First K-band Focal Plane Array Pixel Testing
The first step in developing the full 61 pixel array is prototyping the first pixel hardware. The development of the first pixel required extensive design and construction expertise by the NRAO engineering groups. This web page describes the tests leading to completion of the first pixel verification.
This page is organized in inverse time order, with the latest results at the top of the page and preceding steps following. Major milestones in inverse order:
- Band scan search IF spurs, made during GBT Maintenance time - Dec 30, 2008
- GBT Tests of second Down converter unit - Dec 28, 2008
- GBT Comparison of 4 Pixel (old) performance with Single Pixel - Oct 12, 2008
- GBT Tests of Feed in 3 locations of the array - October 6, 2008
- IF Spur investigations - Return FE to Lab - September 18, 2008
- First Astronomical Tests - September 11, 2008
- First Laboratory Cold Tests - August 28, 2008
GBT Maintenance day check of K band 18 to 27 GHz - December 30, 2008
The K1 feed was available during GBT maintenance time and we performed a band scan searching for
IF spurs in the spectra. We were able to take data over the whole frequency range, but later
found that the IF Rack was set to narrow frequency range, which excluded the required data.
No IF spurs were detected, but only extremely strong IF spurs would have been seen.
GBT Test of second down-converter Unit - December 28, 2008
The single pixel (K1 feed) and receiver front end system was installed on the GBT for an
evening of observations. These tests were performed with the second down converter unit, which
only has a single stage of downconverter. Previous lab tests had shown the converter did
not produce strong IF spurs. The goal of the tests were to confirm good system temperature
over the observing band, by comparison of observations between the K0 (current GBT) receiver
and the K1 receiver.
GBT 4 beam Rx Comparison - October 12, 2008
On October 12, 2008, Glen Langston performed a sequence of observations of 3C48 with the 4 Beam (Current) and Single pixel receiver. For these observations we performed Peak observations on 3C48 with both systems, in order to compare the system temperatures.
LO equations for centering a Sky Frequency in the Spectrometer Band center.
Three LOs are adjusted in order to bring a particular sky frequency into the center of the spectrometer bands. The First two LOs, LO1A
, are located in the receiver room at the GBT. LO1A
is used to set the center of the band within the band pass filter at 6.8 GHz. LO1B
is always set to a fixed frequency 8.9 GHz with +8 dBm output power. LO2 is located in the GBT Equipment room and is used to select a band for input to the spectrometer samplers.
The Spectrometer has four different observing bandwidths, 800, 200, 50 and 12.5 MHz.
The equation for the KFPA is
KFPASky[LO1A_, LO1B , LO2
, LO3_] := LO2 - LO1B
+ 2 LO1A
- 10.5 + 1.2 [sky frequency in GHz]
For the usual fixed value of LO1B
of 8.9 GHz, and equations below for the different spectrometer bandwidths:
- 800, LO2=11.4 GHz: SkyFreqGHz = 2*LO1A - 6.8
- 200, LO2=11.1 GHz: SkyFreqGHz = 2*LO1A - 6.8
- 50, LO2=11.4 GHz:
Feed Gain Tests at Different Array Location
On October 6, 2008, Steve White, Bob Simon and Glen Langston performed GBT observations using the first pixel. At the start of the tests, the first pixel was located at the center location of the front end. A series of measurements of the 3C48 brightness were made. Next the feed was moved to an azimuth offset location, approximately 3.7 arcminutes from the center. We repeated the 3C48 measurements. Next the front end was rotated so that the feed was located 3.8 arcminutes higher in elevation, relative to the center pixel location.
The measurements indicate there is a slight reduction of gain for the outer feed locations, but the reduction with offset is approximately what was expected based on antenna models.
- We report the details of the tests in a separate page.
IF Spur investigations
The First pixel tests revealed a number of IF spurs that are images of the LOs in the down converter chain. While the GBT was stowed for maintenance, Steve W. and Glen L. ran a number of scans, with a variety of combinations of LO1, LO2 and LO3.
- The spectra are were performed using Astrid and the data are in project T_KFPA_08SEP18
First Telescope Tests
The first pixel hardware was built in 2008 and put on the antenna on September 9, 2008. The first astronomical tests were from 1:00 AM to 9:00 AM on September 11, 2008. In general, these tests were successful, and pointed out additional efforts that are required to completely validate the first pixel hardware. Below is a list of tests/checks that were performed, with notes on results.
The First Pixel Dewar was placed in the C-band slot in the GBT turret. Since the C-band Front end is a single pixel front end, the GBT observing system was configured, using ASTRID as though the C-band Rx was still in place, then CLEO was used to update the signal path connections to route the signals to the back-ends.
There were extensive tests of the First Pixel hardware, required to set the proper voltages, gains and temperatures for proper functioning of the Dewar and electronics. A web page was created to allow monitoring of the hardware (See 172.23.1.79/index.html, only visible from an NRAO network)
These tests indicated that there was insufficient gain in one of the two polarizations. The cause of the low power was unknown at the time of the tests. The remaining results are based on use of a single good channel.
The weather was overcast, but the system temperature were reasonable, ranging between 50 and 100 K, in agreement with the CLEO weather forecast for this date.
The data were taken as project T_KFPA_08SEP11.
- Scans 1-70: Peak and focus observations of 3C123, 3C48 and 0501-019 at a variety of center frequencies.
- Scans 71-139: Spectral line observations of Orion A. All observations were with 800 MHz bandwidth
- Scans 140-145: Peak and Focus on 0410+76
- Scans 145-152: Spectral line observations of W3(OH) to observe a very strong water maser line
- Scans 153-246: Baseline stability tests. Observations were toward TMC-1, with 800 MHz bandwidth
- Spectrometer base line stability curves kfpastab155.ps. One minute scans, 10 second integrations: Differencing scans 155 to 210.
- Spectrometer base line stability curves kFPAstab246.ps. One minute scans, 10 second integrations: Differencing scans 210 to 246.
All ASTRID scripts used to run the tests are in the directory /home/astro-util/projects/TKPFA
Peak and Focus
The peak and focus tests were relatively smooth. The Astrid software worked well and, in fact, every peak and focus measurement was successful. Below are two example peak and focus measurements. The system temperatures were calculated using a T-Cal value of 1, so all results should be scaled by the actual cal values.
| || |
|Peak with First Pixel||Focus with First Pixel|
We stepped across the frequency band 18.0 to 26.5 GHz in 1 GHz steps, making peak and focus observations of 3C48. We plan to use the known brightness of 3C48 to compute system temperature vs frequency.
Spectral Line observations: W3 (OH) and Orion A
Previous to these astronomical tests, it was found that a number of IF spurs were being introduced into the band pass, due to the down conversion unit. Modifications to the down conversion unit are planned. The plot below shows a band scan of Orion A, taking one minute on source followed by one minute off source observations. Several spectral lines in this band are due to IF spurs, except the water line at 22.235 GHz.
| Strong (greater than 4000 K) Water line in W3 OH
|| Band Scan of Orion, showing water line, hydrogen recombination lines and IF spurs
A median filter baseline has been subtracted from the spectra, and the end channels have been flagged. The width of the spectral baseline shows the noise in the band. The noise is strongest in the center of the band due to the higher system temperature in this region.
Note that the FITS files contain frequencies corresponding to the C-band LO equation. In the observing configuration we used. (IF center frequency 3 GHz, LO2 Frequency 8.9 GHz, Converter rack frequency 11.4 GHz) the equation converting C band sky frequency to K band sky frequency is: Freq_K = (2 * Freq_C) - 2.6 GHz
Examination of a Single IF spur at 22300 GHz.
The locations of image lines will be important for the re-design of the down-converter unit. Below, the raw and (Signal-Reference)/Reference calibrated spectra are shown.
| Intensity versus frequency plot for zoom in region near the water line. There is an IF spur in the center of the band. The upper curve shows the on Orion Signal. The lower curve shows the Off Orion signal. Note that there is clearly signal at the location of water line and the H alpha,66 recombination line on the other side of the IF spur.
|| Calibrated, (Signal - Reference)/Reference, spectrum of Orion. Note the detection of the water line, the IF spur and Hydrogen recombination. The other lines marked in the plot are too weak to be detected in a 1 minute integration.
List of Sky frequencies for water line and IF spurs, based on topocentric measurements
- H2O: 22234.764 +/- 0.016 MHz (delta between line and spur: -64.98 MHz)
- Spur: 22299.734 +/- 0.008 MHz
- H_alpha_66: 22364.466 +/- 0.015 MHz (delta between line and spur: 64.72 MHz)
The first telescope tests used no "K-band" specific software, but rather pretended the "C-band" front end was in place. This worked reasonably well, but our observing efficiency could be improved by more specific software support. Items include:
- K-band single pixel manager, including LO1 setup. Implementation of the IF configuration
- K-band support for auto-peak focus
- K-band support for Doppler tracking.
- The K-band LO equation has two LO's, LO1 and LO2.
- LO1 is Doppler tracking LO and is multiplied by 2 in hardware
- The IF band pass center filter is at 6.9 GHz
- LO2 is at a fixed frequency 8.9 GHz and downconverts the IF to 2 GHz.
- For our tests at 800 MHz, the converter rack LO was at 11.4 GHz.
Table of tests completed during the telescope test.
- Signal checkout (partial, low signal level in one polarization)
- Step across band with spectrometer 800 MHz mode, towards astronomical source.
- On/Off with spectrometer across band
- System temperature across band (Partial, only one polarization)
- Frequency switched line obs across band; too many IF Spurs.
Installation Tests, September 9, 2008
- Passband shape: ripple and slope across the band
- Baseline structure (Spectrometer)
- Baseline stability (Spectrometer)
- Total power stability (DCR)
- Noise calibration checks (Partial, observations at selected frequencies)
- Efficiency at specific focal plane sites.
- Center Pixel
- ~ 3 HPBW at 4 quadrants.
A large number of laboratory tests were performed before putting the receiver on the antenna. These include:
- Issues(?) noted during lab testing (RDN 04Sep08):
- LO1 +/- 3,4 X LO2 mixer spurs in IF band up to 30dB above TP in 3MHz RBW.
- Ripple and slope in IF output total power: Ripple ~5dB p-p; slope ~-5dB 1.2-2.8 GHz.
- IF rolloff 2.8-3.0 GHz.
- Noise increase at high frequency end. Need to understand cause at least.
- LO1 frequency range. Desirable to tune 18 & 26.5 GHz to center IF?
- Series resistance high in noise module driver circuit.
- Multilevel Tcals? Is this a requirement?