Home Featured Articles A Classic LNA Design Challenge: Meeting the Satellite Digital Audio Radio Service (SDARS) LNA Specification

A Classic LNA Design Challenge: Meeting the Satellite Digital Audio Radio Service (SDARS) LNA Specification

56
0

by Alan Ake, Vice President of Applications Engineering/Technical Marketing at Guerrilla RF

This article will focus on one high-volume (primarily automotive) application that goes against this trend and this is the multi-stage LNA associated with the SDARS radio capability that is included with millions of new cars every year. It requires a cascade of multiple high-performance amplifiers and filters to meet a very difficult set of requirements.

These SDARS LNA/filter cascades are located within the ubiquitous “shark fin” antenna modules that can be seen on most newer automobiles. Often, these modules also contain transmit antennas for cellular phones, thus the radio frequency (RF) environment can be very challenging. For this reason, it shouldn’t be surprising that the LNA specification requirements are difficult to meet.

The SDARS radio channel occupies the 2320 to 2345 MHz band and for SDARS LNAs, the specifications are written in terms of the cascaded performance of the required amplifiers, filters, cables and connectors. The gain and noise figure (NF) requirements are fairly straightforward, but there are several linearity metrics which must be met to include requirements for third order intercept (IP3), 1 dB compression point (P1dB), out-of-band (OOB) blockers, second and third order intermodulation rejection ratio (IMRR2 and IMRR3), etc.

The following discussion will focus on these interrelated gain, NF and linearity requirements.

Gain
The SDARS application requires a typical LNA gain of 22 dB after considering all filter, cable and connector losses. Given the lossy coaxial cable connecting the SDARS radio to the roof-mounted antenna, the antenna module gain for an application typically needs to be in the 30 to 35 dB range and this must include roughly 4 dB of surface acoustic wave (SAW) filter losses. This net gain requirement drives the need for an architecture with two high-gain LNAs, as can be seen in Figure 1.

Figure 1: Generic High-performance SDARS LNA Block Diagram

To get an idea of the total gain required from LNA1 and LNA2, consider the case where the antenna module must deliver 35 dB gain. Accounting for insertion losses from SAW1 and SAW2, the gains of LNA1 and LNA2 must add up to around 39 or 40 dB. Achieving this high level of gain over 2320 to 2345 MHz from only two amplifiers requires today’s modern pseudomorphic high electron mobility transistor (pHEMT) process technology and, to further enhance the gain, these amplifiers are typically designed with cascode architectures.

For LNA design in general, gain partitioning is an important consideration and that is especially true here. If the design of an SDARS cascaded LNA could be distilled down to its essence, it would be the interaction between the OOB rejection of SAW1 and its associated insertion loss with the gain and NF of LNA1. As can be seen in Figures 2 and 3, a fairly high level of gain from LNA1 is required in order to achieve some margin to the NF requirement. It is precisely this NF margin which makes possible the use of a SAW1 device with improved OOB rejection characteristics.

Linearity
Out of Band (OOB) Blocker: For antenna modules with co-located cellular/Wi-Fi transmit elements, the specification requirement is that the gain of an in-band reference signal compress <1.0 dB when an OOB blocker signal of +10 dBm is applied. These blocker signals are defined for a wide range of frequencies both below and above the SDARS band and an inherent problem here is that the gain of the LNAs increases significantly at lower frequencies such as 800 MHz. This provides a glimpse into why the OOB rejection of SAW1 is important since it must reduce the incident power of these blockers a great deal in order to protect the limiting stage which is LNA1.

The Guerrilla RF applications team finds that, of all the various linearity requirements for these SDARS LNA cascades, this blocker specification is the most difficult to meet. The good news is that a receiver chain that can pass this test usually has no problem with any of the remaining OOB interference requirements.

Critical point: This blocker requirement implies that the SAW1 prefilter has an excellent OOB rejection characteristic as it is essential that it adequately protect LNA1 from these interferers. The rub is that there is a tradeoff between the OOB rejection characteristics of SAW1 and its in-band insertion loss. Because of this, a designer is forced to balance OOB blocker performance with cascaded NF and, as was shown in Figures 2 and 3, there is precious little room to give in terms of SAW1 insertion loss when it comes to NF.

IMRR2 and IMRR3: The response of the LNA cascade to these OOB interferers is typically controlled by the OOB rejection characteristics of SAW1 and LNA1. As noted above in the blocker discussion, passing the blocker requirement typically assures plenty of margin to these IM requirements.

IP3 and P1dB: These in-band linearity metrics are typically easily met given the linearity of the typical LNA1 and LNA2 solutions. For this reason, these parameters typically do not significantly influence the design of the cascaded LNA.

Noise Figure
The cascaded NF requirement of 1.0 dB is difficult to achieve given the losses associated with SAW1 and SAW2 and these SAW filters are necessary in order to meet the various linearity specifications. Here, the NF performance of LNA1 is critical since this NF essentially adds to the insertion loss (NF) of SAW1. Its gain is also important as this minimizes the NF contribution of the lossy SAW2 and the relatively low NF of LNA2.
To better illustrate the importance of the LNA1 NF, please look at Figures 2 and 3. These plots show cascaded NF versus LNA1 gain for an LNA1 NF of 0.35 dB and an LNA1 NF of 0.45 dB. To isolate the effect of LNA1, the NF of SAW1 and (SAW2+LNA2 NF) are assumed to be 0.5 dB and 4.5 dB, respectively, for the sake of the cascaded NF calculations.

The takeaway from these plots is that it is critical that the LNA1 NF be <0.45 dB. Per Figure 2, if LNA1 has a NF of 0.45 dB, regardless of LNA1 gain, the noise figure of the LNA cascade can’t get significantly below the 1.0 dB limit whereas an LNA1 NF of 0.35 dB as shown in Figure 1 allows the solution to achieve some margin to the NF requirement. If such a device exists, a part with a NF of 0.30 dB would be even better, but it should be noted that all these NF values discussed are pushing the capability limits of today’s best pHEMT processes.

These NF curves show that there is a definite window as to the amount of LNA1 gain is desirable. As the gain drops below 19 dB, the NF begins to rise rapidly towards the 1.0 dB limit and as gain rises above 21 dB, NF improvements are minimal with cascaded LNA linearity degrading rapidly.

Notes on Component SelectionSAW1:
This filter is critical for setting the cascaded NF and the OOB blocker linearity performance of the LNA solution. The issue here is that improved OOB rejection comes at the expense of higher in-band insertion loss. This OOB rejection is critical to the overall linearity performance since LNA1 is typically the limiting stage in terms of linearity. From our experience, the lowest loss SAW1 solutions have an insertion loss of around 0.45 dB with an associated rejection at 800 MHz of only around 10 dB. The highest loss SAW1 device the NF requirement will allow can be no more than 0.6 dB. Devices with this level of insertion loss can have higher 800 MHz rejection of around 20 dB, thus offering superior blocker performance. Please consult with the Guerrilla RF applications engineering team for manufacturer and part number suggestions.

LNA1: As shown from Figures 2 and 3, the NF of this ultra-low noise device is critical to achieving the 1.0 dB cascaded NF requirement. Having the lowest possible NF here allows the use of a SAW1 device with better OOB rejection characteristics, with the automotive-qualified GRF2073-W and GRF2093-W being the industry’s premier LNAs for this application. Over the 2320 to 2345 MHz band, GRF2073-W and GRF2093-W have de-embedded NF values of 0.37 and 0.30 dB, respectively. These ultra-low NFs, combined with their high gain, allow the cascaded LNA solution to meet the NF requirement while still using a SAW1 device with adequate OOB rejection. Specifically, GRF2093-W, with its 0.30 de-embedded NF, allows for the use of a 0.6 dB insertion loss SAW1 device along with its superior OOB attenuation characteristics. This results in the industry’s best combination of cascaded NF and linearity performance.

SAW2: This filter supplies the required rejection for the Wireless Communications Service (WCS) bands which occupy 2305 to 2320 MHz and 2345 to 2360 MHz. Achieving this close-in rejection leads to a SAW filter with relatively high insertion loss of around 3.5 to 4.0 dB. Not trivial, this loss is a major factor in determining how much LNA1 gain is required to meet the overall NF requirement. The high WCS band attenuation of this filter usually makes it the source of any excessive gain ripple within the SDARS band, thus some minor external matching is sometimes required. Please consult with the Guerrilla RF applications engineering team for manufacturer and part number suggestions.

LNA2: Well-protected by the preceding SAW filters, this LNA primarily provides any necessary gain to meet the design requirement with only modest linearity requirements. The amount of gain required for LNA2 varies by the length of coaxial cable required for a specific application and by the gain of LNA1. Typical LNA2 gain requirements are in the 14 to 20 dB range with NF of <1.0 being sufficient so as to not significantly impact the cascaded NF. Guerrilla RF has many component options for this location, with the automotive qualified GRF4002-W or GRF4014-W being excellent choices.

Summary
This discussion focused on helping design engineers understand the challenges associated with LNA/filter architecture and device selection for an SDARS LNA module design. To further assist with this, Guerrilla RF offers tuned evaluation boards for the LNA components noted here and for the rest of our product portfolio. Specifically for SDARS, we offer tuned reference design boards containing the required SAW filters, LNA devices and discrete RLC components that can serve as the foundation of a specification-compliant design. At these frequencies, layout and board parasitics/grounding are critical elements of the design, thus final optimization will be needed on any customer application board.
As always, the Guerrilla RF applications engineering and sales teams are happy to assist you with this or any other design challenges you have. For further information or support needs please contact us at:
sales@guerrilla-rf.com or applications@guerrilla-rf.com

(56)

print

LEAVE YOUR COMMENT