First-time radio astronomer, looking for advice

[Robert Hamers]

I'm a total newbie at radio astronomy but have been doing astrophotography for years and have a strong background in electronics and some prior experience in radio and programming. I built a modest-size rectangular horn (~ 1 m x 0.8 m top dimension) and am using the Nooelec H1 Sawbird, Airspy R2, and a headless Raspberry Pi 5 (which I connect to via TigerVNC).   For software I'm using the gnuradio-companion, DSIPRA software (spectrometer_w_cal.grc) with the gr-osmocom radio block. I've gotten some decent first results (below) looking at the MW cutting through the south meridian, AM and PM.  (spectra are 2-minute averages) I'm pretty pleased with this as initial results, but have some issues and looking for help. 

1) I'm getting a lot of "spikes" noise in the spectra, which you can see below. The individual frequencies seem to be quite reproducible but intensities vary. I'm not sure if this is an issue in the electronics or what to do to get rid of it.  

2) Using the Gnuradio-companion and gr-osmocom radio block,  setting the RF, IF, and BB gain seems to be very irreproducible...sometimes it seems the gain "takes" and sometimes it doesn't.   I'm wondering if there is better/different software out there compatible with the Airspy R2 , ideally compatible with RPI/Raspbian, but generally open to all kinds of suggestions. 

 

[fasleitung3]

Hi Robert,

You are getting pretty good results, congratulations.

Your spikes in the spectrum are most likely caused by some interfering device in the vicinity. A typical offender are monitors. You could try relocating the monitor or if possible turn it off when taking data to try that. If that is not the cause, than look at other devices you have around, including the Pi. Keeping digital electronics away from the antenna as far as possible is a good thing to do.

As far as I know, spectrometer_w_cal.grc is an older version. You could try their newer version gr-radio_astro, https://github.com/WVURAIL/gr-radio_astro. Maybe that cures your problem.

Other than that, you may want to look at some other software which supports the Airspy and will most likely run on a Pi 5:

Virgo: https://github.com/0xCoto/Virgo

RadioPy: https://github.com/byggemandboesen/RadioPy

Spectro_radiometer https://github.com/ccera-astro/spectro_radiometer

You may also want to check whether the parameters for the gain you are using are all in the permissible range. I could imagine that if that is not the case, the setting may just be ignored.

Best regarsd,

Wolfgang

--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/sara-list/59342166-2654-484d-a1c4-07b60c724ae8n%40googlegroups.com.

[Robert Hamers]

Thank you, Wolfgang- this is all quite helpful !

best wishes,

Bob Hamers

[Alex P]

Hello Robert,

Here is info on software easy to use and from which you can obtain good results ..

( It does require MS Win 7,10,11 )

The processing sequence is  AirSpy SDR# > IFavg > HLine3D > Rinearn (graphics )

https://github.com/AP-HLine-3D/HLine3D

Regards,

Alex Pettit

[b alex pettit jr]

The processing sequence is  AirSpy SDR# > IFavg > HLine3D > Rinearn (graphics )

https://github.com/AP-HLine-3D/HLine3D

Regards,

Alex Pettit

[Ayushman Tripathi]

Hi Robert,

I'm also running a fully remote & automated system using a Raspberry Pi 500. I wrote a script, later converted to a systemd service, which runs continuous observation loops, each loop is 24 hours of 5-minute integrations. It uses VIRGO software and an Airspy SDR. The system automatically records both raw and calibrated data on the Pi.

On my main PC, a second script handles everything else, reprocessing with VIRGO, baseline correction, and plotting. It supports multiple plot types, including 3D VLSR-corrected stacked plots, etc.

The whole pipeline is largely automatic once running. It's not perfect and it has bugs for sure, which is why I haven't published it on GitHub. But if you're interested, I'm happy to share both scripts on GitHub public repo.

The attached plots are from this same pipeline.

Thanks

[Robert Hamers]

Thanks for the helpful comments !  I've tried the IFAvg plug-in for SDR# and that is definitely an improvement over what I was using previously.  I"m still open to any additional suggesions and would like to try the software from Ayushman as well. 

[b alex pettit jr]

Robert, 

If you have SDR# & IFavg installed and can save *.txt files, HL3D is a near trivial operation to remove offset drift and

calibrate the HLine data files in dB normalized to Cold Sky. There are stepbystep instructions for using Rinearn to create some nice plota.

HL3D

Find the Data Files load & click Ok

Hit  PROCESS.

The results are written into  data subdirectory 

Load them into either Excel or Rinearn for plotting

( Rinearn 2D )

Raw Data

Drift  ( Sun drifted into side lobe of antenna @ ~19-23 hrs  )

Corrected, Normalized,  dB Scaled Data 2D

above data w/ Rinearn 3D

Alex Pettit

GitHub - AP-HLine-3D/HLine3D: Beginners Low Cost 21cm Radio Telescope Project b.alex.pettit.jr.

GitHub - AP-HLine-3D/HLine3D: Beginners Low Cost 21cm Radio Telescope Pr... Beginners Low Cost 21cm Radio Telescope Project b.alex.pettit.jr. - AP-HLine-3D/HLine3D

[b alex pettit jr]

Another plug for  SDR# > IFavg > HL3D : The data is Calibrated and Scaled in engineering units.

Meaning :  You can compare changes in your hardware setup and compare the performance of your system against others.

I have data sets from the US & Europe on systems from simple 1m antennas to a 

          12m radio telescope in Neu Golm, Germany.

Alex Pettit

========================================

On Wednesday, March 4, 2026 at 01:17:57 PM EST, Robert Hamers <rjha...@gmail.com> wrote:

[b alex pettit jr]

Robert, 

If you have or can record a set of IFavg data files, send them to me.

I'll Process them via HL3D & Plot w/ Rinearn .

Alex

On Wednesday, March 4, 2026 at 01:17:57 PM EST, Robert Hamers <rjha...@gmail.com> wrote:

[Robert Hamers]

Hi-

Thanks for the offer! I have rather limited visibility from my home so I”ve been just doing short spectra (~ 2 min) as the MW crosses the south meridian and different elevations in the evening and then again in the morning.  So not enough to make a 2D map.  There is another location I can move to that will give me broader visibility but I want to have everything else optimized before doing that.  But that’s my goal!  

So far the IFAvg plugin is working well.  It sometimes crashes inexplicably when I choose 1024 or 2048 FFT length,  but if it doesn’t crash immediately it seems to be stable afterwards… I’ll probably try another set of meridian-crossing data files later this evening and tomorrow, weather permitting…

Thanks again for your help and offer!

Get Outlook for Mac

--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/sara-list/1151188160.3407657.1772653852181%40mail.yahoo.com.

[b alex pettit jr]

" Thanks for the offer! I have rather limited visibility from my home so I”ve been just doing short spectra (~ 2 min) as the MW crosses the south meridian and different elevations "

That's how a drift scan it done : Set the dish pointed along your Meridian at some Elevation & Let the Sky Drift By .. for 24 hrs or so

Earth's Rotation provides the RA axis ..

I typically acquire 5 min samples w/ 512 FFT freqs.

Watch this perhaps ?

Alex Pettit: Galactic Hydrogen 1.42 GHz RF Emission Radio Astronomy for $300

Alex Pettit: Galactic Hydrogen 1.42 GHz RF Emission Radio Astronomy for ...

Alex

[b alex pettit jr]

This is a good chart to use for finding bright areas of the MW H Line

( Dec+20 & Dec+40 are the best if you can see them with your antenna )

Set up your system for 2min or 4min samples & let it run for 24 hrs ..   Send me the 360 or 720 set of spectra

( A pre-test IFavg background correction will remove the gain vs freq curve of the LNA/SDR )

Alex

[Ayushman Tripathi]

Hi Robert,

I uploaded both scripts to GitHub https://github.com/ayushman-t/h-line-observer I'd suggest trying h_quick first to make sure everything works before going automated.

I cleaned up the paths and settings that were specific to my Pi and made them a bit more flexible, there may still be some bugs.

h_quick.py — a simple manual testing tool. Just run it, pick a calibration from the menu (or record a new one), then choose single observation or continuous loop. Good for checking everything works before going fully automated. Both scripts will also generate 5-minute plots using the default VIRGO output which should be fine for a quick test.

h_observer.py — the main automated script. Run it once with --install to set up, then --config to pick your SDR and output folder, record a calibration pointing at cold sky, and then start the systemd service. From that point it runs 24/7 on its own, creates a new folder each day and resumes automatically after a power cut.

I'll upload the pipeline script that handles the combined plotting and analysis of the full 24-hour loops soon.

If you face any issues, I'll be happy to help.

Thanks

[Robert Hamers]

Thanks!

Are there any prerequisites beyond ones I might already have (e.g.,  python 3.9,  numpy,  astropy ) ?

bob

Get Outlook for Mac

--

--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/sara-list/45ed4eec-fd5a-4ca2-9767-39053a36d091n%40googlegroups.com.

[Ayushman Tripathi]

Hi Robert,

For the prerequisites, just GNU Radio and VIRGO. 

Installing VIRGO automatically includes numpy, matplotlib and astropy, so no additional packages needed beyond what you already have.

For installation: https://virgo.readthedocs.io/en/latest/installation.html

Thanks​.

Sent from my iPhone

On 5 Mar 2026, at 20:04, Robert Hamers <rjha...@gmail.com> wrote:



You received this message because you are subscribed to a topic in the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/sara-list/3RC4V2ZxCQI/unsubscribe.
To unsubscribe from this group and all its topics, send an email to sara-list+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/sara-list/PH0PR05MB7590B0BF472586109CB11C98A17DA%40PH0PR05MB7590.namprd05.prod.outlook.com.

[Ayushman Tripathi]

Hi Robert,

I’ll be sharing the automated processing and plotting pipeline for h_observer.py data on GitHub this weekend. It generates multi-hour/multi-day plots with VLSR correction and RFI removal using a median filter, reading directly from the .dat files. I’m also thinking of adding ezRA integration in the same script.

Once the weather clears up, give it a try and if you’d like to improve the script, that would be great!.

Thanks

[b alex pettit jr]

Hello Ayushman,  

Impressive Software Capability !

I've never seen a *.py script . 

I wrote signal analysis software in Matlab for in-house aerospace use.

The original Hline software was developed (5 yrs ago ) in Matlab,  

 and have since painfully adapted these concepts into Java. 

Alex

=================================

[Robert Hamers]

Hi, Ayushman-

Great!   I look forward to seeing it and trying it out !    (60 mph /  100 kph wind gusts here today….definitely no data collection !).

Best wishes,

Bob

Get Outlook for Mac

From: 'b alex pettit jr' via Society of Amateur Radio Astronomers <sara...@googlegroups.com>
Date: Friday, March 13, 2026 at 5:31 AM
To: sara...@googlegroups.com <sara...@googlegroups.com>

--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/sara-list/653566060.468559.1773397895060%40mail.yahoo.com.

[Ayushman Tripathi]

Hi Alex,

Thanks! Your H-Line 3D plots are really good. This is just a Python script I put together to automate the observations on a Raspberry Pi and handle the plotting/analysis, trying to make it easier to run 24/7 unattended.

[Ayushman Tripathi]

Hi Robert,

Hope the wind calms down soon! I pushed the pipeline script to GitHub, this is the one that handles the plotting and analysis of the collected data:

Setup and command reference: https://github.com/ayushman-t/h-line-observer/blob/main/PIPELINE.md

Overview: https://github.com/ayushman-t/h-line-observer#hline_pipelinepy--analysis--plotting

One thing to note: the pipeline currently needs VIRGO installed on whatever machine you're doing the processing/plotting on, with a small patch applied (included in the repo as virgo_patch.py).

If you're running everything on the same Pi that collects data, you're all set, just patch that VIRGO and it'll work, though plotting will be slow on the Pi. If you're using a separate machine for analysis (laptop/PC), you'll need to install VIRGO + apply the patch there too and copy the loop folders and calibration file over.

I'll update this soon, so only the Pi's VIRGO needs to be patched, and the pipeline on PC will just do the plotting without needing VIRGO at all, I didn't set it up that way originally since I was experimenting on my local PC rather than the Pi as my Pi is fully remote. Fairly easy fix, I'll try to get to it this week. But it works fine for now as-is just it's not perfect :)

Let me know if you run into any issues.

Thanks

[Ayushman Tripathi]

BTW the VIRGO patch is just a one-line change on line 887 in virgo.py. Stock VIRGO writes 4 columns to the CSV, the patch adds spectrum_clean as a 5th column which is the median-filtered calibrated spectrum, this is what the pipeline uses for plotting as it reduces RFI and gives a cleaner signal. The raw spectrum VIRGO plots by default is still unfiltered.

Later on I'll add an option for fully raw multiday plots in the pipeline, I'm working on it as I need fully raw data for integrating turboSETI into this.

[Ayushman Tripathi]

Hi Robert,

If you're experiencing any issues with this processing pipeline, I can update the Pi side observer script to output the processed CSVs directly during collection. That way you'd just copy the CSVs over and run the plotting, no VIRGO needed on the other machine at all. You can let me know if that would be easier, and I'll make the change.

Thanks.

[Robert Hamers]

Hi, Ayushman--  Thanks!   No, not having any problems right now.  If it's easy for you, then yes it would be great to be able to output CSVs directly. It's been poor weather here (snow, wind) so not much opportunity to gather data. Mostly I've been trying to identify noise sources and just make sure I can get the best data possibe when the weather cooperates more...f

Best wishes,

Bob

[Ayushman Tripathi]

Hi Robert,

I hope the weather clears up soon!

Yes, I'll add CSV output to the observation script soon.

Thanks.

[Robert Hamers]

Hi Ayusman-

Things have cleared up here and I’m starting to take some data again.  I’ve been getting good results with SDR# with the IF averaging plug-in but would really like to migrate to Raspberry Pi.   I had your program working on my PI5 but had a problem that caused me to re-install the OS and so I had to start over with the installation of everything.   I have it mostly working again, but I’m still getting error related to the plotting.  (See error msgs below). 

I recall that you had a fix for that,  but I can’t seem to find in my emails and  so I’m wondering if you can please refresh my memory ?  I think it involved deprecating matplotlib…

Also — I’m having problems with the large spike at 1420 MHz that comes from the AirSpy itself.    I’ve played around with different gain settings and nothing much seems to help reduce it , (That is, it’s not from external noise but is coming from the air spy itself).    Do you have any suggestions on that ?   Is there a way to shift that (say, by changing the clock frequency ever-so-slightly) so that it’s not right at 1420.00?  

Thanks much!

Bob

Recording 20s...

gr-osmosdr 0.2.0.0 (0.2.0) gnuradio 3.10.12.0

built-in source types: file fcd rtl rtl_tcp uhd miri hackrf bladerf rfspace airspy airspyhf soapy redpitaya freesrp xtrx 

Using AirSpy NOS v1.0.0-rc10-0-g946184a 2016-09-19, samplerates: 2.5M 10M 

Traceback (most recent call last):

  File "/home/hamers/virgoenv/bin/virgo", line 8, in <module>

    sys.exit(main())

             ~~~~^^

  File "/home/hamers/virgoenv/lib/python3.13/site-packages/virgo/virgo.py", line 1304, in main

    plot(obs_parameters=observation, n=args.n, m=args.m, f_rest=args.f_rest,

    ~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

         dB=args.dB, obs_file=args.obs_file, cal_file=args.cal_file, waterfall_fits=args.waterfall_fits,

         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

       spectra_csv=args.spectra_csv, power_csv=args.power_csv, plot_file=args.plot_file)

      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "/home/hamers/virgoenv/lib/python3.13/site-packages/virgo/virgo.py", line 1060, in plot

    ax5.get_shared_x_axes().join(ax5, ax4)

    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^

AttributeError: 'GrouperView' object has no attribute 'join'. Did you mean: 'joined'?

  Saved: /home/hamers/hydrogen_obs/cal_20260327_181144/calibration.dat

=======================================================

  obs_0001

=======================================================

Recording 20s...

gr-osmosdr 0.2.0.0 (0.2.0) gnuradio 3.10.12.0

built-in source types: file fcd rtl rtl_tcp uhd miri hackrf bladerf rfspace airspy airspyhf soapy redpitaya freesrp xtrx 

Using AirSpy NOS v1.0.0-rc10-0-g946184a 2016-09-19, samplerates: 2.5M 10M 

Traceback (most recent call last):

  File "/home/hamers/virgoenv/bin/virgo", line 8, in <module>

    sys.exit(main())

             ~~~~^^

  File "/home/hamers/virgoenv/lib/python3.13/site-packages/virgo/virgo.py", line 1304, in main

    plot(obs_parameters=observation, n=args.n, m=args.m, f_rest=args.f_rest,

    ~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

         dB=args.dB, obs_file=args.obs_file, cal_file=args.cal_file, waterfall_fits=args.waterfall_fits,

         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

       spectra_csv=args.spectra_csv, power_csv=args.power_csv, plot_file=args.plot_file)

      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  File "/home/hamers/virgoenv/lib/python3.13/site-packages/virgo/virgo.py", line 1060, in plot

    ax5.get_shared_x_axes().join(ax5, ax4)

    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^

AttributeError: 'GrouperView' object has no attribute 'join'. Did you mean: 'joined'?

  Saved: /home/hamers/hydrogen_obs/cal_20260327_181144/calibration.dat

=======================================================

  obs_0001

=======================================================

Get Outlook for Mac

To view this discussion visit https://groups.google.com/d/msgid/sara-list/0e7ab215-dbef-4853-9c56-9a571daabbe5n%40googlegroups.com.

[b alex pettit jr]

Hi Bob,

I have Never seen a "center spike"  issue with the AirSpy SDR products,  RTL-SDR YES

FYI :  AirSpy mini Specs :

No IQ imbalance, DC offset or 1/F noise at the center of the spectrum that plagues all the other SDRs

Alex

[Robert Hamers]

Hi, Alex-

It’s not a center-spike—  it is always at exactly​ 1420 GHz,  no matter where I tune the frequency.   I see a spike  like this every 10 MHz throughout the entire Airspy frequency range.   It also scales with gain,  so that lowering the gain also lowers it, but just by the same amount as the noise floor, so there’s no improvement  in S/N by lowering the gain.

I did a lot of tests (taking off the radio horn and substituting a 50-ohm load, for example) and it’s still there.    I’ve seen others mention this also.

I also SOMETIMES see some obvious RFI around 1422, but that comes and goes and is either present or it’s completely absent.  

I’m attaching here a PNG file showing an averaged plot (2-minute average).  You can clearly see the H spectrum, but there’a a spike at 1420.000 MHz.  Other than the 1420 MHz spike it’s very clean.  This is an uncalibrated spectrum, obviously — I prefer to calibrate after-the-fact in software, using a “middle-of-the-night” spectrum.   

Also attaching a screenshot of data that I’m taking right now​…(just an iPhone photo of the screen).    I have some RFI right now around 1422 MHz that comes and goes, and I know that this is local RFI because I can see it disappear at certain times of the day and it depends on where I point my horn.     On the other hand, the one spike at 1420 GHz does not depend on where I point the horn and is also there if I replace the horn with a 50-ohm terminator..

Maybe somethings’ fishy with my AirSpy.       This spectrum I got using SDR#  + IF Average on Windows11,  but I also see it on a Raspberry PI using Gnuradio-companion.

Any thoughts appreciated !

Bob

Get Outlook for Mac

From: 'b alex pettit jr' via Society of Amateur Radio Astronomers <sara...@googlegroups.com>
Date: Friday, March 27, 2026 at 6:50 PM
To: sara...@googlegroups.com <sara...@googlegroups.com>
Subject: Re: [SARA] Re: First-time radio astronomer, looking for advice

--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/sara-list/828916768.662915.1774655408041%40mail.yahoo.com.

[b alex pettit jr]

Try running @ 3 msps and max all 3 gain sliders

Any change ?

On Friday, March 27, 2026 at 08:39:31 PM EDT, Robert Hamers <rjha...@gmail.com> wrote:

Hi, Alex-

It’s not a center-spike—  it is always at exactly​ 1420 GHz,  no matter where I tune the frequency.   I see a spike  like this every 10 MHz throughout the entire Airspy frequency range.   It also scales with gain,  so that lowering the gain also lowers it, but just by the same amount as the noise floor, so there’s no improvement  in S/N by lowering the gain.

I did a lot of tests (taking off the radio horn and substituting a 50-ohm load, for example) and it’s still there.    I’ve seen others mention this also.

I also SOMETIMES see some obvious RFI around 1422, but that comes and goes and is either present or it’s completely absent.  

I’m attaching here a PNG file showing an averaged plot (2-minute average).  You can clearly see the H spectrum, but there’a a spike at 1420.000 MHz.  Other than the 1420 MHz spike it’s very clean.  This is an uncalibrated spectrum, obviously — I prefer to calibrate after-the-fact in software, using a “middle-of-the-night” spectrum.   

Also attaching a screenshot of data that I’m taking right now​…(just an iPhone photo of the screen).    I have some RFI right now around 1422 MHz that comes and goes, and I know that this is local RFI because I can see it disappear at certain times of the day and it depends on where I point my horn.     On the other hand, the one spike at 1420 GHz does not depend on where I point the horn and is also there if I replace the horn with a 50-ohm terminator..

Maybe somethings’ fishy with my AirSpy.       This spectrum I got using SDR#  + IF Average on Windows11,  but I also see it on a Raspberry PI using Gnuradio-companion.

Any thoughts appreciated !

Bob

To view this discussion visit https://groups.google.com/d/msgid/sara-list/PH0PR05MB7590E1C819C1738CA6BB6E37A154A%40PH0PR05MB7590.namprd05.prod.outlook.com.

[Ayushman Tripathi]

Hi Robert,

Glad weather is good now!

For the plotting error, run this to fix it:

pip install matplotlib==3.7.5

On the 1420 MHz spike, Could you share one of the plots generated by my script once you fix the matplotlib error? My script should already run at maximum gain settings so it should give a good reference to compare against.

If the spike persists, also worth trying moving the Pi and Airspy away from any nearby electronics as a test, and putting a ferrite choke on the USB cable just in case though it’s unlikely.

If we can’t fix then also we can mask that frequency out in VIRGO so it doesn’t affect the data.

Thanks

Sent from my iPhone

On 28 Mar 2026, at 06:34, 'b alex pettit jr' via Society of Amateur Radio Astronomers <sara...@googlegroups.com> wrote:



--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---

You received this message because you are subscribed to a topic in the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/sara-list/3RC4V2ZxCQI/unsubscribe.
To unsubscribe from this group and all its topics, send an email to sara-list+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/sara-list/1866023780.677477.1774659874039%40mail.yahoo.com.

[Robert Hamers]

Thanks for your help!   

I’ve done some measurements and posted the results on a google photos site here: https://photos.app.goo.gl/6kRpDJQet7XKxwPr9

At 10 MHz or 2.5 MHz sample rate, both show the 1420 MHz spike, both a low and high gain. (Images 2359 - 2361).  The noise spike is about 25-28 dB above the noise floor. 

I did a series that you might find interesting, all at 10 MHz sample rate:

IMG 2362: All sliders lowest gain:  -100 dB noise, flat across the spectrum, and NO SPIKE..

IMG2363: Slight amount of RF gain,  the 1420 MHz spike begins to come up before the noise floor moves.  Noise floor is still ~ -100 dB.  

IMG 2364: Slightly more RF gain, the 1420 MHz spike is now 25 dB above the noise floor. 

IMG2365: More typical gain settings, noise spike still 25 dB above the noise floor   Noise floor has moved up about 30 dB, but spike is still 25 dB higher than noise floor. 

This is pretty consistently what I’ve been seeing — the spike is always about 25 dB above the noise floor, no matter what the gain settings are.

I’ve tried a few other things, including:

1. Putting in an additional  GPIOlab narrowband filter before the LNA on the theory that maybe this is impacted by strong RF sources outside of my ROI
2. Putting in an additional  GPIOlab narrowband filter after the LNA (instead of before)
3. Running the laptop  (a high-quality Lenovo) from batteries instead of its charger
4. Laptop with both bluetooth off, wifi off, Network off.
5. Running the Elenoo amplifier from a separate battery-powered DC supply with a DC block between it and the Airspy (rather than using the Airspy’s internal bias tee)
6. Using a USB isolator. https://www.amazon.com/DSD-TECH-SH-G01L-Isolator-Isolation/dp/B0CSYNSZ3K/ref=sr_1_1?crid=1MGPQJ55D4A39&dib=eyJ2IjoiMSJ9.cTUkPKswxFLeQYMsyZ359cO04sJVaqa_Z99w1RUNVXueKD2J6GXTxC-1Csi2_OhAuxHrAtB4AzOe6j73mU98hvbR8laQVhTbCIJzcebSuvc4rqCXAmh0tTzGmio42RV5nb0TMh5s1w15d3UQ2OcIln7aLQUWS736KYMJzejO2YGYXrkBIVgfi_cuw18LylsChC7XNoX4Z2OZi1tup3IHDi5ARriPAAwzdSMmTaXGARP8e_e4ruvYzSYwaOPv8l-JVC9X9MjZPr4QXDPTSLkNkuxi4Cw7wpn46QcX5wqTnik.4Fu6V_8yPhAwXFp79z8jteBFbe3QRQGWiTqWXOD7Bsc&dib_tag=se&keywords=DSD+TECH+SH-G01L+USB+Isolator+Supports+USB+2.0+480Mbps+High+Speed+Isolation&nsdOptOutParam=true&qid=1774669941&s=electronics&sprefix=dsd+tech+sh-g01l+usb+isolator+supports+usb+2.0+480mbps+high+speed+isolation%2Celectronics%2C134&sr=1-1

   

7. In the SDR# software I’ve tried using the Linear / Sensitive settings vs. the “Free” settings 

None of this seems to have made a difference.  (Maybe the USB isolator made a small improvement)

I have ferrites on both ends of the laptop-to-Airspy USB cable, and also on the power supply to the Elenoo LNA (when not powering it via the Airspy’s internal bias tee)

I’ve tried moving the laptop as far away from the amplifier as possible

With all this said, I’ve made to get some decent drift spectra and made a 2-D that I’m pretty happy with, but only up to 1420 MHz, and losing the red-shifted parts of the spectrum.  I have written some routines in python to selectively remove the 1420 spike, but they’re not perfect and it’s not just very palatable to have to do this.

Any thought appreciated !

Thanks much!

Bob

Get Outlook for Mac

From: 'b alex pettit jr' via Society of Amateur Radio Astronomers <sara...@googlegroups.com>
Date: Friday, March 27, 2026 at 8:04 PM
To: sara...@googlegroups.com <sara...@googlegroups.com>

--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/sara-list/1866023780.677477.1774659874039%40mail.yahoo.com.

[Robert Hamers]

Slight correctIon:   The LNA I’m using is the Nooelec Sawbird+.    (I said “elenoo” in my previous post…that was incorrect).

Get Outlook for Mac

[b alex pettit jr]

Hello Robert,

I'd suggest : Purchase an AirSpy mini &  try it.

IF it works, you probably can send all this info to AS & get a replacement on the R2,.

You can remove narrow spikes pretty well with this automated process:

" RFI Eraser " : SDR# > IFavg > HL3D > Rinearn Software : Update

" RFI Eraser " : SDR# > IFavg > HL3D > Rinearn Software : Update

Alex

[Robert Hamers]

Alex and Ayushman,

 

Thanks again for all your help and advice.

 

I managed to get my hands on another AirSpy R2, and it is behaving significantly better than the one I was using.  There is still a peak at 1420.0000 MHz, but it is more manageable.

 

I spent some time today comparing SDR# with IF averaging (windows 11 laptop), and   gnuradio/DSPIRA flowgraph, and H-quick.  (both on Raspberry Pi 5).

 

Some observations:

1. With the SDR#/IF averaging on Windows 11, the data using this new Airspy R2 are generally good.   The 1420 GHz is reasonably narrow (but not just a single point) and likely possible to remove.  However, I’d like to move away from having laptop dedicated , so I spent more time on the RPI solutions.

 

2. With the Raspberry PI and the Gnuradio/DSPIRA flowgraph,  the 1420 GHz peak is there and modestly large, but is exactly one point wide. The data points are separated by only 250 MHz in frequency space, and so a simple 7-point median filter does a very good job of completely removing the noise spike. Most of the RFI peaks near 1422 are also just a couple points wide and so are quite effectively removed.  I’m sure there are better ways to filter, but having the noise peaks very narrow certainly helps to discriminate between noise and real  H emission.  

 

3. With H_quick, the 1420 spike is significantly lower in intensity than with Gnuradio/DSPIRA  but is also broad and seems split into twocomponents (that is, like a double-peak) and with a broader base.  The width of the 1420 peak is similar to the width of the envelope of peaks from the Gnuradio/DSPIRA.

 

I’m attaching some graphs showing these results.   The H_quick y values are almost exactly 32 times those of the DSPIRA values, so I think this is just some difference in bit-conversions, as multiplying the DSPIRA values by 32 makes them almost exactly overlay the H_quick values.   In the attached graphs I displaced the graphs vertically a small amount so that you can see them more clearly.    Also attaching an excel file with the original data.    I generally do all my analysis and plotting in a different program (Igor Pro) that I use a lot for my day job, so I’m pasting the graphs from that.

 

AYUSHMAN- For some reason the obs and cal  files produced have 2048 points but appear to be two identical 1024-point spectra in each  file.   I thought perhaps these are Real^2 and Imaginary^2 parts of the power spectrum but not sure, so I just took the first 1024 points for these graphs.  I’m also still getting an error message about plotting, which is a problem with my local configuration.  I have a separate environment on the RPI for the virgo/H_quick software and in that subfolder it has Matplotlib ver 7 (and no evidence of ver 10)  but for some reason it’s still finding matplotlib 10 on the system. I’ll have to dig deeper in to how the RPI does paths to figure this part out…. For now I’m just taking the raw data from H_quick and converting to txt files so I can import into Igor.

 

Happy to hear any thoughts/suggestions….       I’m going  to try another 24-hour drift scan and see how it goes.  See graphs below!

 

Best wishes,

Bob

From: 'b alex pettit jr' via Society of Amateur Radio Astronomers <sara...@googlegroups.com>
Reply-To: "sara...@googlegroups.com" <sara...@googlegroups.com>
Date: Saturday, March 28, 2026 at 1:29 AM
To: "sara...@googlegroups.com" <sara...@googlegroups.com>
Subject: Re: [SARA] Re: First-time radio astronomer, looking for advice

 

Hello Robert,

 

I'd suggest : Purchase an AirSpy mini &  try it.

IF it works, you probably can send all this info to AS & get a replacement on the R2,.

 

You can remove narrow spikes pretty well with this automated process:

 

" RFI Eraser " : SDR# > IFavg > HL3D > Rinearn Software : Update

 

" RFI Eraser " : SDR# > IFavg > HL3D > Rinearn Software : Update

 

Alex

 

 

On Saturday, March 28, 2026 at 12:33:38 AM EDT, Robert Hamers <rjha...@gmail.com> wrote:

 

 

 

With all this said, I’ve made to get some decent drift spectra and made a 2-D that I’m pretty happy with, but only up to 1420 MHz, and losing the red-shifted parts of the spectrum.  I have written some routines in python to selectively remove the 1420 spike, but they’re not perfect and it’s not just very palatable to have to do this.

 

Any thought appreciated !

 

Thanks much!

Bob

 

--

--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/sara-list/1771261369.698496.1774679362513%40mail.yahoo.com.

[Stephen Arbogast]

The  1420 Mhz is  a  spur  from  the  AirSpy SDR.  I have  been dealing with this  for over a year now.  I don't see  the spur  when using  my RTL-SDR.  You  might  try  disconnecting  your AirSpy  from your antenna output and  feed into a  Tiny Spectrum Analyzer.  I bet you won't see  the spur.   I use  software  to remove the spur before  further processing.....  no problems.

[Robert Hamers]

Thanks much- good to know-

Bob

On Mar 28, 2026, at 6:54 PM, 'Stephen Arbogast' via Society of Amateur Radio Astronomers <sara...@googlegroups.com> wrote:

The  1420 Mhz is  a  spur  from  the  AirSpy SDR.  I have  been dealing with this  for over a year now.  I don't see  the spur  when using  my RTL-SDR.  You  might  try  disconnecting  your AirSpy  from your antenna output and  feed into a  Tiny Spectrum Analyzer.  I bet you won't see  the spur.   I use  software  to remove the spur before  further processing.....  no problems.

To view this discussion visit https://groups.google.com/d/msgid/sara-list/6fb4cf98-a8ef-47c3-8431-48b3562a2cffn%40googlegroups.com.

[b alex pettit jr]

An example of my AirSpy Mini SDR  ( 1 of 3 )

( 50 Ohm load on the SawBird LNA )

Alex P

==============================================================================

[Ayushman Tripathi]

Hi Robert,

For the matplotlib path issue: you can try running h_quick.py explicitly using the venv's Python instead of the system one:

/path/to/your-venv/bin/python3 h_quick.py

Or activate the venv first:

source /path/to/your-venv/bin/activate

python3 h_quick.py

That guarantees it picks up matplotlib 3.7.5 from the venv and not the system.

I'll add the CSV export in the next 1-2 days, which will contain raw, calibrated, and median-filtered calibrated data.

On the 1024/2048 question, you're right to take the first 1024. The two halves are the negative and positive frequency portions of the IQ FFT. They're nearly identical (correlation of 0.997) with tiny differences due to IQ imbalance in the SDR. The first 1024 is all you need.

Thanks

[Stephen Arbogast]

Looks  very  good.....   I am using  an  Air Spy, not   mini  or  R2.. but    I did   connect my Tiny Spectrum Analyzer   to  my  Discovery  Dish LNA  output   and did  not  see  a  spike  at  1420 Mhz.....

I  would  like  to  comment  on comments by Marcus  Leech  and Alex  Pettit.....   Maybe  it is  time  to replace "scope  in a  box"   with   the  Discovery Dish   with   HL  feed.

I ordered a  Discovery Drive  today   hoping Kracken  RF   will  meet their goals.....

[Ayushman Tripathi]

Hi Robert,

I pushed a change to h_observer.py so it now outputs the processed spectra CSV and power CSV directly during each observation. If you patch VIRGO on the Pi (one-line change, virgo_patch.py in the repo), the CSVs will have the filtered spectrum ready to go. You'd just copy the loop folder to your PC and go straight to plotting, no need to install VIRGO or run --process on your PC at all.

https://github.com/ayushman-t/h-line-observer

Thanks.