2016 October 13 IVTW - Haystack 1

VGOS broadband delay extraction, analysis,

and results

Arthur NiellMIT Haystack Observatory

2016 October 13 IVTW - Haystack 2

Overview

• What is broadband delay and why?• Dual pol’n broadband delay observable

– Delay and phase calibration needed• across bands• between polarizations

• Geodetic analysis– Cable delay to phase cal generator– Results for VGOS Demonstration Series

• GGAO12M-Westford: 2014 Dec – 2016 Feb

2016 October 13 IVTW - Haystack 3

Broadband delay

• Frequency coverage– Four bands of 1 GHz (ideally; 0.5 GHz for now)– Dual linear polarizations– Mean frequencies (GHz) : 3.3 5.5 6.6 10.5

• Delay precision– VLBI2010 WG3 set a goal of 4 psec– Intention was to make it smaller than other errors

• Delay from pseudo-Stokes II = (XX + YY)*cos(Δ) +(XY –YX) *sin(Δ)

where Δ = difference of parallactic angles

2016 October 13 IVTW - Haystack 4

Broadband frequencies

2016 October 13 IVTW - Haystack 5

Broadband signal chain

2016 October 13 IVTW - Haystack 6

Extraction of delay and phase observables

• Calibration along the bands– Essential to have coherence in each polarization product

separately: XX, YY, XY, YX– Obtain delays and phases from multitone phase cal

• Calibration between polarizations– Must include delay and phase differences between

polarizations in order to combine them coherently– Determine Y-X offsets using a strong source (fourphase)

• Estimate delay and dispersion simultaneously (fourfit)– Correlation increases delay uncertainty by a factor of 2.6

2016 October 13 IVTW - Haystack 7

QRFH feed

couplers

phase calgenerator

5MHz frequency reference

Polarization delay difference

2016 October 13 IVTW - Haystack 8

IXY fourfit

dTEC

Phase per frequencyRMS 5.8°3.0-10.7 GHz

amp

2016 October 13 IVTW - Haystack 9

Delay uncertainty distribution

2016 October 13 IVTW - Haystack 10

Cable delay measurement proxy

• Need to know the delay on the cable carrying the reference frequency (5 MHz) to phasecal– az-el dependence will bias position estimate

• No cable delay for GGAO12M until CDMS is built• Use a proxy for cable delay when it is missing

– Measure dependence of multi-tone phasecal delay on azimuth and elevation in short time

– Use dependence as the GGAO12M cable delay proxy• Reduces GGAO12M-Westford length WRMS • Reduces RMS post-fit delay residuals

2016 October 13 IVTW - Haystack 11

Multitone phasecal delay

2016 October 13 IVTW - Haystack 12

VDS Sessions

• Fifteen months of GGAO12M – Westford baseline – 2014 December through 2016 February

• 13 sessions– All bands worked– Receivers cold– Durations:

• 10 sessions 1-1.5 hours• 2 sessions 4 - 5 hours• 1 session 24 hours

2016 October 13 IVTW - Haystack 13

Geodetic analysis

• Databases are in the new vgosDb format• Geodetic estimation done with nuSolve

– Approximately 45 scans per hour– 30 seconds per scan– Separate analysis for each session

• All sessions analyzed with ‘standard’ PWL parameters– Clock: 15 min; zwd: 20 minutes; gradients: 1 hour– Identical processing for all sessions

• Typical WRMS residuals 10 psec before re-weighting• Additive noise for re-weighting 10-30 psec

– Less than 1% deleted as >5*reweighted delay uncertainty

2016 October 13 IVTW - Haystack 15

Other broadband sessions

• VGOS– Seven sessions that include KOKEE12M with

GGAO12M and WESTFORD– Five sessions to debug the broadband capability at

Ishioka, Wettzell-South, and YEBES13m in fringe-tests with GGAO12M, WESTFORD, and/or KOKEE12M

2016 October 13 IVTW - Haystack 16

2003 2016

2016 October 13 IVTW - Haystack 17

AcknowledgementsThe following people contributed significantly to making the results reported here possible.

Roger CappalloBrian CoreyMike TitusJohn GipsonSergei BolotinEd HimwichAlex BurnsMike PoirierKatie PazamikasJay RedmondChris BeaudoinChris EckertMark DeromeJon Byford

(I apologize to anyone I left out.)

2016 October 13 IVTW - Haystack 18

Rest of talk

2016 October 13 IVTW - Haystack 19

SNR distribution

2016 October 13 IVTW - Haystack 20

Other sessions

• VGOS broadband– Seven sessions that include GGAO12M, WESTFORD,

and KOKEE12M– Five sessions to debug the broadband capability at

Ishioka, Wettzell-south, and YEBES13m in fringe-tests with GGAO12M, WESTFORD, and/or KOKEE12M

• S/X– Four 24-hour tie sessions for KOKEE12M to

KOKEE20M (S/X)

2016 October 13 IVTW - Haystack 21

Broadband correlation and post-correlation

• Correlate all four bands simultaneously.– directly from the Mark6 (last six sessions)– all polarization products within each band for all

four bands• fourfit

– Apply multitone phasecal delays and phases to align the phases of the four bands.

– Correct for uncalibrated delay and phase offsets between polarizations (use one strong source for values).

– Coherently estimate group delay and dTEC(ionosphere).

2016 October 13 IVTW - Haystack 22

Geodetic analysis• Create vgosDb database (MikeT).• Create proxy cable delay values based on

the scheduled azimuth.• Calculate baseline length.

– Use same standard parameterization for the clocks and atmosphere delay for all sessions.

– Typical re-weighting 10 psec of additive delay • group delay uncertainty a few psec

2016 October 13 IVTW - Haystack 23

Backup/additional slides

2016 October 13 IVTW - Haystack 24

Geodetic analysis

• VGOS Data Series– Median delay uncertainty per scan is a few psec.– With re-weighting the WRMS post-fit delay residual

is typically 10-15 psec (compared to a few times 10 psec for current S/X sessions).

– Extra noise likely due to both real variations and to simplified atmosphere and clock models; possibly also to missing cable delay measurement.