductance makes that current rise very, very slow
at this point, through the switch, and back to the power supply. Lock the
current limit knob if possible and turn the voltage back to zero. This sets
the supply so you don't try to run the current up too high and tests all the
connections. Now warm up the switch by applying about 5 V and 20 mA to the
switch heater and waiting 10-30 seconds for it to go non-superconducting.
Now slowly turn the output voltage up to 3.5-4 V (I'm working from old
memories so could be off a volt or an amp here or there). Now the current
will flow from supply through the coil and back to supply, with a tiny bit
going through the switch since it's resistance is now tens of ohms, and the
current will rise at a rate set by the inductance and the applied voltage.
The copper matrix that the superconducting alloy is embedded in will also
take a little current since it is also in parallel, and this current will
make a little heat which must be minimized so that the wire doesn't warm up
and go non-superconducting. Stare at the meters and the clock for a couple
of hours until the current gets up to 20-25 A and the field up to about 2/3
of full value, then turn the voltage down to about 2.5 or 3 V to slow down
the rate of rise and stare some more. Somewhere near 32 A out of 36 A total
slow down again, and let the current get to the final value. Now gird your
loins and grab the current limit knob and tweak it slightly to bump the
current up about 0.005%, leave it for a few minutes, and then turn it back
down to 36 A. This slight overcurrent empirically was found to make the
final field much more stable. The fear is from knowing that you are already
right at the edge of the critical current density at that field and now you
are creeping a little closer. (If you weren't that close to the edge they
would rate the magnet higher and put you right back on the edge; wire is
expensive :-).) If you go too far the wire goes normal in some spot which
then gets hot which then makes the rest of the coil normal, all in about a
millisecond. Enough heat is released from the energy stored in the magnetic
field to boil off all the liquid helium and blow it into the room in about
30 seconds, and to warm the coil so somewhere near 100 K or warmer.
Stopping the current that fast through that large an inductor leads to
horrific voltage spikes, potentially thousands of volts, which in the old
days lead to arcs and destroyed coils. Remember those power resistors?
That's why they are there, to snub the voltage in case of a quench. Okay,
now the current is back down that smidge and the field is stable where you
want it, so turn off the heater on the switch and let it go superconducting
again. Now the current flows out of the coil, through the switch, and back
into the coil, around and around with no resistance. The power supply
current flows to the switch, through the switch in the opposite direction
(so the net switch current is zero but this is the best way to visualize
things), and back to the power supply. Slowly turn the voltage down to
zero, and then the current down to zero, and disconnect the supply from the
magnet. Voila! a persistent mode superconducting magnet lives. With an
Oxford magnet the total time is 2-3 hours. Single filament joints like this
can be very, very good so that the power dissipated in their residual
resistance makes the magnetic field decay maybe 50-100 parts per billion per
hour. Joints with multifilament niobium-titanium aren't as good and there
are usually more of them because you only use that wire when you need higher
field which means more turns, so those magnets can decay a part per million
per hour or so. Remember the superconducting shims? Those are windings of
various shapes to produce orthogonal field gradients to correct defects in
the field from the main coil. Some of them are shaped so that they couple
to the main coil like the secondary of a transformer. Each shim has its own
switch and heater, and some of those must be turned on and left on during
the entire charging process with a resistor across the switch to dissipate
the energy to keep from inducing too large a current in the shim. If
needed, after the main coil is persistent you then connect a separate supply
to each shim and set the currents to the appropriate values, then make those
switches superconducting as before.

Oh, no ferrous material anywhere inside the coil because it would saturate
near 2 tesla and distort and limit the field. This is the same technology
used in MRI magnets. The cost of a magnet is mostly the cost of the wire,
and the amount of wire is some function of the room temperature bore size,
the homogenous volume, and the field strength. MRI magnets tend to be much
bigger bore so people can fit into them, and to keep the cost under control
they tend to be lower field. I think early ones were 1.5-2 tesla and cost
1/2 to 1 million. Nowadays they are 3-4 tesla and cost less - the miracle
of volume production. Anyway, that's enough for now. Be glad to answer any
other questions.

Iggy, that was a great price. I'm hoping for under $200 and I have some
time to be patient.

-----
Regards,
Carl Ijames

"Wes" <clutch@xxxxxxxxx> wrote in message
news:A5Ynm.725189$4p1.664484@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
"Carl Ijames" <no@xxxxxx> wrote:

I told my old adviser
I'd charge a 3 tesla supercon magnet for him when he gets an experiment
ready to go, after I heard that he got a quote of $28k from the
manufacturer. That's a bit extreme considering that they used to charge
$5k
20 years ago and that included a day spent shimming the field with an nmr
probe that they supplied - no shimming needed on this one. I actually
charged this magnet once back when I was a student, after repairing a
vacuum
leak and pumping the cryostat back down, but we borrowed a supply back
then.


Carl,

Could you give us the Cliff notes version of what you are doing? Is this
a real magnet
out of ferous material or a ring with current flowing around in it super
cooled?

Just curious,

Wes


.

Relevant Pages

  • Re: TOTAN Ramp Magnet Coil and Ramp Divertor Problem
    ... first thing I would check the solenoid (and magnet) ... > My ramp magnet coil does not work. ... > I checked continuity from the power wire on coil 8 to J133-1. ...
    (rec.games.pinball)
  • Re: trying to make an electromagnet
    ... > long while means that the shortest piece of wire you can wind on the ... I have built a circuit which sends a short pulse through the ... magnet whenever an LDR goes from light to dark (i.e. to water at the end ... What i am planning to do is to split the coil i have into 4 parts of about ...
    (sci.electronics.basics)
  • Re: Electrical Insanity with 87 CBR600
    ... There is NO moving armature inside the pulser coil. ... In order to mechanically generate electricity, ... wire, a magnetic field, and *something* has to move. ... magnet past the coil of wire to get a pulse, ...
    (rec.motorcycles.tech)
  • Re: Magnetic induction question.....
    ... electrons are moving through the wire in one direction. ... and a magnet is passed near the ... coil, will the coil generate an induced magnetic field opposing that of ... They are called eddy currents. ...
    (sci.physics.electromag)
  • Re: DiMarzio humbucker in an Ibanez guitar (5 way switch)
    ... and then take the bare and solder it onto a separate ground lug. ... Next - take the red wire and wire it the same as the hot wire on the ... your coil tap) to where the white lead was on the Ibanez. ... some pickups DiMarzio makes have good designs for getting both humbucker and ...
    (alt.guitar)