280 likes | 365 Views
Thermal Focus and Pointing Corrections. K. Constantikes. Status. New thermal model, new gravity model parameters Uses 19 structural temperature sensors Linear thermal, tradional gravity Focus peformance ~< 3 mm (excludes midday) during ~30 mm thermal focus shift
E N D
Thermal Focus and Pointing Corrections K. Constantikes
Status • New thermal model, new gravity model parameters • Uses 19 structural temperature sensors • Linear thermal, tradional gravity • Focus peformance ~< 3 mm (excludes midday) during ~30 mm thermal focus shift • Elevation performance ~<3” 1s , <1”/hour (excludes midday) during ~ 30” thermal pointing shift • Azimuth performance ~<3” 1s , <1”/hour (excludes midday) • Unanticipated dominance of horizontal feed arm influence • PTCS/PN/25 “Thermally-Neutral Traditional Pointing Models and Thermal Corrections to Pointing and Focus” • Test of real-time focus corrections (Balser and Prestage,11/20/03)
Structural Temperature • 19 locations, 0.2C interchangeable accuracy, 0.01C resolution, 1Hz, range –35 to 40C. (actual accuracy is ~0.1C, temp control of conversion elex) • Design documentation: • PTCS Wiki (AntennaInstrumentation) • PTCS Project Note PTCS/PN12 • Accuracy tested in lab: • Solar/convective loading • Selected unit-to-unit accuracy, repeatability • Electronics temperature range • RFI mitigated, ESD protected • Two thermistor failures, forensics with YSI • Integrated into M&C • First cut pointing, focus predictive algorithms tested
Algorithms • Use existing GBT gravity pointing and focus models • Structure is linear: Thermal effects superpose • Temperature effect on focus, pointing assumed linear in temperatures • No dependence on air or bulk temps, just differences • Simultaneously estimate gravity and temperature model coefs • Estimate coefs using 9/11, 10/2, 11/10 data • Test models using 9/5, 11/20 data
Term Coefficient Min-Max Significance Parameter M1 1.086 13.1 14.3 SR-Pri M2 -0.697 6.2 -4.3 VFA-Pri M3 3.981 15.6 62.0 HFA M4 -7.326 0.9 -6.8 BUS V1 M5 -0.688 12.1 -8.3 BUS V2 M6 -2.576 12.1 -31.2 BUS F M7 -180.630 0.0 0.0 Offset M8 66.189 .7 43.1 sin term M9 196.949 0.6 110.8 cos term Focus Model
Focus Model Tests • Wind < 2.5 m/s • 15° < elevation < 85° • 9/5 is NCP • 11/20 is all-sky • Excludes 1000-1800 • Graphs show thermal contributions only
Term Coefficient Min-Max Significance Parameter M1 -4.6455 1.2 -5.3 BUS M2 1.7830 15.6 -27.8 HFA M3 4.4488 5.9 26.4 VFA M4 -8.4477 1.6 -14.0 Alidade M5 62.2218 0.0 +0.000 -IE,d(0,0) M6 -55.8624 0.7 -62.792 HZCZ,b(0,1) M7 -22.8268 0.9 -38.216 HZSZ,d(0,1) M8 2.4960 2.0 +2.169 -AW,c(1,0) M9 -1.3360 2.0 -1.750 AN,d(1,0) Elevation Model
Elevation Model Estimation s = 3.6
Term Coefficient Min-Max Significance Parameter M1 5.5862 4.0 22.4 Alidade M2 -8.0331 2.7 21.3 HFA M3 -1.6289 2.4 3.8 BUS M4 1.3683 2.0 2.8 VFA M5 3.4124 0.0 0.0 CA, d(0,0) M6 1.3223 0.7 1.0 NPAE, b(0,1) M7 3.5152 0.9 3.0 IA, d(0,1) M8 -2.4960 1.9 4.8 AW, b(1,1) M9 -1.3360 1.8 2.5 AN, a(1,1) Azimuth Model
Azimuth Model Estimation s = 3.9
Why does it work? • Didn’t for 140’ (von Hoerner), why should GBT? • Thermal diffusivity? • Time constants? • Characteristic length of perturbations? • Surface area of structural supports? • Better temperature measurement technology ! • Better homology?
Conclusions • Focus and elevation greatly improved with thermally-neutral traditional model and temperature corrections • Azimuth performance improvement marginal (but it’s already pretty good) • Use of thermal imaging to improve locations • Add sensors to HFA, BUS • Work on graceful degradations • Production implementation • Further tests for confidence • Thermal stability model
Temperature Sensor Locations TF1 TF5 TSR TF3 TF4 TF2 TH3 TB2 TH2 TE2 TB1 TB3 TB4 TB5 TE1 TA4 TA2 TA3 TA1
Focus Model • Gravity • SR-Primary • VFA-Primary • HFA • BUS
Elevation Model • Gravity • BUS • HFA • VFA • Alidade
Azimuth Model • Gravity • Alidade • HFA • BUS • VFA
Optimization • Focus optimization using pseudo-inverse for LSE solution • Coupled Az and El gravity models (AN, AW constraint) • Gradient descent