VLBA Electronics Memo No. 737

Transcription

1 VLBA Electrnics Mem N. 737 U S I N G PULSECAL A M P L I T U D E S TO D E T E R M I N E SYSTEM T E M P E R A T U R E D.S.Bagri 1993Mar05 INTRODUCTION System temperature is nrmally measured using mdulated nise dide signal injected at the input f a receiver and synchrneusly detecting the nise pwer at the backend in the BBCs. The injected nise dide signal is calibrated in the labratry r in the field using ht and cld lads at the receiver input r at the feed input. Generally the injected nise pwer is abut 510% f the receiver nise pwer and is assumed cnstant with time. Als it is assumed that the spectrum f the calibratin nise is basically same as that f the system nise -which is nrmally assumed flat with frequency (ver the passband f the receiver backend). The integratin time cnstant fr the ttal pwer and the synchrneus detectrs is f the rder f a few secnds. Thus fr a bandwidth f 1 MHz at the backend f the receiver we expect the estimated system temperature t be accurate t within a few percent. The estimates are affected by linearity f the amplifiers and detectrs used, ffsets in the detectrs, and quantizatin and such ther errrs in the mnitring system. The quantizatin errrs fr switched pwer (synchrneus detectr utput) becmes imprtant because its effect n the system temperature estimatin is amplified by the rati f the ttal pwer t the switched pwer. By increasing the injected nise pwer and/r integratin time it may be pssible t imprve the estimatin accuracy. Fr reasns f decreasing the system sensitivity it is nt desirable t increase the injected nise pwer. Increasing the integratin may help smewhat but has prblems - e.g. at lw elevatins. Als it may be pssible t measure the detectr ffsets etc. and apply apprpriate crrectins while calculating the system temperature values using ttal pwer and synchrneus detectr values but it is smewhat cumbersme. Further the nise surce shuld nt vary frm the time it is calibrated. Fr the VLA the relative accuracy f the system temperature estimatin f a few percent is cnsidered adequate and several percent is prbably acceptable. In mst VLA bservatins abslute accuracy is nt imprtant as it is calibrated ut using calibratin surces. Hwever in the VLBA it is smewhat different and it appears that it will be desirable t achieve an accuracy f 1-2% fr the system temperature measurements. In view f the abve it is desirable t examine whether amplitude f the pulse calibratin detectrs can be used fr estimating the system temperature, and what are its merits and limitatins in cmperisin with thse f the nise calibratin system presently used. GENERATION Pulse cal - uses tunnel dide driven by utput pulses frm tw back t back cnnected Schttkey dides, cnnected t the grund, and driven frm utput f an amplifier which is heavily saturated. Essentially the utput pulse amplitude depends n the characteristics f the tunnel dide, and its rise/fall time depends n the dide bulk resistance and capacitances, and ther stray capacitances in the circuit. We knw that the delay/phase f the utput cmb frequency signal is very stable (needed fr phase calibratin) and ne wuld expect that t be

2 mre sensitive t any circuit parameter variatins. Therefre I wuld guess that amplitude f the utput cmb is very stable. Fr sme narrw band peratins cmb frequency signal may nt be available in backend (BBC) passband. Nise cal - uses zenner dide in avalanche mde. Output nise pwer depends n the current thrugh the dide. Other aspects like cnsideratins fr the nise spectrum dependence etc. shuld be similar t thse fr the pulsecal. The current thrugh the dide depends n the drive circuit, and presently it is driven frm a 28 Vlt pwer supply switched thrugh a transistr, which is a crude arrangement fr this purpse. We shuld change this t a cnstant current surce rather than a vltage surce. INJECTION Bth pulse cal and nise cal shuld have similar prblems, except that nise cal is cheaper t build and therefre we have ne fr each receiver. Therefre its frequency ripples may be smewhat smaller than thse f the pulse cal cmb. But since pulse cal has been installed primarily fr phase stability measurements, which axe mre sensitive t passband ripples and their variatins, we have t ensure that passband ripples are nt a prblem. This distributin scheme has nt been bjected fr the phase calibratin purpses, and therefre fr present I will cnsider that this is nt a surce f cncern. DETECTION Pulse cal - uses cherent detectin f narrw band signals frm digitized data. The pulsecal amplitude is nrmalised using ttal pwer in the baseband signal. Nn flat bandpass will therefre cause errrs in the estimated pulsecal amplitude, and if the passband ripple arund the pulsecal tne frequency changes, then the pulsecal amplitude will be affected even withut the system temperature changing. Crrectin fr this can be estimated by measuring the autcrrelatin pwer in the frequency channels arund the tne frequency, and therefre it is easier t accunt fr it in a FX crrelatr. The pulsecal amplitude gives high signal/nise rati fr very small added signal. With limited spectral dynamic range f the crrelatr and limited linearity f the analg signal path, presence f narrw band signals like pulsecal tnes may nt be desirable during sme spectral line bservatins. Nise cal - uses syncrneus detectin f bradband signal pwer level during tw halves f the nise cal switching cycle. Change in pwer level during tw halves f a cycle will aifect the FFT f the FX crrelatr which will be reduced by fringe frequency but still may have sme residual effects (due t cmmn peridicity f 100 msec fr FFT and cal switching). Linearity f amplifier and detectrs, and ffsets in the detectrs etc. need t be accunted prperly t estimate the system temperature value accurately. CONCLUSION I have used pulse cal amplitudes in cnjuctin with the nise cal and cmpared the tw sets f data. Fig. 1 shws the pulse cal amplitudes fr varius BBC channels during a S/X phase stability test at Ls Alams with the antenna stwed. Fig. 2 shws system temperature values calculated using ttal pwer and synchrneus detectrs frm the mnitr data. During this perid the 4 cm receiver was cling. The variatins in the pulse calibratin amplitudes (rather inverse f square f the amplitudes) seem t be tracking the system temperature values well. It seems that use f the pulse calibratin amplitudes fr estimating the system temperature is

3 prmising. It will have t be calibrated using ht and cld lads in the field (fr present). It seems wrthwhile t try this methd ut t learn its limitatins and strengths.

4 LA 19bFEB01 - <D "O D CO h Q. < O. CL ^ CN h 4 Cr* J i i ' Time (days 19^FEB ) Fib. 1

5 ifc*\ hdk LA 19WFEB01 l 1 1 r en <l> "U CO >> m l-c. \-6 4 tra Time (days 1993FEB )