Attendees:
M. Zarnstroff
W. Reiersen
E. Lazarus
D. Johnson (author of minutes)
H. Takahashi
E. Fredrickson
B. Stratton
Sensor Designs:
The meeting began with a discussion of various sensor designs. Examples of NSTX sensors were available for inspection. NSTX coils use single layer bare copper wound on Macor form and stabilized with Fortafix ceramic. They are then wrapped in glass fiber sleeving and covered with thin copper electrostatic shield. Leads are generally twisted pair copper with glass insulation and stainless braid material. Contacts are brazed and stabilized in ceramic adhesive. The assembly is baked before insertion. In many cases such an assembly is potted into a carbon tile with ceramic adhesive. Loops are generally made with the cable described above as leads. Typical size for these sensors was 4x30x20 mm for small coils and roughly twice this size for large coils. NA values were not given at the meeting.
Hiro Takahashi showed samples of ceramic 3-D coils that he developed in collaboration with LHD and companies in Japan. These are compact, high temperature sensors, a few of which have been installed in LHD. Hiro distributed a paper describing this development.
Ed Lazarus described the mineral-insulated (MI) cable-based sensors in use at DIII-D. A subsequent e-mail is reproduced below:
I haven't got the official sizes from the engineers. They seem to have difficulty finding drawings. But I spoke with Ted Strait and I have it pretty much correct. Most flux loops are coaxial lines with a copper center and a solid stainless outer conductor. The size is between 30 & 60 mils, probably 40. The insulator is magnesium oxide. They were fabricated by extrusion. The stainless tube serves as the electrostatic shield. A nickel cladding is was added to the center conductor because the copper showed pitting after braising to it. This was thought to be an effect of the extrusion process.
Other details.
They are not tack welded in place, but have little straps from stainless shim stock spot welded in place. Saddle coils are a similar construction.
The saddle coils on the vessel have not moved by a millimeter in the past 12 years.
The flux loops in the inner F coils are not attached to the coil casing as for the outer ones. The inner F coils do not have casings. The flux loops were wound with the coils prior to potting. These do not have an electrostatic shield.
The segmented Rogowski's are not quite as I thought. I was mixing Doublet and DIII-D together. The old segmented Rogowski's had a high failure rate at the junction of leads near the probes. The new ones have been quite reliable and have no sych junctions. The coax is continuous from the feedthrough to the winding and back to the feedthrough. The feedthrough is a bunch of what I know as "housekeeper seals" in a secondary vacuum. They are quite large, an elongated winding about 1x6 cm and a length of 14 cm. These are 1/16" coaxial line.
The Rogowski outside the vessel is copper wound on teflon with a center conductor as well. There have been problems in the past with the teflon losing structural integrity and deforming. They are carefully kept off the vessel wall and heat sunk to cold structure.
It was generally agreed that we should look closely at the use of small MI cable and understand its limits and termination techniques.
Numerical Approach to
Sensor Optimization:
Ed Lazarus then summarized his approach to sensor optimization, which involves computing the signals on an overdense array of sensors on several control surfaces due to the coil currents and the plasma for a wide range of equilibria. Based on these responses, he will then numerically optimize to a finite set of sensors for both plasma control and equilibrium reconstruction. This optimization was the topic of several meetings held this same week at PPPL. It is hoped that preliminary optimizations would be available in the next 6-12 months.
Co-wound Sensor
Loops:
There was general agreement that we should be planning on a number of co-wound sensor loops, which would be part of the field coil fabrications. In each case they would likely be MI cable with a diameter of 40-60 mils, which would be embedded in epoxy with protected lead connections. We reached agreement on the following:
Modular coils 2 at base of tee, 2 near top of tee for each coil
TF coils 1 at plasma side of each coil
Solenoid 1 each at top and bottom in gaps between solenoid coils
PF 1 at plasma side of each coil
Trim coils 1 at plasma side of each coil
Interfacing with
Vessel Design:
We discussed with Wayne Reiersen the inputs required by the vessel designers prior to the PDR. These are listed below. Specific action items were not defined at the meeting, although some target dates were given.
Spatial build for internal and external sensors and leads - For sensor loops on exterior vessel surface, we discussed the use of MI cable with .040-.060 diameter, which would be wound in various shapes on the outer surface underneath the vessel cooling tubes. On the inside of the vessel, a combination of pickup coils and loops would be required, probably attached directly to the wall by spot welding.
Lead protection and termination - Since many sensors will have to be installed before the machine is assembled, protection of the sensors and also the leads will be important to avoid costly retrofits. Concepts need to be defined for routing of sensor leads and for termination at feedthrus (for in-vessel sensors) and terminals (for external sensors). A concept to provide access to terminals through modular coil structure would also be desireable. It was generally agreed that we should avoid connections between vessel segments for outboard sensors, unless they can be easily accomplished by interconnecting terminals through modular coil structure. Some in-vessel sensors could cross joints, since they could be installed after VV is assembled. If in-vessel ribs are installed, they need to have many slots where they meet the vessel, providing channels to route leads and loops through the rib.
Feedthru locations - The magnetics will require hundreds of conductors with vacuum feedthrus. Providing adequate port space for these feedthrus, compatible with other port needs, requires attention in the near term, to be ready for the PDR. Also, to avoid long lead lengths, the distribution of such feedthrus may be important. Making a 2-D map of the inner vessel surface showing ports, antenna, etc would be a valuable tool in this effort.
Target dates:
Size of wire for outside vessel loops end of April
Guess at number of sensors end of April
Concept for electrical feedthru distribution end of April
Concept for inside loops crossing assembly joints end of April