The antenna is connected and noise rises. What have we proved?
A coloured band appears on a laptop screen. Disconnect the antenna and the background darkens; reconnect it and the display brightens. The receiving system responds to an external input. But this experiment has not established whether that input came from the Sun, Jupiter, a neighbouring building’s power supply or a radio station.
A Radio JOVE teaching radio telescope consists of an antenna, a receiver and recording software. Each stage needs testing, while a celestial origin requires separate evidence. We will examine one specific construction: two wire dipoles, matching cables, a combiner and an SDRplay RSP1B. It is a passive receiving system, with no transmitter in the circuit.
From the article on radio telescopes and receivers, we need one idea: a radio wave produces an alternating voltage that the chain selects, amplifies and digitizes. This does not always require a large dish. Wires can be antennas too, provided their geometry suits the frequency range.
Construction status: the documentation and calculations have been checked; the author has not physically built or tested this system, and no personal recording of the Sun or Jupiter is presented. The diagrams explain a documented build; the coloured teaching example below is generated mathematically.
Instead of a photograph: power across frequency and time
Radio JOVE developed as an educational project for people who want to work from components to data; NASA’s project page dates its beginning to 1999. The current RJ2.1 configuration uses a ready-made software-defined radio, SDR. Software performs some tasks handled by separate electronic stages in other radios. The older analogue kits and RJ2.0 with an RSP1A are different versions. NASA: about the project, kit versions.
The receiver observes a band around 20 MHz, where strong solar radio bursts, Jupiter’s decametric emission and radio background can be studied. These are possible targets, not guaranteed events at a chosen hour. The ionosphere and local interference affect reception; Jupiter being visible through an ordinary telescope does not ensure a detectable radio burst.
The result is a spectrogram: time increases to the right, frequency upward, and colour represents the recorded level. The system does not measure a sky position for every coloured patch or photograph a planetary disc. For many radio-astronomy questions, the time and frequency structure is itself the useful measurement. NASA: observing conditions, Radio-Sky Spectrograph’s author: what the program records.
First, fix the entire configuration
A dipole consists of two conducting arms with a gap at the centre. The two conductors of coaxial cable connect to them: the centre conductor and the braid. Two identical dipoles allow their signals to be added with a defined phase relationship. Cable lengths and connection orientation must therefore agree.
We select a version from the Radio JOVE manual: horizontal east–west wires, two rows separated north–south, nominal height 3.048 m, and no extra phasing cable. Its modelled pattern in the manual peaks at the zenith. It provides a useful teaching case for receiving sources high above the horizon. Actual observations begin by checking the target’s elevation for the particular location and date; this configuration is not universal pointing for every latitude. Antenna Manual v2.1, upper fig.3.11, p.15.
| Component | Selected size or type | Why it matters |
|---|---|---|
| Two dipoles | Each spans about 7.09 m between working tips; design frequency 20.1 MHz | Antenna geometry depends on frequency |
| Row separation | 6.096 m, or 20 ft | Sets the relative positions of the receiving elements |
| Suspension points | Nominally 3.048 m, or 10 ft | Height above ground affects the pattern |
| Two feed cables | Each 9.85 m RG-59, Belden 8241, 75 Ω, VF=0.66 | Identical branches give identical cable delays |
| Combining | Kit 2-to-1 combiner, no additional phasing section | In-phase sum for the selected direction |
| Common cable | 4.93 m of the same RG-59 | Carries the already combined signal to the receiver |
| Receiver | SDRplay RSP1B, 50 Ω SMA input | Converts the received signal to a digital stream |
| Recording | 16–24 MHz band, 512 channels in the software bridge | Frequency–time recording |
Feed and common-cable lengths come from table 4.4 of the antenna manual. We do not mix them with longer cables or phasing sections from other configurations. Nor is an arbitrary television splitter automatically a suitable combiner: its operating band, ports and phase behaviour must be appropriate.
Branch · With formulasOptional dimension calculations at 20.1 MHz; the main build continues without them.
Wire, a wave in air and a wave in cable have different lengths
At frequency f₀=20.1×10⁶ Hz, the free-space wavelength λ₀ follows from the speed of light c=299792458 m/s:
λ_(0) = (c)/(f_(0)) ≈ 14.915 m,; (λ_(0))/(2) ≈ 7.458 m.
A real resonant wire dipole is shorter than this half-wavelength: end effects and conductor geometry change its electrical behaviour. The JOVE manual gives an empirical starting estimate with f expressed in megahertz:
L_(dipole) ≈ (142.65)/(f_(MHz)) m,; L_(dipole)(20.1 MHz) ≈ 7.097 m.
The documentation rounds the finished span to about 7.09 m. This does not guarantee an exact resonant frequency without measuring the particular antenna: ground, suspension and nearby objects matter. Do not cut each finished arm to 7.09 m; that dimension describes the whole dipole.
In cable, propagation speed depends on VF, the ratio of wave speed in that cable to c. With VF=0.66:
λ_(cable) = (VF c)/(f_(0)) ≈ 9.844 m,; τ_(cable) = (L)/(VF c).
One wavelength in our RG-59 is therefore approximately 9.85 m of cable, and half a wavelength is 4.93 m. The 14.915 m free-space wavelength cannot be used as the coax length. Exact c and nominal VF give about 360.22° for a 9.85 m section rather than mathematically exact 360°: these are rounded assembly dimensions and an approximate material coefficient. For in-phase combining, what matters is that the two feed cables match. JOVE manual, §3.1–3.2 and table 4.4.
The site must accommodate guys as well as wires
The two rows of wire could fit in less space than the complete installation. But the supports must remain upright, with guy ropes anchored outward. Our 14×11 m site plan contains four masts, the wires and eight stakes. Mast spacing within a row is approximately 7.62 m; short insulating ropes connect the supports to the dipole ends. This reproduces the geometry of manual fig.7.1, rather than calculating an allowable wind load.
A site containing supports, guys and two dipoles
- E–W wire
- Mast
- Stake and guy
- Cable route
North · N · East · E · Combiner: 2→1 · RG-59: 9.85 m + 9.85 m → 4.93 m
Choose an open ground-level site away from overhead power lines, metal fences and electrical noise sources. Neither wire nor supports must be able to fall onto a power line. Even a low antenna needs carefully chosen attachments; mast sections, anchors and suitability for a permanent installation depend on local soil and weather. The manual shows metal and PVC supports. For the selected height, reproduce its appropriate 10-foot arrangement rather than inadvertently using a telescoping configuration intended for another height.
Mark out the site in daylight. The manual requires at least three people to raise the masts; mark guy ropes with visible tape. Avoid ordinary walking routes where people would trip over them. Do not start installation or testing when a storm approaches; disconnect the antenna beforehand. Disconnection alone is not a complete lightning-protection design for a permanent outdoor installation. Antenna manual, §6–7, especially pp.38 and 42.
In a city, the largest difficulty may be strong interference rather than a weak antenna: switching power supplies, lighting, electronics and radio transmitters. Surveying the radio background at an accessible safe site can therefore be a useful first step. Extra amplification does not separate celestial signals from interference and can overload the receiver.
Build two matching electrical assemblies
The antenna requires copper wire, RG-59, six insulators, six ferrite toroids, a combiner, F-connectors, tie-wraps, insulating rope, four supports and hardware. The standard kit specification includes 15.24 m of seven-strand #14 AWG copper wire and approximately 29 m of Belden 8241. Our three cable sections total 24.63 m before allowances for the particular end preparation. Four F-connectors suffice without a phasing cable; the kit’s extra connectors and coupler remain spares. Manual tables 4.3–4.4 and §5.
Assembly order matters because an electrically sound connection can still fail mechanically:
- Measure the two identical feed sections and the common cable. Label them before installation. Do not replace one RG-59 with another cable type whose VF is unknown.
- Prepare four wire blanks with allowances for wrapping around insulators. The manual specifies approximately 3.759 m each, including about 0.127 m at each end for fastening. These are blanks, not finished electrical arms; after assembly, check the overall span of about 7.09 m against figs.3.1/7.2.
- Fasten two arms to each centre insulator and an end insulator at either tip. Preserve the central gap: do not bridge the two arms with a wire.
- Connect the coax centre conductor to one arm and the gathered braid to the other. Choose the same side on both antennas: for example, the centre conductor always connects to the east arm. Solder joints need good electrical contact, but the insulator and tie-wrap must carry mechanical load.
- Secure three kit ferrite rings around the whole coax near each dipole’s centre. They restrict unwanted current on the outside of the braid; they are not components inserted in series with the centre conductor.
- Fit F-connectors without allowing any braid strand to touch the centre conductor. Connect the matching feeds to the combiner inputs and the common 4.93 m cable to its output. There is no additional phasing section.
Matching polarity, separated conductors and strain relief
- Centre conductor → E
- Identical at both dipoles
- Braid → W
- Insulator / gap
- 3 toroids around coax
- Strain relief
Avoid nicking the centre conductor while preparing the braid: damaged wire may break as the cable moves. Provide strain relief at the suspension point; solder must not support the entire coax weight. Weather-seal feed points, connectors and combiner with suitable outdoor coax sealing materials. A loose wrap of ordinary tape does not guarantee lasting protection. Keep the receiver, USB connections and computer dry, and ensure the physical cable route actually reaches them with slack. Manual figs.5.3–5.5 and §7.1.
Branch · DeeperOptional idealized voltage-sign check, not a complete model of the real antenna pattern.
Reversing one dipole can cancel the wanted sum
For a source directly overhead, matching antennas meet the same wavefront simultaneously. Matching cables preserve that relationship. But reversing the arm connections on only one dipole changes its voltage sign: contributions intended to add begin to subtract.
Let both amplitudes equal V, with phase difference Δφ. Normalize the sum by dividing by two for comparison; this is mathematical normalization, not the combiner’s amplification factor. Relative mean power for one coherent signal is:
u_(1) = Vcosωt,; u_(2) = Vcos(ωt + Δϕ),; (P_(sum))/(P_(aligned)) = (cos)^(2) ((Δϕ)/(2)).
At Δφ=0 the ratio is 1; at 90° it is 0.5; at 180° ideal cancellation occurs. Receiver noise, unequal amplitudes and signals from other directions do not vanish in the same way. Thus “there is still noise after connecting both” does not prove correct polarity.
The same signal adds or cancels
- u₁
- u₂
- Normalized sum
Matching orientation · Δφ=0° · P/P₀=1.0
One dipole reversed · Δφ=180° · P/P₀=0.0
An extra ΔL metres of cable would introduce Δφ=2πfΔL/(VF·c). The common cable after the combiner delays the already combined signal without creating a difference between its two inputs. These two elements form one receiving sum, not the complete set of correlations for imaging; that distinction is explained in the article on interferometers. The matching-orientation requirement comes from JOVE manual fig.7.2.
Receiver and software must form a continuous chain
The common RG-59 ends in an F-connector. The documented assembly uses an F-female ↔ BNC-female adapter and a short flexible BNC-male ↔ SMA-male lead, about 15 cm long, into the RSP1B. A USB 2.0 type B↔A cable connects the receiver to the computer and carries both data and power. A large heavy adapter directly on the small SMA socket would add unnecessary mechanical load. NASA receiver manual, pp.7–8, RSP1B datasheet.
The antenna coax is 75 Ω and the receiver input is 50 Ω. This is the documented connection, but a connector adapter does not make it perfectly impedance-matched. Some signal can be reflected; the exact response depends on all components. Keep Bias-T disabled for this passive antenna: it needs no power for an external amplifier through the coax.
As of 14 November 2026, the documentation for this article supports the following software path on Windows 11 x64, an x86-64 PC:
| Part | Function |
|---|---|
| SDRplay Hardware API 3.15 | Driver and RSP1B access |
| SDR Console 3.3 | Receiver control and digital data transfer through Data Repeater |
| SDRc2RSS | Conversion to spectral data in the format RSS requires |
| Radio-Sky Spectrograph, RSS | Display, saving and reopening spectrograms |
NASA’s 2025 manual installed the API through SDRuno. The manufacturer now marks SDRuno as end-of-life; SDR Console’s author permits a separate API installation instead, and the current API explicitly supports RSP1B. We use that separate path here. The old SDRplay2RSS for RSP1A does not replace SDRc2RSS in this configuration. Availability of Linux/macOS APIs also does not establish compatibility of the full Windows chain on those systems. SDRuno: current status, Hardware API, SDR Console: SDRplay connection, Console system requirements, stable Console release, RSS author’s documentation, p.68.
This checks documented compatibility, rather than reporting our own hardware run. Obtain software from its authors’ sites and verify its publisher before installation; the article does not suggest bypassing security warnings to run an unknown file. Once installed, one application should control the receiver: Console. Another open controller may occupy the device.
What travels from antenna to saved spectrogram
- Two dipoles → equal feed cablesRadio-frequency voltages · V
- Combiner → common cable → adaptersOne radio-frequency voltage · V
- RSP1B → USB → Hardware APIDigital I/Q samples · ADU
- SDR Console → Data RepeaterI/Q sample stream · ADU
- SDRc2RSS → local stream 127.0.0.1Spectral levels · relative units
- Radio-Sky Spectrograph → .sps fileUTC time, frequency, level and metadata
Test from the cable to the saved file
Faults are easier to locate when each step has a separate observable result. Do not begin troubleshooting by waiting for a Jovian burst: a working system may go a long time without seeing one.
- Before connecting the receiver: inspect assemblies and connectors. On an individual disconnected dipole feed, a multimeter between centre conductor and braid should show no short circuit. Do not extend that requirement to the entire chain through combiner or receiver: their circuitry may provide other DC paths.
- Check USB/API: select SDRplay V3 in Console and find the RSP1B itself. If it is absent, first examine USB cable, power, API and device ownership, rather than dipole length.
- Check the local digital path: enable Data Repeater, connect SDRc2RSS and check its RSS box. In RSS choose RTL Bridge/TCP with local address 127.0.0.1. That mode name does not introduce an RTL receiver into the system: its communication protocol is being used.
- Fix the observing mode: 8 MHz bandwidth with 16–24 MHz limits and 512 bridge channels. Disable automatic gain control AGC, noise reduction NR and noise blanking NB for astronomical recording: they can alter the variations of interest. Set gain so that strong interference does not overload the chain; someone else’s numerical settings are not your receiver’s calibration.
- Test antennas separately: with unchanged settings, compare a disconnected input, the first dipole, the second, and the normal combined connection. Mark switching times. Both separate branches should show a meaningful response; an increase after connection shows an external contribution without identifying its origin.
- Test recording: save a short .sps file, stop acquisition and reopen it. Do time, frequency and switching events agree? A moving display without a usable file is not a complete recording check.
This is a procedure for real equipment, not a list of hardware tests already performed by the author. Its basis is NASA receiver startup, pp.10–15, astronomical settings, antenna checks, §7.5–7.6.
Branch · With formulasOptional detail about frequency spacing and the limits of a digital spectrum.
512 channels do not imply 512 independent narrow measurements
If 8 MHz is divided uniformly among N=512 channels, the average frequency spacing is:
Δf_(grid) = (8 × (10)^(6))/(512) Hz = 15625 Hz.
That is approximately 15.625 kHz between grid elements, not automatically the effective width of each channel or the ability to resolve two lines separated by that amount. FFT window, overlap and averaging change the response. Save information about any window change alongside the data.
The antenna’s design frequency, 20.1 MHz, and the recording band’s midpoint, 20.0 MHz, are different parameters: (16+24)/2=20. Do not shift the limits to 16.1–24.1 simply to make these numbers match. Antenna resonance is not a rectangular filter; its response varies across the band. Check file-header labels against the actual operating mode rather than filling them from memory. NASA: 8 MHz and 512 channels, pp.11–12.
Calibration checks the scale; a palette changes appearance
Two operators can see equally bright bands despite different receiver gains. Conversely, the same recording can appear dark or bright after a colour-scale change. “Brighter” alone therefore does not mean a known physical power.
Calibration applies a known reference input through an appropriate calibrator and checks how the recording responds to its changing level. An arbitrary generator is unsuitable: its output must stay within the receiver’s input limits. NASA links a guide for the separate RJ DIY Cal 1, describing a Console/SDRc2RSS/RSS procedure. A link does not mean a calibrator is already in the kit or can be replaced by a colourful image.
The practical sequence is: check the chain and let it stabilize; apply the calibrator’s documented levels; set gain without saturation; record the transfer curve and settings; check repeatability; reconnect the antenna; keep calibration with the observation. Changing RF/IF gain or SDRc2RSS gain requires another scale check. RSS Color Offset and Color Gain control display mapping; changing a palette alone creates no physical reference. NASA calibration materials, Larry Dodd: Calibration Guide, 15 July 2024, pp.1–3.
Without a known reference input, useful relative recordings can still be made with fixed settings and honestly labelled units. Do not label them as absolute flux density in janskys or precisely measured sky temperature. Even proper electronics calibration does not automatically include all cable losses, frequency response and antenna-pattern effects.
Branch · With formulasOptional teaching model at one frequency; numbers are invented for calculation, not measured on an RSP1B.
Two known inputs define a simple linear scale
Suppose the recorded number y depends linearly on a reference input’s equivalent temperature T as y=aT+b. Here a is a scale factor and b an offset; they may differ by frequency. If known T₁ and T₂ produce y₁ and y₂:
a = (y_(2) − y_(1))/(T_(2) − T_(1)),; b = y_(1) − aT_(1),; T = (y − b)/(a).
In our hypothetical example, 100 K produces 25 units and 500 K produces 105. Thus a=0.2 units/K and b=5; a recording of 65 corresponds to 300 K at this model’s input. This explains the need for a known input, rather than calibrating a real station.
Once the chain saturates, the same maximum can represent many stronger inputs; an inverse formula cannot recover the lost information. Likewise, calibration at one frequency does not establish correct scaling in every channel. Equivalent input temperature and flux density from a particular celestial source require different additional antenna information.
A band reveals signal structure, not its identity
The following diagram contains no observational data. We set a background, raise it after a hypothetical antenna connection at 20 seconds, and add a persistent narrow interferer, a frequency sweep and a brief broadband enhancement. The example teaches how to read axes and distinguish features.
Synthetic example: antenna connection and three different features
t=20 s: Antenna connected. Scale · relative units: 0…12.
- A · fixed frequency
- B · changing frequency
- C · brief broadband event
Origin requires independent station timing and frequencies + source visibility + terrestrial-interference checks.
A horizontal line indicates a contribution at nearly constant frequency. A sloped line shows frequency changing with time. A nearly vertical broad stripe marks a fast event across much of the band. Terrestrial electronics also produces short broadband impulses; a silhouette alone does not identify the Sun or Jupiter. NASA’s real examples show radar sweeps and other interference alongside solar events. NASA: spectrograms and origin caveats.
A stronger conclusion requires comparison of UTC time, frequencies and event development with an independent station; check whether the source was above the horizon and within a useful part of the pattern, and whether the event fits other astronomical context. Agreement between two stations helps, but a shared distant terrestrial transmitter can also appear at two locations. A combination of evidence is needed, rather than “two screens prove it”.
Keep the original .sps file, timing and synchronization method, station coordinates, antenna orientation/height, cables, frequency limits, window, gains, calibration state and a log of local switching. A screenshot helps discussion, but repeat analysis needs the data and metadata themselves. RSS Help: Identity/Antenna, pp.27–28; timing and metadata, p.67; image comparisons, p.72.
A working station has several levels of evidence
A consistent build starts with two matching dipoles and feed cables. Correct dimensions, matching connection orientation and reliable mechanical assemblies provide the conditions for reception. The RSP1B and compatible software turn that input into a file that can be reopened and checked.
Different checks answer different questions: device detection establishes USB/API access; file playback checks recording; a changed level after antenna connection shows an external contribution; reproduction of a known reference input checks the scale. Only further analysis with independent data can support a celestial origin for an event.
Without equipment, one useful step is already possible: trace both branches to the combiner on the diagram and verify that the centre conductor connects to the same side of each dipole. With equipment, the next step is a systematic reference-based test, rather than waiting for a guaranteed discovery. A documented route lets you understand what you are building and what each result actually proves.