, THMP Calibration Miriam K¨ ummel Ruhr-Universit¨ at Bochum Institut f¨ ur Experimentalphysik I PANDA-Collaboration Meeting March 2016 1
, Why Do We Measure Temperatures? PWO-II: LY depends on T with d ( LY ) = 3 %/ ❽ at -25 ❽ dT Goal for PANDA: ∆ T < 0 . 1 ❽ → sensors with σ T < 0 . 02 ❽ 2
, How Do We Measure Temperatures? The resistance of most materials depends on the temperature Platinum best material for resistance thermometers: R vs T relationship quite linear α Pt = 3 . 89 · 10 − 3 K − 1 R ( T ) ≈ R (0 ❽ )(1 + α Pt · T ) , For commercial Pt100 sensors R (0 ❽ ) = 100 Ω, so that R ( − 25 ❽ ) ≈ 90 . 23 Ω For our manufactured sensors R (0 ❽ ) ≈ 105 Ω, so that R ( − 25 ❽ ) ≈ 94 . 7 Ω The measurement accuracy σ T for temperature sensors, translates to a measurement accuracy σ R for the THMP: σ R ≈ ∂ R ( T ) · σ T ≈ 7 . 8 mΩ ∂ T ⇒ σ R R ≈ 0 . 08% = 80 ppm 3
, How Do We Measure Resistances? 4-wire measurement constant current I = U I R I gain G = 5 + 200 kΩ R G offset voltage U off ADC conversion factor C ( R · I · G + U off ) · C = R · I · G · C + U off · C = N � �� � � �� � m n 1 − n ⇔ R = N + m m ���� ���� p 1 p 0 → σ R corresponds to 3.5 ADC channels for optimized values of U I , R I , R G and U off 4
, How Do We Measure Resistances Accurately? ! Electronic components vary within their production accuracy all readout channel k on a PBB are different! U I constant current I k = R I , k gain G k = 5 + 200 kΩ R G , k offset voltage U off ADC conversion factor C + U off · C = N k R k · I k · G k · C � �� � � �� � m k n 1 − n ⇔ R k = N k + m k m k ���� ���� p 1 , k p 0 , k → Calibrate each channel by fitting a 1 st order polynomial to pairs of known resistors and measured ADC conversions! 5
, How Do We Measure Resistances with Respect to the AT? ! resistances depend on the ambient temperature T for R I , k , R G , k is (∆ R / R ) / ∆ T = 10 ppm/K the best available measurement of stable resistors 50 94.70 measured resistance ∼ R / Ω U I THMP temperature ∼ T / ◦ C constant current I k ( T ) = R I , k ( T ) 40 94.68 30 94.66 200 kΩ gain G k ( T ) = 5 + R G , k ( T ) 20 94.64 10 94.62 offset voltage U off 0 94.60 0 10 20 30 40 50 60 ADC conversion factor C t / h R k · I k ( T ) · G k ( T ) · C + U off · C = N k � �� � � �� � n m k ( T ) 1 − n ⇔ R k = N k ( T ) + m k ( T ) m k ( T ) � �� � � �� � p 1 , k ( T ) p 0 , k ( T ) → Fit a 1 st order polynomial at different temperatures, describe dependency on T & monitor THMP temperature! 6
, How Do We Measure Resistances Reliably? ! There is an additional PBB-wide drift of unknown origin ( t )! measurement of stable resistors U I ( t ) measured resistance ∼ R / Ω constant current I k ( T , t ) = 94.48 k=6 R I , k ( T ) k=4 94.47 200 kΩ gain G k ( T ) = 5 + 94.46 R G , k ( T ) 94.45 offset voltage U off ( t ) 94.44 0 2 4 6 8 10 12 14 16 ADC conversion factor C t / d R k · I k ( T , t ) · G k ( T ) · C + U off ( t ) · C = N k ( T , t ) � �� � � �� � m k ( T , t ) n ( t ) − n ( t ) 1 ⇔ R k = N k ( T , t ) + m k ( T , t ) m k ( T , t ) � �� � � �� � p 1 , k ( T , t ) p 0 , k ( T , t ) → Use a reference resistor on each PBB! The 10 THMPs foreseen for the fw. endcap are still sufficient. 7
, How to Take the Reference Resistor Into Consideration? Two possible sources have been identified: U I and U off → Determine the impact of each source U I ( t ) m k ( T , t ) = R I , k ( T ) · G k ( T ) · C = m k ( T ) α ( t ) ⇒ R k · m k ( T ) · α ( t ) + n ( t ) = N k ( T , t ) ⇒ N k ( T , t ) − N j ( T , t ) R k m k ( T ) − R j m j ( T ) = α ( t ) → Long-term measurement with known resistors of different resistances for each channel case 1 U I stable: n ( t ) = N ref ( T , t ) − R ref · m k ( T ) case 2 U off stable: R k = N k ( T , t ) − n m ref ( T ) m k ( T ) R ref N ref ( T , t ) − n case 3 Both unstable: Best ansatz with only one reference sensor: � � �� N k ( T , t ) − n (0) m ref ( T ) 1 − n ( t ) − n (0) 1 − R ref m ref ( T ) m k ( T ) R ref = R k N ref ( T , t ) − n (0) N ref R k m k ( T ) 8
, Summary and Outlook ! The resistance of resistors depends on T . The effect has already been minimized, but still must be compensated. A temperature dependent calibration is ongoing. → The THMP temperature must be monitored. A temperature sensor is included in the final THMP design, but possibly one channel per THMP has to be sacrificed. ! Additionally a slow drift has been observed. It is unknown, how long it continues in the same direction. The source of the drift will be determined by a long-term measurement. → The drift can be compensated by using a reference resistor with different methods, depending on the source of the drift. We can sacrifice one, but not two channels per PBB. The resistance should be as close as possible to R ( − 25 ❽ ) 9
Recommend
More recommend
