Today, a Global Navigation Satellite System (GNSS) is much more than GPS, which was introduced for civilian use more than a decade ago. Nations around the world are working on their own navigation satellite systems for strategic reasons and also to offer improved user experience. Today, two GNSS systems are operational: the United States GPS and the Russian GLONASS. The Galileo positioning system being developed by the European Union is expected to be functional by 2014 and Chinese COMPASS is also expected to follow.
From a civilian usage point, additional systems added to GNSS bring with them the advantages of increased satellite signal reception, increased coverage, higher precision and the facility for additional features such as search and rescue (SAR). This article considers only GPS and GLONASS and Table 1 summarizes some the basic GPS and GLONASS characteristics.
Why the GNSS Front-endis important
Consolidation of different wireless connectivity features and the advent of GLONASS on today’s handheld devices pose challenges on the GNSS receiver design at two levels: to enable the GNSS solution to work uninterruptedly in the vicinity of high power GSM/EDGE/UMTS/LTE signals active on the same device and to design a broadband solution covering the complete frequency range required by the two systems together.
To achieve this, cutting edge-design and technology capabilities are required to offer these features:
- High linearity to avoid interference of high power cellular signals
- Broadband design to cover the frequency range from 1575 to 1605 MHz
- Gain optimization to achieve state-of-the-art sensitivity
- Low noise figure to achieve state-of-the-art sensitivity
- High integration to have a small form factor and excellent cost position.
Although many current GNSS chipsets rule out the need for an external LNA and filters, this article offers a detailed guide through the various challenges faced by the system designer and explains the necessity of a high performance RF front-end to achieve hassle-free user experience.
Figure 1 An application schematic of a GNSS front-end.
GNSS RF Front-End Configurations
Typically, a GNSS front-end consists of an antenna, a pre-filter, a LNA and a post-filter. After this chain, the received signal is fed into the GNSS receiver IC. Figure 1 is an application schematic of a GNSS front-end. Today we can find different integration levels of GNSS front-end modules that suit different phone or personal navigation device (PND) configurations.
These include a pre-filter +LNA+post-filter, which offers highest integration level thus can save PCB space, reduce design complexity and assembly costs, and a pre-filter+LNA, where the module can be placed very close to the antenna to minimize losses, at the same time a discrete post-filter can be placed in front of the GNSS IC. If the GNSS IC requires either single ended input or differential input, only the discrete post-filter needs to change – the GNSS module can be the same for the whole product portfolio, and a LNA+post-filter. This configuration enables the choice of a discrete pre-filter with different rejection levels for different designs, if the pre-filter is already integrated into the antenna switch module.
Design Challengesand Solutions
Sensitivity: GNSS satellite signals are transmitted from orbits 20,000 km away from the earth’s surface. With typical antenna gain, the received power at the GNSS device can be calculated as given in Table 2.1
The receive power of -130 dBm is below the thermal noise level for a bandwidth of 10 MHz. Conventional GNSS receivers integrate the received GPS signals for 1 ms. This results in the ability to acquire and track signals down to around the -130 dBm level. High sensitivity GPS receivers are able to integrate the incoming signals for up to 1000 times longer than this and, therefore, can acquire signals up to 1000 times weaker. State-of-the-art high sensitivity GPS receivers are expected to track signals down to levels approaching -160 dBm.2 The sensitivity of the GPS receiver can be significantly enhanced with a very low noise RF front-end.
Noise Figure: The noise figure of the front-end directly influences the sensitivity of the receiver, as well as the time-to-first-fix (TTFF) and time-to-subsequent-fix (TTSF). These parameters are directly visible to the end-user and, therefore, are a major focus for design. The noise factor, F, of a GNSS front-end, according to Friis’s formula for a cascade of stages, each with its own noise factor F and gain G is given by:
where ‘pre’ denotes pre-LNA filter and ‘post’ denotes post-LNA filter.
The noise figure of the GNSS front-end is given by:
The noise figure of a filter is equal to the insertion loss (IL). As can be seen from the formula above, IL of the pre-filter is directly contributing to the total noise figure and therefore needs to be minimized. However, IL and Out-of-Band (OoB) rejection is a tradeoff in filter design and, therefore, impacts other design constraints like jammer performance of the GNSS module, which will be explained later. The noise figure of the LNA directly influences the system noise figure (Gpre ~ 1), whereas the post-filter being located after the amplifying stage does not contribute to the total noise figure (NF).
Typical insertion loss of a pre-filter can be in the range of 0.5 to 1.1 dB. State-of-the-art LNAs show a noise figure as good as 0.6 dB. The noise figure contribution of the post-filter is typically lower than 0.05 dB – here the tradeoff winner is clearly OoB rejection. In fact, the matching network in front of the LNA can also contribute up to 0.1 dB of noise figure depending on the quality factor (Q) of the external components used.
To summarize, the pre-filter and LNA define the noise figure of a GNSS module. There are modules with a noise figure as low as 1.1 dB (low OoB rejection) and as high as 2.0 dB (high OoB rejection).
Figure 2 Interference from mobile path to GNSS path.
Working near Jammer Signals
GNSS modules with the lowest noise figure satisfy one necessary selection criterion, but this does not necessarily guarantee good sensitivity and TTFF. Another concern regarding low power satellite signals is the presence of high power jammers in the GNSS device. Today’s highly integrated smart phones are subject to a variety of wireless links that are transmitting or receiving signals with high power like GSM, 900 MHz, UMTS, 1.8 to 2.2 GHz, Bluetooth and WLAN, 2.4 GHz and WiMAX, 2.5 to 2.7 GHz. Figure 2 shows a simplified block diagram of a mobile phone with the main path (GSM, UMTS) and the GNSS path.
If the strength of these signals is sufficiently high, it can drive the GNSS LNA into saturation and the gain in the GNSS band is lowered. The lower gain and increased total noise figure of the system is called desensitization.3 Depending on the isolation conditions in the phone a certain OoB compression point is required. The built-in GSM power amplifier can transmit power levels up to 33 dBm.
If we assume an antenna isolation of 10 dB between main antenna and GNSS antenna, we can directly derive the necessary GNSS module compression point at 900 MHz to be P1dB_900 = 33 dBm – 10 dB = 23 dBm. The GNSS LNA cannot achieve this compression point performance and, therefore, the pre-filter needs to attenuate the 900 MHz signal. Assuming the LNA has a compression point of 3 dBm at 900 MHz the pre-filter needs to have a rejection of minimum 20 dB at 900 MHz. At this point, it should be mentioned again that in filter design the IL and OoB rejection is a tradeoff (see Noise Figure section).
The second gain stage in the chain is the GNSS IC, which has a compression point of ~ -40 dBm and is much lower compared to a GNSS module. Here the post-filter comes into play, which should attenuate the jammer by 40 to 60 dB. However, desensitization is not only caused by driving the gain stages into saturation. Much before compression, an increase of noise figure can be noticed in many LNAs in the presence of a jammer.
Figure 3 Desensitization performance of Infineon's GNSS modules:noise over 1710 MHz jammer power.
Figure 3 shows the desensitization performance of a GNSS module. The jammer, in this case, is at a frequency of 1710 MHz. The acceptable noise figure increase here is 0.1 dB, which is shown as a black line. The jammer, in this case, is at the frequency of 1710 MHz. The acceptable noise figure increase here is 0.1 dB shown, as black line. The newest generation of modules is hitting this line beyond the 20 dBm jammer power level. We can see that this module is hitting the black line a little earlier, at +17 dBm jammer level.
One more detailed study that illustrates the everlasting demand for more wireless bandwidth and higher data rates in mobile phones should be discussed. Just recently the Federal Communications Commission (FCC) gave conditional approval to the use of the L-band for terrestrial LTE applications. The downlink frequency is 1525 to 1559 MHz, which is directly neighboring the frequencies of GPS and GLONASS. The base station will transmit a maximum power of 62 dBm equivalent isotropically radiated power (EIRP).
Depending on the base station antenna elevation and tilt angle, a jammer power of up to -20 dBm can be received at the GNSS antenna in close proximity (~200 m).4 The GNSS module itself will withstand that jammer level but it must protect the GNSS IC against the jammer. If we again assume a maximum allowed jammer level of -40 dBm at the GNSS IC and 20 dB gain of the GNSS LNA, we can calculate the necessary rejection of pre-filter + post-filter together:
This requirement is not fulfilled for existing GNSS equipment and it is a very hard requirement for new GNSS modules, due to its proximity to the GNSS frequencies.
Although there are no disturbing signals inherently in the GNSS band, due to nonlinearity effects in the LNA as well as in the filter, there are mixing products that can exactly fall in the GNSS band. Both 2nd order and 3rd order mixing products are relevant and should be taken into account. Table 3 is a list of user cases that are critical in a mobile phone.
The OoB intercept point 3rd order (IP3) and OoB intercept point 2nd order (IP2) for the user cases needs to be optimized. The mixing can take place in each stage (pre-filter, GNSS LNA, post-filter, GNSS IC) due to nonlinear effects. The passive filters normally show higher linearity compared to the active devices. However, the pre-filter gets the biggest jammer level and can, therefore, also be the source of the mixing signal. The pre-filter OoB rejection can improve linearity if the dominant mixing already happens at the GNSS LNA. To improve OoB IP3 of the GNSS module, a high OoB rejection pre-filter should be used to lower the jammer power level at the LNA and, therefore, the power level of the mixing product. The post-filter can only limit the jammer power level to the GNSS IC. Typical maximum allowed mixing product levels at the GNSS IC input is -90 dBm.
Prior to the introduction of GLONASS, the filter in a GNSS module only had to cover the GPS center frequency and, therefore, bandwidth was not a problem at all. If we look now to GPS and GLONASS to be served by one filter, this needs to cover a bandwidth of ~40 MHz. The variation in IL within this bandwidth must be below 0.3 dB. Together with the same or even higher attenuation levels outside the band, this becomes a challenge especially for pre-filter design.
Yet, the focus in joint GPS/GLONASS designs is on the GPS system. The lowest IL and noise figure of the module can be found at GPS frequency. But not only a low noise figure over the complete GPS + GLONASS band is desirable, also a flat gain of the module is important to get good system sensitivity over all bands. Gain flatness with a typical value of <0.5 dB is necessary to achieve an acceptable performance over process variation. One significant difference between GPS and GLONASS is that, unlike GPS, each GLONASS satellite is transmitting on a different center frequency. The position acquisition of these systems is based on signal runtime from the satellites to the receiver. The GNSS module now has a group delay, which is not flat over the bandwidth – so different GLONASS satellites see a different delay through the module.
This was suspected to directly translate into position uncertainty. However, these small band dips (i.e. 5 ns within 1 MHz bandwidth) have a much smaller bandwidth compared to the GLONASS spectrum, so a strong averaging effect on the measured filter delay occurs. Therefore, the in-band group delay dips will not be detected by the matched filter or correlation receivers for GLONASS and will not significantly impact the accuracy.
Figure 4 Noise figure over frequency of GNSS/GPS modules.
Figure 5 In-band gain of GNSS/GPS modules.
This section shows RF characteristics of GPS and GLONASS modules for in-band and out-of-band operation to meet the requirements described earlier. In Figures 4, 5 and 6, the GPS/GLONASS modules referred to are: GNSS Mod1: pre-filter + LNA for LTE platforms; mid-gain, GNSS Mod2: pre-filter + LNA for standard platforms; mid-gain, and GNSS Mod3: pre-filter + LNA for standard platforms; high-gain. The GPS Module is GPS Mod1: pre-filter + LNA + post-filter.
Figure 6 Wideband perforamnce of GNSS/GPS modules.
Figure 4 shows the noise figure of the modules over frequency. The GNSS modules are required to demonstrate a low noise figure in the whole bandwidth covering GPS and GLONASS and not exceeding 2 dB. Figure 5 shows the in-band gain of the modules. GNSS Mod3 and GPS Mod1 are high-gain modules, whereas, GNSS Mod1 and GNSS Mod2 are mid-gain modules. The gain ripple of GNSS modules is expected to be below 0.5 dB.
For good linearity and jammer performance, the OoB rejection is the key factor. Figure 6 shows the OoB performance of the modules. The GNSS modules, which are pre-filter + LNA configuration, offer an excellent suppression of all OoB jammers. The GNSS Mod1 is designed to address the LTE Band 13 jammer at 787 MHz, which is exactly half of GPS frequency. The 2nd harmonic of this LTE signal, generated at any non-linear device in the system would interfere directly with the GPS satellite signal. As a result, strong rejection is required to prevent the 787 MHz going into the GNSS path. GNSS Mod1 offers a 75 dBc rejection at 787 MHz and is, therefore, optimized for LTE Band 13 phones.
The GPS Mod1 is a pre-filter + LNA + post-filter module and, therefore, offers the highest rejection over the whole frequency band. It features more than 80 dBc rejection below GPS frequency and 70 dBc rejection above GPS frequency, and with its high integration level, it is an easy part to use.
Integration of various GNSS systems and other wireless connectivity standards onto one platform results in increasing requirements on GNSS system design. The receiver performance can be significantly improved with proficient front-end design. The key performance criteria for the front-end are noise figure, gain, linearity and suppression of out-of-band signals.
- GPS Principles and Applications, E.D. Kaplan and C.J. Hegarty.
- California Eastern Laboratories CEL: AN1050, “Using the UPC8232T5N Discrete LNA to Improve GPS Signal Performance in Mobile Handsets.”
- Letter to FCC: “Notice of Ex Parte Presentation in LightSquared Subsidiary LLC Application for Modification of Authority for Ancillary Terrestrial Component.”
Daniel Kehrer gained an MS in communications engineering and then a Ph.D. in electrical engineering from Vienna University of Technology, Austria. He joined the high frequency circuit corporate research division of Infineon Technologies AG, Munich, Germany in 2003. Until 2005, he was engaged in the design of high speed CMOS electronics for wireline communications. From 2006 to 2009, he worked with the graphics DRAM development team on GDDR5 memory devices for Qimonda AG, where he was managing graphics DRAM design projects. In 2009, he joined Infineon Technologies, where he is leading an R&D team developing complementary wireless front-end solutions.
Deepak Bachu is Product Marketing Manager at Infineon Technologies AG for complementary wireless MMICs, which include GPS, GLONASS, Galileo, WLAN, WiMAX, Mobile TV and FM LNAs. Previously, he had nearly five years experience with diverse RF front-end products as application engineer at Infineon with focus on RF switches and low noise amplifiers. He holds a master’s degree in Electromagnetics, Optics and Microwave Engineering from the Hamburg University of Technology, Germany, and a bachelor’s degree in Electronics and Communication Engineering from Osmania University, India.