HF Receive Loop Antenna and Man-Made Noise Detector

Published on by pa1ejoLeave a comment

At both radio clubs where I do my contesting and meet with fellow amateurs, a frequent topic of discussion is the noise we all experience on the shortwave bands. During contests, there is almost always some level of man-made interference, even when operating far away from populated areas. This raises the question: where is the noise coming from in the clubhouse itself?

To investigate this, I built a simple portable loop antenna. It is a fun project, and chances are that you already have most of the required materials at home.

Receive loop antenna, good for listening on shortwave and detecting noise sources in your neighborhood.

At the top of the loop, make a crossover by connecting the centre conductor to the shield of the coax on both sides. At the bottom, connect the two shields together and install a 1:1 balun wound on a multi aperture ferrite binocular core (a pig nose ferrite). The balanced side of the balun connects to the two centre conductors of the loop, while the unbalanced side connects to an RG-58/U coax cable leading directly to an SDR.

Modern SDRs, such as the Airspy HF+ Discovery or a Malahit receiver, are sufficiently sensitive that no additional amplification is required. A spectrum analyzer shows signals from -130 dBm upwards to -80 or higher dBm’s, which may be a challenge on your analyzer but not a real issue for an Airspy HF+ discovery. This antenna can be used on the 20 m and 40 m amateur bands. But it is also very effective for locating sources of radio-frequency interference, such as noisy switch-mode power supplies and LED lighting. At both club houses I did locate them, what we do with this knowledge is a separate story.

Smaller versions of the loop can be built from semi-rigid coaxial cable. These are particularly useful for pinpointing interference sources at close range. If you want to use the loop as a receiving antenna some distance from the receiver, you can add a wideband 20 dB amplifier after the balun. This allows a longer coaxial cable to be used and makes it practical to place the loop outside, for example in the garden. In that configuration, the design closely resembles the well-known LZ1AQ receiving loop concept.

Last update: 24-Jul-2026, tools used to write this text, Google translate from Dutch to English and AI (ChatGPT) to check for readability.

QO100 FT4/FT8 frequency stability

Published on by pa1ejoLeave a comment

The waterfall of the wsjt-x decoder of QO100 provides an indication of the ground transmit/receiver oscillator frequency stability of users at that time working over the microwave transponder. The decode window shows a 3 kHz receive window starting at 10.489540 GHz. We can see that some of the FT8 signals slowly wander in frequency space probably due to temperature effects. The worst ones drift by 250 Hertz in several minutes, it comes down to a drift of 0.5 Hz per second. A crystal oscillator that heats up will decrease in frequency, but we see something else, a number of oscillators cool down so that the frequency increases.

A better solution is in my opinion to rely on ovenized oscillators. Our own frequency reference is an oscilloquartz 8663xs running at 10 MHz and it reaches 5e-13 for df/f at an integration time of 400s. This results in a frequency drift of better than 0.005 Hz, so a factor 50000 better than what you see for some oscillators in the WSJT-x waterfall which are more on the level of 2.5e-8 in terms of df/f.

The WSJT-x FT8 decoder is able to decode signals with a frequency drift in Hz/s, beyond 3 Hz across the 13s FT8 cycle the sensitivity of the decoder rapidly decreases. Another practical problem is that a transmission with a large frequency drift will overlap with other transmissions which will also affect the probability that the FT8 signal is decoded correctly.

In other words, the succes of FT8 on QO100 increases when you use a high quality OCXO. Best practice is therefore to use a better local oscillator both for the transmitter and the receiver, so the QO100 up and down converters for 2.4 and 10.5 GHz should be locked to a 10 MHz frequency reference with a performance better than 1e-10. Allan deviation analysis shows that this can be achieved with a number of oscillators.

In fact all of the oscillators in this graph have a sufficient performance, you want to stay as much as possible below the 1e-10 level, it can be achieved with a decent OCXO, a Rb cell or a GNSS disciplined oscillator. Synchronizing your OCXO to a 1 pps signal can also to the job, but you need to give it some time (like 3 to 5 minutes) to fully stabilize reaching the 1e-10 level.

Last update: 10-jan-2026 12:30

Allan deviations of precision oscillators

Published on by pa1ejoLeave a comment

Since I got the TinyGTC instrument of Erik Kaashoek I’ve collected frequency and phase data from precision oscillators. This idea is to get an impression of the relative noise for different integration times. My definition of a precision oscillator it that it will give you an error of less 1e-9, equivalent to better than 10 Hertz at 10 GHz which is the receive frequency of the QO100 satellite. All collected phase differences by the TinyGTC are put in the TimeLab software where a fe5680a is the frequency standard to compare to:

Integration times tau start at 1s and they are all collected at 10 MHz, except for the 1 pulse per second (PPS) experiment from a Ublox GNSS receiver. If you integrate longer than 10s than all oscillators satisfy the criterion of a precision device. If you are not disciplined to GNSS then Allan deviations will show a noise floor somewhere around 400s where the deviations vary between 1e-11 and 1e-12. This level is well above the expected noise floor of the phase detector in the TinyGTC. In the end beyond 3000s of integration time GNSS disciplining will win from all my local oscillators. This is the regime where temperature variations and other local disturbances take over.

Last update: 22-Dec-2025 6:30 LT

Lightshift jump of my Rubidium frequency standard

Published on by pa1ejoLeave a comment

The attached timelab plot shows the phase difference via different combinations observed with a TinyGTC, a dual port frequency counter with the ability of observing phase difference measurements between the ports. The GTC analyzer includes a numerically controlled internal oscillator (NCO) which is disciplined to GNSS. The GTC analyzer is built by Erik Kaashoek PD0EK.

I ran the experiment for 8 hours, port 1 (P1) is connected to an oven controlled oscillator from the Oscilloquartz company, port 2 on the GTC analyzer is connected to a 10 MHz rubidium standard, and there is a NCO disciplined to GNSS in the GTC analyzer.

Figure 1, phase difference plots of three oscillators

Figure 1: The green curve shows how the OCXO compares to the GNSS, the blue curve is the Rb standard relative to the GNSS, and red is the phase difference between the OCXO and the Rb standard.

At t around 7 h 7 min we see a discontinuity in the phase difference of the blue and the red curve; in figure 2 this is visible as a discontinuity in the frequency difference representation at 7hr and 7min.

Figure 2: frequency discontinuity observed in the 10 MHz Rubidium frequency reference.

The phenomenon is known as a light shift jump [1] which is caused by intensity variations in the lamp. The analysis also shows that light shift jumps are correctable with the help of other oscillators, for instance the availability of a GNSS disciplined NCO and a free running (not disciplined) OCXO.

Light shift jumps are small (in my analysis approximately 0.2 milliHertz) but clearly visible, and should be corrected if you intend to realize a frequency reference with a short term accuracy better than df/f = 2e-11. In the allan deviation representation light shift frequency jumps show up at integration time greater than 1000s, if we would correct light shift jumps then we would be able to extend the noise flicker floor in an Allan deviation analysis.

[1] https://ieeexplore.ieee.org/document/8255660

Last update: 3-Dec-2025 9:36

Audi Map updates

Published on by pa1ejoLeave a comment

Audi offers navigation map updates via a paid subscription, but the updates are infrequent and quite expensive. Fortunately, you can perform the update yourself for free using the official myAudi website. Here’s the exact process:

  • Log in to myAudi
    Go to the myAudi website, register yourself and your vehicle, and navigate to Map Updates.
  • Download the update package
    Download the ZIP file for your region. The file is large (around 29 GB), so this may take some time depending on your internet speed.
  • Prepare the USB stick
    • On your PC, extract the ZIP file to a new folder (this can take up to an hour).
    • Copy the contents of that extracted folder (not the folder itself) to an empty USB stick formatted as FAT32.
  • Install the update in your Audi
    • Switch on the ignition.
    • Insert the USB stick into the USB port in the center console.
    • Open the Navigation → Settings → Software Update menu.
    • Select the available update and confirm when prompted.
  • Wait for completion
    The installation takes about one hour. Keep the ignition on during the process. You’ll see a progress bar on the screen, and once complete, the system will confirm that the update was successful.

Tip

If you have a longer drive planned, start the update during that trip so the ignition stays on naturally.

Result

You’ll have the latest maps installed — without paying your dealer’s typical €160 service fee. The process is straightforward, entirely official, and keeps your navigation system up to date for free.

Last update: 31-oct-2025 (enhanced by AI)

M31 Andromeda Galaxy

Published on by pa1ejoLeave a comment

I observed it for three nights, once with the Seestar S50, once with the Askar FRA500 in mosaic mode (F5.6) and once with a reducer (F3.4) without a mosaic. Only the Optolong Quad-l Enhance filter was used, in the end you combine all three nights, and that results in this image:

M31 observed in three nights

The astrometric evaluation yields this:

Annotated image of M31

In total there is 2 hours of exposure by the Askar FRA500 data, and 3 hours of data from the Seestar S50. The combination of all images yields more details that any of the single nights. All processing was done in Pixinsight with the RCastro tools.

M31 Andromeda is 2.5 million light years from us, it is our nearest neighbor on a galactic scale. In addition we you also capture M32 and M110, so we jump three steps ahead in the Messier marathon.

Combining several nights

In August and September I collected 3 nights of Askar FRA500 data and one night of SeeStar S50 data. The S50 data is left out because it is noisier and there are dead pixels in the images. From the remaining nights I stacked all frames that belong together, rather than making the mosaic by night and possibly by exposure setting I now make a composite mosaic over all nights, and that probably looks best, but I admit you would need to look at the details to see the difference

Composite over three nights, all frames in one mosaic

Last update : 7-Sep-2025 7:00 LT.

Cygnus loop

Published on by pa1ejoLeave a comment

With the 90mm aperture scope (The Askar FRA500) I have a reducer that brings it back to F3.9 so that the Cygnus loop can be observed in a mosaic with two images. Tried that earlier and it ended up in a disaster without the reducer or even the C8 with reducer since you have to map the CL with 6 or more overlapping images. In Stellavita it turns out to be too complicated, no way to automate the procedure but in NINA it is doable with the sequencer. I had to understand that part first, next in pixinsight the problem is to combine LP Ha Oiii and Sii in individual mosaics, perform a star alignment between mosaics for LP and all narrow band exposures and finally use pixelmath to create the composite. What you get is a false color map with Ha in red, Oiii in blue and Sii in green. This is what you get for the full Cygnus loop, for more details check astrobin (I’m missing a small part in the bottom, maybe in the future I will redo this system, but on the 12/13th August the night turned out to be too short).

The full Cygnus loop: there are two panels, 4 filters and 5 exposures of 90s each, total exposure time is 3600s, post processing goes via Pixinsight. and RC astro tools and Siril. Recorded on 12/13 Aug 2025 in my backyard.

In the end I performed some very modest histogram stretching, deblurring and denoising with the RC-astro tools and GraXpert. Overall I’m quite pleased with the result. the last Sii panel almost disappeared behind the roof and there was a meridian flip which makes everyone nervous.

This last image is to check once again whether we did not hit the tripod, it is a tricky thing and I don’t have a good solution, wish the scope was further away from the mount. Will see whether there is an easy solution for that. The Cygnus loop is well documented on wikipedia. This system is 6 times larger than the moon, and a challenge to get it right because of its size.

Interpretation

The Cygnus loop is a remnant of an old supernova that happened 10 to 20000 years ago, it is at a distance of 2600 lightyears. In the image you see a shockwave expanded to around 3 degree in diameter, or 1.5 degree at 2600 lightyears away from us. The radius of the ring is therefore 68 lightyears, also it means that the shockwave is traveling at around 0.3 to 0.6 percent of the speed of light, or between 900 and 1800 km/s. Faster than anything we ever launched.

If we would be within 100 ly from a supernova then we wouldn’t survive it, this is part of the Drake equation. Fortunately you need a star 8 times as massive as our sun for a supernova, and there aren’t that many of those around in our neighborhood. You need to turn the story around, our sun is in a part of the galaxy where there are no heavy stars (>8 solar masses) within about 100 ly. Otherwise we wouldn’t have survived earlier supernovas.

Status

Last update: 14-Aug-2025 16:30 LT, update 18:35, it is Sii and not Siii, moreover I added text like an interpretation of the system.

NGC6888 Crescent Nebula

Published on by pa1ejoLeave a comment

Some call it a baked patato or brains or even a soap bubble, the system is discovered in 1792 by William Herschel and it is at a distance of 4700 light years. NGC 6888, also known as the Crescent Nebula, is caused by the collision of stellar winds from a Wolf-Rayet star and the material it previously ejected as a red giant (according to google AI, I simply googled it together).

I recorded this system in three ways using the SeeStar 50mm scope and the Askar FRA500 F5.6 scope where I can change between a LP and a UHC filter from Optolong using a ASI183MC pro camera from ZWO cooled to -10C. Here is the SeeStar S50 result post processed in Pixinsight, 35min exposure time, in PI i reduced the stars are enhanced the nebula, otherwise everyone says that there are too many of them. Well dim a few and you are fine.

For the 90mm scope I used 1800s exposure but we had a waxing moon at the same time, here is the version where I used an Optolong light polution filter, the LP 90mm version clearly has some more detail.

And there is the UHC filter, a dual narrow band filter from Optolong, also 1800s exposure.

The S50 image is not far from what you get on an Askar FRA500 F5.6 with more aperture, filters and cooled ZWO color camera.

Update on 12-Aug-2025

Next night I went back with the same scope and a monochrome camera, the ASI 1600 mm pro and narrow band filters. Next I combined hydrogren and oxygen and I got this:

This is NGC6888 in Ha and O3, Ha is red and O3 is blue, the red stuff around the backed patato (or better strawberry in this color representation) is real, it is interstellar hydrogen. The tendency with LP images is that background gradient removal kills this signal, don’t do that, it is real signal.

Last update 12-Aug-2025 9:52 LT.

Caldwell 33, in hydrogen alpha with a different scope

Published on by pa1ejo1 Comment

This summer my vacation time went into setting up a new scope and mount controller. The scope became an Askar FRA 500/90 refractor controlled by N.I.N.A. running on a miniPC using an iOptron HAE18c mount. Furthermore I installed a ZWO auto focusser and their filter wheel. The filters come from various sources, the narrow band filters come from AliXpress. Last night I went back to the supernova remnant Caldwell 33 and recorded a 1200s image with a hydrogen alpha (Ha) filter.

1200 second Ha recording of C33, post processing in pixsinsight, RC-astro tools and Siril, 29-Jul-2025

The telescope/mount set-up looks as follows:

Carbon tripod, iOptron HAE18c mount with counterweight, windows 11 pro miniPC running N.I.N.A, ZWO focusser and filter wheel, ASI1600 MC pro camera cooled to -10C, SVbony powered USB hub, dew heater. and some Velcron straps to keep everything together.

The C33 image shows you more details compared to what I got from the Seestar S50, this is to be expected mostly because of the sensor differences and the aperture of the scope. What you see in figure 1 is only 20 minutes of of observation time while the S50 required a lot more photons to make a nice image (more than 1 or 2 hours I thought).

My big problem this summer is the weather, I had more observing opportunities in the spring and the winter, in this July 2025 you really need to look for suitable nights, often there is only an open cloud deck for an hour or so. My ambition to map the entire Cygnus loop may be difficult to achieve, please, a few more clear nights so I can complete this project.

Three scopes

If you are new in this hobby (I consider myself to be a newbee) then start with studying F numbers, aperture of the scope, the use of a focal reducer, exposure times, filters and camera’s. For all scopes I have I used an internet astro tool to compute the field of view for different systems [link]. And for this you take the largest object in the sky, M31 which is something like 8 times the size of a full moon. So what does M31 look like when you observe it, and, what do you like best for your activity? What are your ambitions? Ask yourself such questions before starting on this hobby.

Field of view calculations for different systems, the fun of the hobby is largely determined by the FOV and the time you need to be able to observe an object.

Some considerations:

  • The scope I started with, a C8 with a reducer needs a handful of images (like 10) to map M31 completely. I will lead to frustration and you will probably not complete it easily unless you put time in the details within pixinsight. So it becomes a multinight project.
  • My second scope, a S50 would map it at once, but it is a non-cooled camera with a 2MP resolution so you will need a lot of time to figure out how to do it right. The challenge is, to reduce the noise and to catch enough photons.
  • The current Askar FRA500/90 refractor will do the job for you in much more details than the S50, and probably you don’t need a lot of time to process it. There are maybe some things to improve but this is currently my preferred system.

It is entirely up to you what you want to do, and what effort you are willing to put to come to a nice result where you catch some old photons. I find the youtube videos of Cuiv the Lazy Geek quite helpful in this.

Last update: 29-Jul-2025