Successful 21cm detection with flat reflector dipole? Hard to say.

[pipnina]

 I have been wanting to dip into radio astronomy for a while, and now that my sister has finished her electrical engineering degree she is helping me with it (in hopes of adding something interesting to her future job applications).

We have started as small as possible, a dipole tuned for 1420.4 with a flat plane reflector behind it (designed in matlab), some RG58 coax (which wasn't the ideal type but, lessons learned and all that), a RTL-BLOG V4 SDR and Sawbird H1 LNA. Collecting data with my ubuntu laptop and GQRX software.

We actually were able to get into the university communications lab and use one of their VNAs, which suggested our antenna was reasonably well tuned and had a SWR of about 1.4 at our target frequency, not bad!

I wrote a basic python script to process the data in the simplest way possible: calculate the PSD of the recording and show me a graph. However I quickly noticed it would not be simple as the sample aliasing is swamping all the signal, and some of the signal is obvious and loud RFI.

I figured I could correct the former with a topic familiar to me already from astroimaging, and recorded a "dark frame" about an hour long, tuned to my chosen frequency (centered at 1420.6mhz, 3.2mhz sample rate, 2x decimation for effective 1.6mhz sample rate and bandwidth).

I then recorded about 1h 22m of light data, pointed roughly at cygnus (though with this antenna, precise pointing is thankfully not necessary!)

Below we see the PSD of my light recording, my dark recording, and the corrected PSD. I suspect a further correction is necessary as the final recording slopes downward, however I am not sure how to achieve this.

The most interesting thing is the bump at 1420.4 and 1420.8~, which would be the right places for spikes in the hydrogen line in cygnus, however they are still noisy bumps and quite a bit further spread out compared to other people's higher quality data.

So I figured I would post here as a mark of my first steps into the (mine)field that is radio astronomy, perhaps someone here has interesting thoughts about what I have.

It's possible due to the extremely wide field of view, that hydrogen from other areas with different doppler shifts is spreading the peaks out? Or perhaps my calibration method is simply imperfect. (I haven't done anything besides calculate both PSDs, then divide the magnitude of the light by the magnitude of the dark).

Clear skies

James

[Victor Boesen]

Nice start to your radio astronomy adventures!!

It's hard to say, whether or not you have successfully detected the hydrogen line. You can always use this source to check if the profile matches your observation in the corresponding area of the sky:
https://www.astro.uni-bonn.de/hisurvey/euhou/LABprofile/index.php

I don't think it's necessary to sample for more than a couple minutes to achieve a decent result - if anything you're just more at risk to let noise influence the observation during that time.
With an RTL-SDR, H1 LNA and a wifi grid antenna I have had very good success with just a minute or so of observation time.

You mention that you divide the light and dark frame PSD. Since they're already in units of power, I would recommend you to just subtract the two, ie light-dark.
For the results above I perform one observation at the center frequency and another observation with similar parameters 3MHz away from the H-line center frequency. The latter is then subtracted from the on-line observation.
In case it's of any interest, you can find the software I wrote/used for the observation above from this link: https://github.com/byggemandboesen/H-line-software

Kind regards
Victor