I How feasible is home radio astronomy?

[berkeman]

will make my neighbors sad.

Or mad. Do they already have their own Tiki Torches...?

 

[Baluncore]

Why collect and combine a year, when it can all be done in 12 hours?

Because a 15m disk is awkward, expensive and will make my neighbors sad.

It will certainly be more interesting, Doppler adjusting the pulse rate, then accumulating the time shifted pulses through the different seasons.

 

[ohwilleke]

TL;DR Summary: Is a homemade radio telescope realistic?

Is a homemade radio telescope realistic?

There seems to be a confluence of multiple technologies that makes the situation better than when I was a wee lad: software-defined radio (SDR), the easy availability of satellite dishes, surveillance drives, and fast CPUs.

Let's take a step back - it is trivial to see the sun in radio. An old analog TV, a set of "rabbit ears" antenna, and you're good to go. Point the antenna at the sun (i.e. the ears are perpendicular to it) and there is noticeably more snow and static than when pointing it away from the sun (i.e. lines up with it). But I am looking to see what else can be done.

I imagine getting a couple of DishTV dishes, and mounting them in the corners of my house or yard,. This gives the directionality of a house or yard sized dish, but of course not the sensitivity. Ballpark a few degree resolution for the array (more like 30 for one dish) It is likely easier to point with phase than with motors. Use SDR as receivers, record every night to disk and "stack" days or weeks of exposure together. Because its SDR you can look, e.g. on and off the 21 cm peak and map out hydrogen.

FWIW, the EDGES 21cm radio-telescope, which has gotten very serious academic and scientific attention, was very basic and could have been done as a home astronomy project. A picture of it is at their website:

Of course, part of this boils down to where your home is. A critical feature of this radio-telescope is that it is located in "a radio-quiet zone in western Australia". The EDGES group is building a sister radio-telescope in "radio-quiet—Devon Island in Nunavut, Canada".

So, if you live in the middle of nowhere, this can work. If you live in the East Village of Manhattan, on the other hand, don't bother. The local radio background noise will dwarf anything interesting you can pick up from space aside from the Sun.

 

[Baluncore]

So, if you live in the middle of nowhere, this can work. If you live in the East Village of Manhattan, on the other hand, don't bother. The local radio background noise will dwarf anything interesting you can pick up from space aside from the Sun.

That is one experimental antenna element being evaluated. I believe there will be two in the array, separated by 150 metres. That makes it an interferometer for low-band VHF. While the elements are low-profile, the array is ten times larger than a 15-metre dish.

 

[parac-EU]

I found this forum and clicked on the item.
I was sad to see that some people were so negative about receiving radio astronomy signals.
See below picture of a minimal bucket antenna in the backyard of a house in a city in the Netherlands.
Also a result plot is given.
This can also be a setup for a school project.
I do not know if it is allowed to post links, so I wait and see

 

[Baluncore]

Welcome to PF.
Yes, you can post links.
https://public.nrao.edu/ask/galacti...s-with-a-software-defined-radio-sdr-receiver/

 

[parac-EU]

the admin just posted a link to me

https://public.nrao.edu/ask/galacti...s-with-a-software-defined-radio-sdr-receiver/

In this link there is again the denying of the possibility to receive neutral hydrogen 21cm signals when not using big parabolic dishes.

This is false.

The consequence is that interested persons are blown away by the "experts" like Jeff Mangum and people give up any experiment.
On that site comments are not possible so this lie is there forever.

This is contradicting the principle of science; idea, test, adjust, etc.

Here is a (short) list of maser frequencies which can be persued to probe the forming and death of stars (and the forging of the elements we are made of).
You can see that the lower frequencies are also possible to capture with a standard low cost SDR.
more info on
www.parac.eu

Ps; Hi Marcus, how are your projects going.

 

[berkeman]

In this link there is again the denying of the possibility to receive neutral hydrogen 21cm signals when not using big parabolic dishes.

This is false.

You can see that the lower frequencies are also possible to capture with a standard low cost SDR.
more info on
www.parac.eu

In your link, they use a "9.3 meter solid surface satellite dish"...

 

[parac-EU]

Ha, another one,
this is what I said:
"In this link there is again the denying of the possibility to receive neutral hydrogen 21cm signals when not using parabolic dishes.
This is simply false."

 

[renormalize]

Ha, another one,
this is what I said:
"In this link there is again the denying of the possibility to receive neutral hydrogen 21cm signals when not using parabolic dishes.
This is simply false."

No, what you actually said was:

In this link there is again the denying of the possibility to receive neutral hydrogen 21cm signals when not using big parabolic dishes.

(Emphasis added.)
So please clarify: how big must must a parabolic dish be to receive astronomical 21cm signals?

 

[parac-EU]

We used no dish and no low noise amplifier; just a bucket as shown in the picture.
The project description can be found on
http://parac.eu/projectmk9.htm

 

[Baluncore]

We used no dish and no low noise amplifier;

Was the antenna followed by three satellite TV amplifiers in series ?
What was the RTL dongle plugged into, and controlled by ?
Where was your mobile phone at the time ?

 

[Paul Colby]

We used no dish and no low noise amplifier; just a bucket as shown in the picture.
The project description can be found on

So, the plot shown in post #55 shows a peak. How many dB above the background is it? I ask because RTL dongles have pretty high noise figures. In this link

https://www.idc-online.com/technica...se_Figure_Measurements_of_RTL_SDR_Dongles.pdf

The author measures a noise figure of 17dB at 700 MHz for a garden variety SDR. This means the radios noise temperature is 17dB hotter than room temperature. That’s pretty hot. Is the peak you’re seeing consistent with the noise temperature of the hydrogen you’re looking at?

Did some digging. This paper

https://web.mit.edu/lululiu/Public/8.14/21cm/21cm.pdf

Is an interesting read and quite basic. If I understand Figure 8, the gas you’re observing is only 45 or 50 K which is colder than 300K room temperature by a good amount. Makes me question the results in #55.

 

[Baluncore]

Is an interesting read and quite basic. If I understand Figure 8, the gas you’re observing is only 45 or 50 K which is colder than 300K room temperature by a good amount.

That estimate for galactic source temperature is about right. When the Sun is quiet, it has lower flux than the galactic centre.

The observations from the SDR only cover 13 hours. If it was an astronomical emission, galactic or solar, then we would expect to observe a daily cycle with sidereal or solar periodicity. That is not evident in the short record presented. Even with a 13-hour observation of astronomical signals, I would expect to see some artefact of the antenna pattern, swept across the source by the rotating Earth. A two element interferometer, arranged E-W, with only a couple of hours of observation, would be conclusive. Maybe they use one broad-beam bucket, rather than two, to avoid disappointment.

Until we see a longer record, something that shows a repeating daily periodicity over at least two days, we have to assume the signal being presented is local RF interference, thermal noise in the TV amplifier chain, or in the front-end of the SDR.

At this stage, all the evidence says the project is not radio astronomy.

 

[Paul Colby]

There are much simpler tests I would definitely recommend first. Trying 0 buckets would top my list. These SDRs have a wealth of artifacts, noise sources, gain drift and other limitations that need to be understood first. They could well find this peak shows up anyway. It's quite typical that not all signals detected come in through the SDR RF input.

Long term averaging wouldn't be done this way even with the best of receivers. In the simplest cases people use RF switches to continually cycle the receiver input between very stable known source and the antenna feed. I've actually attempted this using resistive loads and mechanical coaxial switches. I'm not convinced I ever got this working.

 

[parac-EU]

You said that you were not successful with your method.
So, just try our method.
If still sceptical, read att paper written by Jon Wallace (SARA).

 

[Paul Colby]

Thanks. This will get a good read. Nice antenna design. I've dusted off my coaxial switch box which allows me to pragmatically switch between antenna and a 50ohm load. This should make differencing quite a bit easier. This should help reduce the effects of receiver drift. I also have boatloads of RF amplifiers laying about.

 

[parac-EU]

You do not need an RF switch.
The plot indicates that frequency shifting is used. You compare "on" the bump with "off" the bump.
For stability; just enable the Automatic Gain Control on the SDR# gui.

Later you can experiment with all other settings.

 

[parac-EU]

Ps; That horn is not my design.
I am not the author of that paper.
I am the person who is mentioned (about) 10 times.

 

[Paul Colby]

The plot indicates that frequency shifting is used. You compare "on" the bump with "off" the bump.

I'd prefer to have some assurances that any bumps measured actually came in through the antenna feed and not some random interference or drift introduced by moving things like antennas and cables about.

This paper definitely used 2 LNAs and a bandpass filter. LNAs can significantly reduce the effects of system noise temperature, which are considerable with RTL-SDR dongles. Given that any out of band saturation of the amplifiers will add considerable noise in band, the bandpass would be better placed before and not after the LNAs.

 

[AlexB23]

Hey guys. I understand 0% of this article, but the guy made a home radio telescope using an old TV receiver.

https://www.arrl.org/files/file/ETP/Radio Astronomy/Build a Homebrew Radio Telescope-QST-0609.pdf

 

[Baluncore]

In the paper written by Jon Wallace (SARA), it appears, from the 24-hour records, that radio astronomy is being done. That system employs LNAs, based on ultra-low-noise HEMT FETs. Each stage has a gain of about 19.5 dB at 1420 MHz, at a cost of US$150 each. Those LNAs lift the RA signal to just above the thermal noise of the SDR mixer front-end. The low-cost signal amplifiers available for TV, do not have that expensive ultra-low-noise advantage.

The SDR software will also have a transform gain from the use of an FFT, which is then followed by noise reduction through Power Spectrum Accumulation, PSA. At some point in the software process, the narrowband local interference is being removed, the resulting holes in the spectrum are then being smoothed over.

 

[Paul Colby]

Here’s a reference to some SDR evaluations.

https://www.klofas.com/blog/2020/sa...odifications/Evaluation_of_SDR_Boards-1.0.pdf

They look at noise figures. I post this because the SDRs evaluated preform much better than I was expecting at 21cm. I also happen to own several of these devices. The Sdrplay radios come with internal LNAs which typically provide 4.4dB ish NF at 21 cm. This isn’t anywhere near as bad as I was expecting. I spend most of my radio time in the HF and VHF bands where the NF is quite a bit worse. Some of the newer RTL-SDR dongles are also not so bad.

 

[Paul Colby]

Before I get out a hacksaws and take multiple trips to the hardware store building an antenna, I thought I would first see if my Sdrplay RSPdx is even capable of detecting the small temperature changes needed to make observations of the 21cm hydrogen line. The radio has a LNA built in making life simple. I'm using a coaxial switch box to alternate between a 50 ohm load at room temperature and one I've placed in ice water. The temperature difference, $\Delta T = 22.2$ degrees kelvin. The change in power output expected is,

$\Delta P = 4 k_B \Delta T R$

where $R=50$ ohms. The x axis is in MHz while the y axis is mW. The blue curve is the difference in the measured power in mW with both terminators at 72 degrees F. The displacement from 0 is due to small differences in the cables, connectors, switch positions etc. Ultimately I hope to calibrate these out. The yellow curve is the same with one terminator is immersed in ice water. A ballon is used to keep it dry.

This is still very much a work in progress. It's unclear I have all my factors of 2 in place. The observed change seems smaller than expected. Anyway, I'm convinced this test shows enough life that astronomical observations using this setup are in fact possible.

The power spectra shown are 2048 bins wide. Data is collected 1000 times for each switch position and 100 such cycles were taken and averaged. Each curve is about 6 min of data collecting. Between each switch change a dwell of 10 spectra are taken and discarded to ensure switching noise has died out between data sets. The IQ sampling rate is 2MHz.

 

[Baluncore]

If you get serious about doing home RA, you will build your own two channel receiver, from FET based LNAs, with matched band-pass filters to remove the nearby and adjacent channel interference. Once sufficient gain has been applied, you can down convert the signal to an IF or baseband. By using two channels with crossed dipole antenna elements, you can select polarisation, linear or circular. Your two front-end mixers will need to use the same LO, so pick your SDR wisely.

Depending on how serious you are, those LNAs will be cooled by a cheap bar fridge, a stack of Peltier junctions, LN₂, or a helium pump refrigerator, if you should be so lucky.

For a 2MHz wide observation band, you will digitise the two signals at about a 10MHz conversion rate. That could be done by an A-D converter, or an SDR module.

Having gathered two channels at 10 megasamples each, in let us say 0.1 second, you have two records of 10⁶ samples each. You then take the FFT of the 10⁶ samples, (using a PC to compute the FFT in real time), and see a reduction in the noise floor of; √10⁶ = 10³, about +60 dB of transform gain.

Over 5 minutes, you Power Spectrum Accumulate 5 * 600 = 3000 of those spectra. That lowers the noise floor again by √3000 = 54 times, yielding another +34dB of gain, making a total of 60+34 = 94 dB in the processing after the LNAs. That is why you need multiple external LNAs in the receiver at the focus, it will get you the cool gain you need later to make deeper observations, like looking for chemical MASERS.

With a dish, bigger than a bucket, you will then be able to observe more than just our galactic centre once each day.

 

[Paul Colby]

Well, being serious about doing science and being interested in learning how things work by doing them really aren’t the same. I had serious doubts I could even detect the thermal noise from a 50 ohm terminator let alone a 10% change in said noise with the equipment on hand.

Having gathered two channels at 10 megasamples each, in let us say 0.1 second, you have two records of 106 samples each. You then take the FFT of the 106 samples, (using a PC to compute the FFT in real time), and see a reduction in the noise floor of; √106 = 103, about +60 dB of transform gain

So, this is an elementary point I’d really like to understand. I chose averaging 2048 wide FFTs rather arbitrarily. Each is scaled by 2048 to obtain mW/Hz as a vertical axis. Wider FFTs would just yield finer frequency bins. If the phenomena one’s interested in isn’t narrow band, why select finer grain bins?

 

[Baluncore]

If the phenomena one’s interested in isn’t narrow band, why select finer grain bins?

You are free to select your transform size.
For one project, I was monitoring the background noise floor across the HF band from 1 MHz to 30 MHz. I needed to look between the transmitters, which required narrow bins. There was some pulse noise, so I also kept score in a counterintuitive way, by following the minimum of every bin.

A noise pulse that falls in the time record, will raise the noise floor across the spectrum, unless you can overwrite it with zero in the time record, before the transform.

Any interfering carrier or spur that falls in your data needs to be removed by zeroing those bins in the frequency domain.

If you look at the pass-band of an SDR you will see the in-band ripple of the digital BPF, and alias spurs. That digital BPF is a source of noise that is added to the signal. For RA, it is better to build a matched pair of LC filters for the duplicated front-end LNAs.

 

[parac-EU]

The very first time I could detect hydrogen with an antenna I was amazed and overjoyed.
That is why I offer,whoever want to do likewise, my support.
Dont build complicated systems; just start simple and have success.

I have repeated the project
http://parac.eu/projectmk9.htm
I am now in Portugal and now the bucket is replaced by a tube on the ground and an ali amplifier.
The tube was positioned first to the polar star, so the milky way is circumpolar and always visible.
The beamwidth of the tube is 57*lambda/tube diameter, so about 80 degrees.

Next the tube was placed to the zenith.
I have put a flare onto it, and again according to Huygens principle with 320mm diam outer rimm, it should become 40 degrees.
This narrower beam will give more details of the milky way arms.

Directing the tube with flare more south will give a transit twice through the milky way plane, giving seperate bumps on a different frequency.

The signal is strong enough; if you wat more detail, then make the beam more narrow, for instance by adding a flare or dish.


I want to add that the s/n ratio of the sdr is not so important; the first amplifier is.
See the excellent explanation by Marcus in this thread.

Further it is important that fast sample and FFT software is chosen; I use free CFRAD2.exe
This software will sample and add 100000 spectra in 5 minutes.
Also use a multicore PC, because one core will be fully occupied.
see www.parac.eu

Concerning masers; see for instance the project 'masers with a 1m dish'
http://parac.eu/projectem01.htm

 

[parac-EU]

forgot to add