Instalation of the COMPASSD tokamak in IPP Prague

1
Zpetnovazební rízení vertikální polohy plazmatu
tokamaku COMPASS-DJan Horácek
2
Poloidal coils on COMPASS-D
Jen nekolik cívek, ale 4 ruzné obvody "M"
vinutí, kontrolované pomoci Magnetising Field
Power Supply (MFPS), poskytuje nezbytnou kontrolu
loop voltage plazmatu, tj. dIM/dt?Vloopgt0 "E"
vinutí, kontrolované pomoci Equilibrium Field
Power Supply (EFPS), reguluje polohu a tvar
plazmatu, "S" vinutí, kontrolované pomoci
Shaping Field Power Supply (SFPS), poskytuje
ruzné tvary plazmatu, "F" vinutí poskytuje
kontrolu polohy plazmatu pomocí zpetné vazby.
3
Zpetná vazba

• Klícová soucást tokamaku je schopnost kontrolovat
  polohu plazmatického sloupce. Ta je
  diagnostikována merením magnetických polí vne
  plazmatu pomocí flux loops, saddle coils (Bz) and
  partial Rogowski coils. Zatímco horizontální
  poloha je stabilní (závisí jen na tlaku
  plazmatu), poloha vertikální je nestabilní a
  vyaduje tudí rychlou zpetnou vazbu s odezvou do
  100 mikrosekund.

4
Questions for Culham

• Is the electronics inside the MR cubicles still
  in place or even operational as described by W.
  Morris in 1997?
• Where to find electrical circuits of components
  of the feedback diagrams? What are the settings
  of the PID controllers, time constants of the
  integrators of the coil signals ...
• Can W. Morris come to COMPASS in Culham and help
  us to identify all the diagram symbols within the
  existing hardware inside the MR cubicle?
• How can we read the waveforms? Either through SAS
  (please provide the corresponding code) or using
  file SHOTnnnnn.M01 but where is this file?
• The F-coils have inductances L50mH and
  resistances R1mW. This yields the fastest
  responds of L/R0.05s which is too slow for a
  feedback. Strange
• If the actual analogue system exists and is still
  operational, then use it. If not, develop a new
  one, fully digital.

5
Status of the feedback group 18.1.2007

• We know spatial positions of the measuring and
  acting coils
• We can read all the signals concerning the
  feedback subsystems for several shots
• We are now identifying what these signals mean
  and how they correspond to the existing documents
  (diagrams)
• Olena found what waveforms mean

6
Radial feedback affected by ELMs

• ELMy H-mode, ELMs at 350-700Hz
• This frequency is strongly visible on the radial
  feedback, even at the Fast Amplifier input.
  Signal from the amplifier itself not available ?.

7
Waveforms

• Waveform 4 is not just a signal, its an active
  device that modifies a signal, something like
  outdin/dt.
• In contrary, block of Waveform 3 and 11 and Opto
  modifies the signal only by a constant factor
  -17x.

8
Jirka Vlcek

• He works practically with operational amplifiers,
  PID controllers, fast amplifiers

9
Matlab Simulink PID controllerMichal Stránský

• Simulink a Matlab umonují simulaci rízeného
  systému, návrh a odladení PID controlleru.
• We use it currently from Lausanne but I recommend
  to buy it for Prague

10
Michal Stransky

• Pro identifikaci modelu cástí rízeného systému
  lze pouít nástroj Ident z programového balíku
  Matlab, který najde prenosovou funkci v
  Laplace-prostoru pro nejaká data vstupu a
  výstupu.
• Zkoueli jsme pro radial position control to
  identify what the PID controller transfer
  function is but the result was only that there is
  no integral-term in the PID during shot 30868.

11
What we should do?

• Transfer all COMPASS database to Prague and read
  it in IDL/Matlab as soon as possible. In
  principle, we know reading it by XPAD in IDL so
  should be easy to write a reading script

12
08/02/2007

• Found and put on web new documents
• Dizertace Luigi Scibile Non-linear control of
  the plasma vertical position in a tokamak, mj.
  Compass-D
• Clanek od P. Vyase 1997 na Vertical position
  control on Compass-D. Napsal jsem mu email jestli
  nechce spolupracovat. Bydli v Mnichove. Jeho
  disertace se hleda (Mlynar)
• V clanku Gossnera je preklep 20MHz-gt20kHz. Tudiz
  casove skaly vertikalni nestability jsou 2500s-1,
  za 10ms je to pryc.
• Zjistil jsem, ze 12x50600Hz sitova porucha
  nerezonuje s ELMy ? pro 30868
• J. Vlcek navrhl rychlý zesilovac na zpetnou vazbu
  a vypékání tokamaku.
• První cást odpovedi z Culhamu jsme dostali

13
ELMs drop plasma pressure by 6seen on
conditional-averaged bpol
Time delay between pressure drop and ELM is
likely due to incorrect data acquisition
synchronisation.