A Low-Noise SDR-Based Receiver for the 21 cm Neutral-Hydrogen Line

Abstract : after the article on the pyramidal horn antenna : Horn Antenna for the 21cm Neutral-Hydrogen Line, we now pass to describe the design, construction and analysis of the hardware system for receiving the 21cm radio-emission of neutral hydrogen.


What we intend to create is a receiver centered on the frequency of 1420 MHz, with a bandwidth of about 50 MHz. The signal we have to receive is very weak, we expect an overall gain of about 50dB, and is practically indistinguishable from electronic noise, for this reason it is necessary to use high gains, while keeping the noise as low as possible in order to maximize the signal-to-noise ratio. The first component will then be a low noise amplifier (LNA) with a noise figure <1dB.
After the amplification stage it will be necessary to insert a band-pass filter centered at 1420 MHz in order to select only the portion of the spectrum that interests us.
The actual receiver will be a software defined radio (SDR) module connected to the computer’s USB port. The SDR module must have good frequency stability over time in order to obtain well-defined and repeatable signal spectra.
These constraints led us to outline the following receiver design.

The components we have identified are the following:

  • LNA amplifier, prefiltered at 1420MHz, gain=30-40dB, FN<1dB
  • Wide band amplifier, gain = 10-20dB
  • Band pass filter (passband of about 50MHz) centered at 1420MHz

1420 MHz Prefiltered LNA

The first component of the signal reception chain is the low noise amplifier (LNA), it is the most critical and most important component because its noise figure (NF) determines the noise figure of the entire apparatus. It is therefore important to choose a good LNA with a low noise factor in order to maximize the S/N ratio.
In our receiver we have adopted a NooElec amplifier specially designed for receiving the frequency of 1420MHz. The NooElec SAWbird+ H1m is an Ultra Low Noise (LNA) amplifier module with premium cascaded SAW filter. The amplifier, centered at the 1420 MHz frequency, is designed to receive the 21cm neutral hydrogen line. The image below shows the amplifier in its aluminum case with the two SMA connectors and the USB socket for power supply.

The following image, taken directly from the NooElec documentation, shows the amplifier diagram. The filter is connected between the two amplification stages. It should be noted that the RF output is connected directly to the power supply, this means that the amplifier can also be powered via a bias-T directly from the receiver stage downstream of the amplifier. For convenience we have powered the component via the USB socket, but in doing so there is a direct voltage (DC) of about 5V on the RF output which must be filtered with a suitable DC-Block (or with a good capacitor).

Technical Data :

  • +40dB of RF gain at 1420MHz
  • 65MHz 3dB bandwidth
  • 0.8dB noise figure at 1420MHz
  • 50Ω gain block
  • +3.3V-5V single supply
  • 122mA Current Draw
The following table and diagram show the frequency response of the amplifier. You can see how the passband goes from a minimum of 1380MHz up to a maximum of 1440MHz, with a gain of 40dB. Outside the band, the gain is reduced to -20dB.
Frequency (MHz) Gain (dB)
1300 -7
1420 40
1500 0

Wideband Amplifier

This unit HAB-FLTNOSAW built by UPUTRONICS is a preamp designed to go between a software defined radio receiver and an antenna. The LNA used inside is a MiniCircuits PSA4-5043. This particular model has the SAW filter removed to cover the 0.1MHz to 4GHz. There are 2 options for powering the unit : either by the USB header or via bias-tee. Devices such as the Airspy can enable bias-tee and power the device. Alternatively any mini USB cable can be used to power the device. We chose to power the unit via USB line.

Technical Data :
Gain 24db @ 100MHz -> 15.2db @ 1415MHz
NF 0.75dB
Supply Voltage USB or Bias tee 5V

In the images below we show the unit and its frequency response.

Frequency (MHz) Gain (dB)
1300 16
1420 15
1500 14

1420 MHz Band Pass Filter

This filter is dedicated to amateur radioastronomers interested in the hydrogen line observations. It uses the TA2494A SAW component and measures only 50 x 10mm. It features edge pads for an easy soldering of a RF shield. Insertion loss is typically less than 3.5dB and bandwith 80MHz.

Technical Data :
Center Frequency 1420MHz
Usable Bandpass 1380-1460MHz
Insertion Loss, 1380 to 1460 MHz 3.5dB
Amplitude Ripple, 1380 to 1460 MHz 1.0 dBpp
VSWR, 1380  to 1420 MHz 1.9:1
Rejection referenced to 0dB :
DC to 1300 MHz 28dB
1550 to 3000 MHz 30dB
Impedance 50Ω
Maximum Input Power Level 10 dBm

In the images below we show the unit and its frequency response. We have soldered two wires between the SMA female headers and we wrapped the filter with aluminum tape in order to shield the filter.

Frequency (MHz) Gain (dB)
1300 -50
1420 -3.5
1500 -50

Airspy R2 SDR  Receiver

From the manufacturer’s site : The Airspy R2 sets a new level of performance in receiving the VHF and UHF bands thanks to its low-IF architecture based on the Rafael Micro R820T2 chip and a high quality 12-bit Oversampling ADC and state-of-the-art DSP. In Oversampling mode, the Airspy R2 applies analog RF and IF filters to the signal path and increases the resolution up to 16 bits using software decimation. Coverage can be extended to HF bands via the up-converter companion SpyVerter (not used by us). Airspy R2 is 100% compatible with all existing software, including the SDR # scan standard, but also with a number of popular software-defined radio applications such as SDR-Radio, HDSDR, GQRX and GNU Radio. The stability and precision of the clock for the local oscillator, given at 0.5ppm, is also important for our application.

Key Features of the AirSpy SDR Receiver :
● Continuous 24 – 1700 MHz native RX range, down to DC with the SpyVerter option (not used)
● 3.5 dB NF between 42 and 1002 MHz
● Maximum RF input of +10 dBm
● Tracking RF filters
● 35dBm IIP3 RF front end
12bit ADC @ 20 MSPS (10.4 ENOB, 70dB SNR, 95dB SFDR)
● 10MSPS IQ output
0.5 ppm high precision, low phase noise clock
● 10 MHz panoramic spectrum view with up to 9 MHz alias/image free
No IQ imbalance, DC offset or 1/F noise at the center of the spectrum1 x RF Input
● 4.5v software switched Bias-Tee to power LNAs and up/down-converters (not used)
● Operating temperature: -10°C to 40°C

In the configuration of the device (done through the osmocom driver in GNU radio) the RF gain is set to 0 (default setting), while the IF and BB gains are each set to 10 dB. These very low gain values ​​show the effectiveness of the components placed upstream of the receiver : from the antenna to the LNA and Wideband amplifiers. The bias-T option is also disabled.

The Complete Receiver

All components of the receiver have been placed in a sealed box, lined internally with aluminum tape in order to increase the shield against RF interference. As you can see, all the modules are equipped with a metal container so in this way we have created a sort of double shielding. The antenna is connected directly to the LNA module and, in cascade, all the other modules are connected up to the Airspy SDR receiver. For the connection between the modules we used straight and right-angled SMA male-male and male-female connectors and a length of cable. The power supply of the amplifiers is obtained via USB from an externally placed “power bank” battery, while the Airspy receiver is powered by the PC via USB. For the connection to the PC we used a USB cable with extension, the laptop PC is used with the battery disconnected from the power supply : we have taken all the precautions to minimize noise and RF interference. The image below shows the receiver from the antenna to the Airspy module.

Analysis of the Receiving system


For each of the components we add the gain :

Frequency (MHz) Gain (dB)
1300 -7+16-50 = -41
1420 40+15-3.5 = 51.5
1500 0+14-50 = -36

We see how the gain at our frequency has the value of about 50dB while the adjacent frequencies are considerably reduced by the filtering operation. Given the rather high gain it will be necessary to be careful not to work in saturation regime. We also note that the above does not take into account the “gain” of the antenna and its further filtering operation with respect to the other frequencies (the antenna works as a high pass filter).

Naturally, the contribution of the SDR receiver must be added to these values, in which the RF gain, the IF gain and the BB gain can be configured. In our case we set the RF gain equal to 0, while the other two parameters we set them to 10dB.

Noise Figure

For the noise figure, the Friis formula must be used :

Where the values ​​of F and G are not expressed in dB but as ratios.

G1 = 40db = 10000
F1 = 0.8dB = 1.2
F2 = 0.75dB = 1.2

Ftotal = F1 + (F2 – 1) / G1 =1.2 + (1.2-1)/10000 ≅ 1.2 ⇒ Ftotal = 0.8dB

It can be seen that the noise figure is practically equal to the value of the noise figure of the LNA. We therefore understand the importance of the first stage in the RF signal amplification chain. The noise contribution of the SDR receiver must also be added to the RF part, which is given equal to 3.5dB, however, given the high gain of the RF chain this contribution is negligible.

With these data we can also calculate the equivalent noise temperature Te of our receiver. The noise temperature is a theoretical concept and refers to the noise that would be generated by a resistor brought to temperature Te. The temperature Te is related to the noise figure by the following relationship :

Te = (F-1)*To

Where To is the reference temperature of 290°K. For our system we obtain Te = 58°K. This is a somewhat optimistic evaluation, probably the real value is little higher.


The Intenet contains numerous examples of antennas and receivers for the emission of neutral hydrogen at 21 cm. At the following link there is the description of an excellent similar project : probe-the-galaxy-on-a-shoestring-with-this-diy-hydrogen-line-telescope, and this is the related documentation : Hydrogen Line Project Documentation. A site very rich in information (I would say indispensable ..) is the following DSPIRA

Next Step

the next step consists in setting up the GNU radio software part for the reception, recording and analysis of radio data : GNURadio Software for 21cm Neutral-Hydrogen Line

If you liked this post you can share it on the “social” Facebook, Twitter or LinkedIn with the buttons below. This way you can help us! Thank you !


If you like this site and if you want to contribute to the development of the activities you can make a donation, thank you !

Check Also

KC761B: the new Gamma Spectrometer from DEEPACE

Abstract: in this article, we present an interesting new apparatus dedicated to gamma spectrometry and dosimetry measurements. It is a device based on a CsI(Tl) scintillator coupled to solid-state photomultipliers: SiPM. In addition to the scintillation sensor, the instrument has a PIN diode sensitive to beta radiation.