ETA: Dipole / Active Balun Testing @ PARI: Oct 15, 2005

Overview: Shown here are the results of testing of various combinations dipole and active balun choices in situ at the PARI ETA site.

Maintainer: Steve Ellingson (Virginia Tech)
This page is http://www.ece.vt.edu/swe/eta/AB_051015/
Updates:
17 Oct 2005 UT 1100. Initial posting.

Introduction: This testing follows up the first round of testing done near Blacksburg using the copper wire dipole and active dipole #1. That testing made it apparent that I had goofed the match design in active balun #1. But, the testing also revealed that intermod generation was not as bad as I had feared, so the fancy match (which also did bandpass filtering) seemed less important. So, for this test, I built "active balun #2", which is identical to #1 but has no input match (i.e., only a 1000 pF DC-blocking cap), with good results as documented here. Other issues addressed include the following: (1) effect of adding a small ground screen (small improvement in Galactic S/N), (2) effect of replacing copper wire with aluminum strip (moderate improvement in bandwidth and small improvement in Galactic S/N). Finally, this testing also takes place at stand 1 of the ETA site, so the RFI seen is the RFI ETA must contend with. The local time was 1:30-2:45 PM, and Galactic Center was very low on the horizon, probably blocked by the surrounding terrain for the duration of the measurements.

Summary of Findings:

  1. Active balun #2 (no attempt to match; essentially the original Hicks design) works fine, significantly increases bandwidth, and appears to do OK in the RFI environment of PARI.
  2. We have a nasty reflection on the long coax that needs to be quenched. This should be easy to do.
  3. The copper wire dipole antenna gives us 10 dB Galactic noise dominance in 33-43 MHz, whereas the 3/4-in wide aluminum strip dipole gives it to us in 32-45 MHz. For 6 dB Galactic noise dominance, it is 28-50 MHz and 27-53 MHz respectively. These numbers are conservative. Based on this, either approach would meet requirements for ETA.
  4. For reasons not clear, the aluminum strip dipole also seems to yield an improvement of about 0.18 dB (4.4%) improvement in Galactic noise dominance near resonance. Probably not worth pursuing solely for this advantage, but interesting nevertheless.
  5. The addition of a small ground screen (barely larger than the footprint of the dipole over it) provides 0.35 dB (8.5%) improvement. It may be worthwhile (but not necessary) to pursue a larger ground screen.

Test Instrumentation: The active balun output was connected to a 150-ft section of LMR-400. At the other end, the signal was pre-amplified using a Mini-Circuits ZJL-3G amplifier, and then input to a Rhode & Schwartz FSH3 handheld spectrum analyzer. The spectrum analyzer is connected to a laptop via optical RS232, and data is acquired using a C-language program written by the author. Each plot below represents the linear average of at least 100 sweeps with the spectrum analyzer set to "sample" (as opposed to "autopeak") detection to facilitate meaningful averaging. Results shown are the levels at the input of the spectrum analyzer; to get the levels referenced to the active balun input, subtract about 45 dB.
Wire Dipole: Shown here is the wire dipole arrangement set up at the center of the future location of the core array. 1.5 mm-dia stranded copper antenna wire is used for the dipole arms. The active balun is mounted on a copper pipe and the dipole arms are formed by simply staking them as shown.
Wire Dipole + Active Balun #1. So, this should be more or less the same configuration tested on October 11 near Blacksburg, with different RFI. As before, the hump centered on 38 MHz is the Galaxy, showing reasonable agreement with the "predicted" curve. Also as before, the prediction is obtained from Cane's empirical expression for the intensity of the Galactic background radiation, and is modified to account for the theoretical antenna response and measurements of the active balun and instrumentation responses. And, of course, there is RFI from the usual sources below 20 MHz and above 50 MHz. At PARI the local TV Ch 3 (60-66 MHz) is a relatively weak NTSC station (as opposed to the monster HDTV signal that clobbers Blackburg), and the main linearity challenges come from Ch 4 (around 70 MHz) and broadcast FM.
Wire Dipole + Active Balun #2. Same as above, just replacing the active balun with the new version. The bandwidth appears to have expanded, as expected. However, we also see some ripple with a null-to-null spacing of about 2.5 MHz. I believe this from reflection on the long coax, because a bad reflection on a 150-ft section of coax should theoretically produce an interference pattern with about this spacing. It is probably present for Active Balun #1 but not apparent due to the smaller overall bandwidth. This is probably easily mitigated with some attenuation on one or both ends of the cable; we will have gain to spare for this.
Wire Dipole, 100 kHz resolution. Same as above, in a narrower spectral view for comparison with subsequent plots. Using active balun #2 for the remainder of this testing.
Adding a Small Ground Screen. All we did here is add a 36-in by 84-in section of "bright aluminum" screen door material under the wire dipole, as shown.
Wire Dipole + Ground Screen. Should be the same as the previous spectra, except for the effect of the ground screen. Hard to see any improvement. Using measurements at 3 kHz resolution around 37.9 kHz (not shown here) and a lot of averaging, I was able to convince myself that the improvement in Galactic S/N is about 0.35 dB (8.5%) relative to no ground screen.
Aluminum Strip Dipole. At this point, we replaced the copper wire dipole arms with aluminum strip, 3/4-in wide by 1/8-in thick, as shown here.
Strip Dipole + Ground Screen. Should be the same as the previous spectra, except for the effect of the new dipole arms. It is apparent that the bandwidth is increased, but the ripple makes it hard to quantify using the measurements along. Conservative estimates are provided in the "Summary of Findings" above. Interestingly, the Galactic S/N also seems to improve: Again using measurements at 3 kHz resolution around 37.9 kHz, and a lot of averaging, I was able to convince myself that the improvement in Galactic S/N is about 0.18 dB (4.4%) relative to the wire dipole. In other words, this "deluxe" configuration seems to be at least 0.5 dB (13%) better than the wire dipole over no ground screen, at resonance.