Cellphone radiation detector

Created: May 2019. Minor edits: Jan. 2023

The devices and cellular bands mentioned in this article are somewhat dated as this project is from the year 2000. However the material and design techniques described are still useful today.

Introduction

Semiconductor manufacturers guarantee the performance of their low-noise transistors by measuring the noise figure (NF) of every unit. To closely simulate customer applications, they may measure the transistors at the intended operating frequency. As the majority of these transistors go into cellular phones, the test frequencies are the same as that of the GSM and PCN bands. However, this choice of test frequencies makes the device testing susceptible to interference from nearby users of cellular phones. Although, cellular phones are usually banned from the production floor, this rule is often violated. As a result, over rejection due to interference is a pressing issue. Hence, the need for a low cost and portable detector of cellphone radiation.

Overall function blocks

Two versions of the interference detector were created to cover the cellular bands that exist at the locality. The first version, operating at 900 MHz, detects emission from GSM (890~915 MHz uplink) and ETACS handsets (871~904 MHz uplink). The second version, operating at 1.8 GHz, identifies the presence of PCN band interference (1805~1880 MHz uplink). The differences between the two versions are in the aerials and the diode detectors only.

Fig 1a: Cellular radiation detector's block diagram

Fig. 1b: enclosure front (top) and opened rear (bottom)

Aerial

The aerial is a simple quarter wave monopole over a ground plane consisting of 4 radial wires (fig. 2a). Due to the absence of a test range, the gain of the aerial cannot be measured. It is anticipated that the gain of the monopole should be close to the 2.14 dBi value that is usually quoted by standard references [1]. Two aerials were fabricated for 900 MHz and 1.8 GHz. The aerials were directly connected to the SMA female connector on top of the detector enclosure.

Fig. 2a: details of aerial

Both the monopole and radials are trimmed to position the matching at the respective target frequencies. The measured performances of the fabricated aerials are shown in fig. 2 and fig. 3.

Fig. 2: 900 MHz aerial return loss and impedance

Fig. 3: 1.8 GHz aerial return loss and impedance

RF detector - 900 MHz

A voltage doubler configuration was chosen for the detector because of improved sensitivity and lower input impedance [2]. To eliminate the need for external DC bias, a low barrier Schottky, the HSMS-2852, is chosen as the rectifying element. The HSMS-2852 incorporates two closely matched diodes in one SOT-23 package. However, it must be noted that the HSMS-285 family has poor consistency of key parameters above 900 MHz. So, board to board tuning is required to maintain the same centre frequency.

Fig. 4: Voltage doubling diode detector for 900 MHz

Fig. 5: 900 MHz detector input impedance (top) and return loss (bottom). The good matching points to the accuracy of the input matching network

Fig. 6: 900 MHz detector output vs. frequency. The peak response is correctly centered at the target frequency

Fig. 7: Detection sensitivity of the 900 MHz voltage doubler

The ideal voltage sensitivity of a zero bias detector, g, is given by: -

where B is the current sensitivity (20 A/W), VT is the thermal voltage (approx. 0.026V at room temperature), and, Is is the diode saturation current. However, the voltage sensitivity of a practical detector is degraded by the following factors: - diode parasitics (g1), load resistance (g2), and matching network loss (g3) [3]. Hence, the actual voltage sensitivity of the 900 MHz detector, g3, can be calculated from the slope of the graph: -

RF detector - 1.8 GHz

The 1.8 GHz detector is conceptually similar to the 900 MHz version, except for the matching network's component values (fig. 8).

Fig. 8: 1.8 GHz detector circuit (top) and assembly photo (bottom)

Fig. 9 Equivalent circuit model of 1.8 GHz detector

Fig. 10: 1.8 GHz detector input impedance and return loss. The matching is centered at ~1820 MHz. It is not possible to match at the exact frequency due to the use of standard component values

Fig. 11: Simulated & measured 1.8 GHz detector output vs. frequency. The peak response is slightly offset above 1.8 GHz. Good agreement between simulated and measured results

The voltage sensitivity of the 1.8 GHz detector, g3, is given by: -

Fig. 12: Detection sensitivity of the 1.8 GHz detector. Good agreement between simulated and measured results

Voltage comparator and LED driver

The output from the diode detector is very low (in the milivolt region). This effectively precludes the use of most common voltage comparators (e.g. the ubiquitous LM311). To work directly in conjunction with the detector, the voltage comparator must have a common mode input range, which include ground. The comparator's input offset voltage is another important parameter to consider if detection of signal below -30 dBm is required. The LM393/2903 series (Voffset < 5 mV) is suitable for normal interference detection, while the LM393A (Voffset < 2 mV) is recommended for precision low-level voltage comparison [5]

Fig. 13: Comparator and output driver

Overall system sensitivity

The sensitivity of the detector is dependent on the threshold adjustment of the comparator and the antenna factor. The antenna factor, which is related to the gain and the operating wavelength, is given by: -

where AF is the antenna factor as a voltage ratio, "lambda" is wavelength in meter, and G is the gain as a power ratio.

The monopole over a ground plane has a gain of 2.14 dBi or 1.64 times over an isopole. Therefore, the antenna factor, in dB, at 900 MHz is: -

[edit] An alternate form of the above equation may ease calculation [6]:

AF (dB) = 20 log f(MHz) - G (dB) - 29.8 dB

And a monopole that is resonant at 1.8 GHz will have an antenna factor of: -

The voltage comparators in the interference detectors are adjusted to trigger when the input to the diode detector, Pin, is equal or greater than -50 dBm (corresponding to a Vin of 57 dBuV into 50 ohms).

[edit] The equation below eases conversion from dBm to dBuV:

dBuV = dBm + 107 dB = (-50 + 107) dBuV = 57 dBuV

So, the lowest electric field strength, E, which the 900 MHz interference detector will respond to, is given by: -

E (dBuV/m) = AF (dB) + Vin (dBuV)= 27.16 + 57 dBuV = 84.16 dBuV/m

And for the 1.8 GHz interference detector, the weakest electric field strength it can respond to is: -

E (dBuV/m) = AF (dB) + Vin (dBuV)= 33.16 + 57 dBuV = 90.16 dBuV/m

Future improvements

If an array of interference detectors can be laid out in a grid like fashion, and then linked to a PC, the location of the interfering source can be made rapidly. The interfering source will trigger the detectors that are nearest to it. The pattern of detection can then be superimposed on the building plan and displayed by the PC.

References

[1] Hall, G. (editor), The ARRL Antenna Handbook, 16th. Edition.

[2] Waugh, R., "Designing Detectors for RF/ID Tags", RF Expo, San Diego, February 1995.

[3] Buted, R. R., "Zero Bias Detector Diodes for the RF/ID Market", Hewlett Packard Journal, December 1995.

[4] Hewlett Packard Application Note 1088, "Designing the Virtual Battery".

[5] Data sheet for LM193/293/393/2903 comparator, SGS Thomson Industrial Standard Analogue IC Databook,1st. ed., 1989

[6] Hewlett Packard application note 150-10, "Spectrum analysis...field strength measurement", Sep. 1976