Bandwidth enhancement for thin substrate UHF RFID patch antenna

Most commercial UHF RFID patch antennas are either made from thick substrate (usually 1/10th of the wavelength) or has a stacked patch configuration to achieve the necessary operating bandwidth. This paper presents an unconventional and inexpensive way to enhance the impedance bandwidth of a thin substrate (1/35th of the wavelength) microstrip fed patch antenna. The 902–928 MHz RFID band was chosen for demonstration as it is the widest RFID bandwidth allotted by Federal Communications Commission (FCC). One major operational disadvantage of patch antenna is its very narrow frequency bandwidth (Balanis, 2012). Traditionally patch antenna’s bandwidth can be enhanced by insertion of slits and slots in both radiating patch and ground plane (Pandey and Shanmuganantham, 2016) or using a thicker substrate (Kovitz and Rahmat-Samii, 2014) or by stacking radiators over the driven patch (Liu et al., 2016). Addition of material such as increased substrate or stacked elements increases the cost and size of the antenna. Slotted radiating element and ground planes yield enhanced bandwidth at the cost of poor antenna efficiency and front to back ratio. Bandwidth and gain can also be enhanced with loaded parasitic elements next to the radiating patch (Cao and Hong, 2016) or by loaded loops over the radiating patch (da Costa et al., 2009). Though parasitic elements contribute in gain and impedance bandwidth enhancement of a linearly polarized antenna, a circularly polarized antenna’s axial ratio will get affected. The proposed bandwidth enhancement technique in this paper is unconventional and is focused on the feed rather the radiating patch.

Conventionally bandwidth of a patch antenna can be enhanced by offsetting the microstrip feed to one side of the patch (John et al., 2016) or using multiple feeds where the same patch is fed by two different locations with 180° phase shift (Koutinos et al., 2017). The antenna’s gain was compromised quite a bit in the former method while the antenna’s maximum radiation was not at the bore sight in the later, which means its directivity is altered. The proposed design in this paper would not alter the antenna’s radiation parameters such as beam-width, front to back ratio, directivity, etc., as the shape of the radiating patch remains undisturbed. Bandwidth can also be enhanced by differential feeding technique mentioned in Liu et al. (2017) and Wang et al. (2017). Antennas operate in different modes and bandwidth enhancement is a consequence of dual mode operation. Though the antenna design has a stable radiation pattern, the design required two input feed and not all application such as UHF RFID could afford sacrificing two ports for a single antenna. In addition to this, the antenna in these designs is not fed through a microstrip line but is coax-fed.

This means that the design is not low profile and antenna array realization will be complex. The proposed antenna uses a microstrip line feed, an antenna array realization is practical as the patch and the feed network will be in the same plane. The proposed antenna design is simple, as it does not use differential inputs. Electromagnetic coupling is also one among those traditional bandwidth enhancement techniques where coupled lines are used to feed the patch antenna (Van Wyk and Palmer, 2001). Antenna’s resonant frequency is deviated when this method is used, and H-plane radiation squint is observed due to unsymmetrical feed (Van Wyk and Palmer, 2001). The proposed antenna has a symmetrical feed as the patch antennas are edge fed. Defected ground structure (DGS) is an evolving technique to enhance the performance of patch antennas (Dash et al., 2016). The antenna’s ground plane is defected either symmetrically or asymmetrically to obtain a DGS. DGS creates a parallel-tuned circuit that offers multiband operation (Dash et al., 2016). As the ground plane is defected, the antenna’s radiation efficiency and its directivity are altered. The proposed antenna’s radiation parameters were not altered yet bandwidth enhancement is achieved. Unconventionally bandwidth enhancement was achieved by using a holey superstrate, described in (Asaadi and Sebak, 2017). The problem with this design is when this antenna is covered by a radome there is no guarantee that the enhanced bandwidth will remain unaffected. The proposed bandwidth enhancement technique is based on a single substrate design and does not incorporate a superstrate yet unconventional.

The proposed design can be distinguished from the traditional bandwidth enhancement techniques due to the following reasons:

No significant addition of material such as increased substrate or addition of a superstrate or stacked elements. Size and cost of the antenna is not increased therefore.

A simple microstrip fed rectangular patch antenna is used in this bandwidth enhancement experimentation. The patch antenna operates only in one mode and thus it is linearly polarized. The antenna is designed in polyethylene substrate. This substrate is preferred over other traditional substrates (such as air) in RFID applications for physical robustness. HDPE’s dielectric constant (ε_r) is 2.5 and loss tangent (tanδ) is 0.0005 (Przesmycki et al., 2012). The patch antenna’s length and width are calculated through the following steps (Balanis, 2012).

Step 1: Patch antenna’s width (W) is calculated using the free-space velocity of light (ν₀), resonant frequency (f_r), and dielectric constant (ε_r) (Balanis, 2012), shown in the following equation: (1) W = ( ν 0 2 f r ) ( 2 ε r + 1 )

Step 2: The effective dielectric constant (ε_reff) is calculated using the formula shown in the following equation (Balanis, 2012). Substrate thickness (h) is 6 mm which is a thin substrate as it is 1/35th of the wavelength in polyethylene substrate: (2) ℰ r e f f = ε r + 1 2 + ε r + 1 2 ( 1 + 12 h W ) − ( 1 2 )

Step 3: The free-space velocity of light (ν₀), resonant frequency (f_r), and effective dielectric constant (ε_reff) are used to find the effective length (L_eff) (Balanis, 2012): (3) L e f f = ( ν 0 2 f r ( ε reff ) )

Step 4: Length extension (ΔL) is calculated using the width of the patch antenna (W), height or the thickness of the dielectric substrate (h), and the effective dielectric constant (ε_reff) (Balanis, 2012), shown in the following equation: (4) Δ L = 0.412 h [ ( ε r e f f + 0.3 ) ( W h + 0.264 ) ( ε r e f f − 0.258 ) ( W h + 0.8 ) ]

Step 5: Actual length of the patch is found using the effective length (L_eff) and the length extension (ΔL) (Balanis, 2012), shown in the following equation: (5) L = L e f f − 2 Δ L

Based on the calculation, patch antenna’s length and width is found to be 100.45 mm and 126.9 mm for 915 MHz resonant frequency. The patch antenna is fed by a microstrip line. The microstrip line is also a quarter wave transformer transforming the input impedance 50 Ω to the patch antenna’s edge impedance 137.8 Ω. The quarter wave transformer is 7.08 mm wide and 60 mm long (see Fig. 1(a)).

Figure. 1

Traditional bandwidth enhancement techniques are based on impedance matching feed network (Pues and Van de Capelle, 1989) where the antenna’s edge impedance is matched to the input impedance through tuned stubs, etc. The proposed bandwidth enhancement is based on intentional impedance mismatch at the patch antenna’s point of excitation. The radiating rectangular patch antenna’s shape and size are unaltered. Thus, intentional impedance mismatched excitation was applied on the same rectangular patch antenna geometry. The antenna designed in Figure 1(b) has an abrupt impedance mismatch at the edge of the patch antenna. The input impedance 50 Ω was carried all the way through a 50 Ω microstrip line and terminated at the edge of the patch antenna. The 50 Ω microstrip line was made quarter wavelength long to maintain the consistency with the matched patch antenna shown in Figure 1(a). The 50 Ω microstrip line is −64% of the patch antenna’s actual edge impedance, 137.8 Ω. Antennas designed in Figure 1(c) and (d) has intentional impedance mismatches in its quarter wave transformer’s impedances. The antenna shown in Figure 1(c) uses a quarter wave transformer whose impedance is 68.33 Ω. This transformer will be ideal to transform the 50 Ω input impedance to 93.5 Ω load impedance. As the designed patch antenna’s edge impedance (load impedance) is 137.8 Ω, the quarter wave transformer designed for a −32% edge impedance produces an impedance mismatch. The antenna shown in Figure 1(d) uses a quarter wave transformer whose impedance is 95.50 Ω. quarter wave transformer is suitable to transform the 50 Ω input impedance to 181.9 Ω. As the designed patch antenna’s edge impedance (load impedance) is 137.8 Ω, the quarter wave transformer designed for +32% edge impedance produces an impedance mismatch. The designed antenna was fabricated (see Fig. 1) in polyethylene substrate. The substrate is 6 mm thick. Radiating patch is made from strips of copper tape. Ground plane is made from tinned steel sheet. The conductive layers are adhered to the dielectric substrate using non-conductive acrylic adhesive. A SMA flange mount connector is used for excitation. In total, +64% mismatch (136.7 Ω transmission line) is not implemented in the study because it was practically not possible to fabricate.

The fabricated antennas are measured using TR1300/1 vector network analyzer for its return loss. The antenna’s path loss was measured across the frequency using a receiving dipole antenna. Fabricated antenna’s gain is found by comparative method (Przesmycki et al., 2012). Using a reference patch antenna (whose gain is known), −29.5 dB path loss measured is translated to 0 dBi antenna gain. Figure 2 shows the return loss and path loss measurements for all the fabricated antennas. The impedance matched patch antenna resonated at 915.0 MHz with a −26.12 dB return loss. The antenna’s −5 dB return loss bandwidth is 79.13 MHz. The return loss at low (902 MHz) and high (928 MHz) RFID frequencies were −12.64 dB and −12.02 dB, respectively (see Fig. 2). Most fixed RFID antennas bear long coaxial cables when installed in sites and the reflections happening within the cable gets attenuated (Walraven, 2006). Moreover, antennas with good SWR will undergo SWR deterioration due to the way assets are tracked in RFID applications (assets sit right on top of the antenna in a point of sale, shelving systems, etc.). Modern RFID interrogators have an embedded SWR tuning module and thus they can handle reflected power by the antenna. So, the bandwidth of the antenna is measured at −5 dB return loss. The +32% and −64% impedance mismatched antennas resonated at 918.0 MHz and 928.7 MHz, respectively. This resonance is higher than the intended resonant frequency. This is because the quarter wave microstrip lines were not only feeding the antennas but were also acting as tuning stubs, thus changing the F_c (Tahir and Brooker). The −32% impedance mismatched antenna was found to be the optimal design as it did not induce a shift in resonant frequency. The antenna resonated at 915 MHz with a return loss of −14.27 dB. Table 1 lists the return loss, path loss and the gain parameters for center, low and high frequency of different designs, respectively. The −32% mismatched antenna’s −5dB return loss bandwidth is 88.98 MHz. The return loss at low (902 MHz) and high (928 MHz) RFID frequencies were −11.44 dB and −11.55 dB, respectively (see Fig. 2). This shows that with a −32% intentional impedance mismatch, the impedance bandwidth is enhanced by 12.5% (9.85 MHz) without significantly altering the return loss at the operating frequencies.

Figure. 2

The +64% mismatch exclusion from the antenna design can also be rationalized through the results obtained from +32% mismatched antenna, as the antenna resonates at a different resonant frequency than the intended operating frequency, 915 MHz.

The linear gain of fabricated antennas at their resonant frequencies; F_c remained unaltered between matched and the −32% mismatched antenna whereas the other two designs has their Fc shifted to higher frequencies namely, 928.7 and 918 MHz, respectively. The gain of the −32% mismatched antenna is 0.7 dB less than that of the matched antenna at 915 MHz (F_c). At −5dB bandwidth points, the −32% mismatched antenna has 0.07 dB (at 849.58 MHz) and 0.04 dB (at 955.92 MHz) less gain compared to the −5dB bandwidth points of the matched antenna (at 857.32 and 951.93 MHz, respectively). Antenna gain at the edge of the band is not deteriorated significantly. Figure 3 shows the azimuth and elevation radiation pattern of the −32% mismatched antenna. Gain was ~3 dB high at 915 MHz compared to the low (870.8 MHz) and high (959.7 MHz) side of the band and thus their beam is smaller than 915 MHz’s pattern. The front to back ratio is −10 dB for 915 MHz. The front to back ratio can be improved by using a larger ground plane (Balanis, 2012). Table 2 shows the measured beam-width for low, mid and high frequency in both azimuth and elevation planes. Beam-width remained the same regardless of the frequency. Figure 3 as well reveals that the antenna’s directivity is not spoiled. Peak radiation is at the antenna’s bore-sight. Both matched and −32% mismatched antennas are tested for UHF RFID applications. Antennas can be deployed in portals, shelves, benchtops, etc., where assets may directly be placed over the antennas for inventory tracking. When an object is in contact with the antenna’s radiating element, the antenna’s parameters such as return loss and gain may get affected. Common objects that come in contact with the reader antennas are plastic, glass, and wood (see Fig. 3). Figure 4 shows the frequency sensitivity of matched and mismatched antennas when objects are in contact with the antennas. The return loss and the path loss of the matched antenna varied quite a bit compared to the intentional mismatched antenna. The frequency shift of a matched antenna varied upto ~30 MHz where as for the −32% mismatched antenna it was limited to ~13 MHz. Results in Figure 5 also show that the antenna’s gain is insensitive when items contact the −32% mismatched antenna.

Figure. 3

Figure. 4

Figure. 5

The antenna is tested with “Impinj speedway r420” UHF RFID reader for FCC frequencies (Fig. 4). The items were tagged using an “Alien Squiggle Higgs-4” tag. Table 3 shows the tag’s return signal strength indicator (RSSI) when the antenna is powered by the Impinj reader (with no cables) at 10, 20, and 30 dBm levels. When the tag was tested at full power (30 dBm) in free space, it reported a highest RSSI value of −47 dB at 928 MHz. When items were placed over the radiating patch, the reported RSSI for different materials remained similar for different frequencies. This is because the antenna’s gain remained similar across the band.

This paper presented a novel and unconventional bandwidth enhancement technique for RFID applications. This single substrate, less complicated technique enhances bandwidth without substantially degrading other antenna parameters such as gain, directivity, radiation pattern, etc. This impedance mismatch technique is frequency independent as the microstrip line feed’s impedance stays constant across the frequencies for a given substrate thickness and its dielectric constant. This method is recommended for transceivers like UHF RFID readers that can handle reflected power. As part of future work, the same technique will be applied for circularly polarized antenna and antenna array. A circular polarization antenna design using intentional mismatch will be challenging because of the two modes existing in the same radiator.