Signals and Sensitivity
In wireless communication, there is always some level of error rate, or at least the chance of some signal error. In this article, Robert revisits the topic of wireless receiver sensitivity, this time looking at how different receivers react differently to electronic perturbances.
Welcome to “The Darker Side.” I previously devoted a column to wireless receiver sensitivity: “Long-range RF for the IoT-Part 1” (Circuit Cellar 313, August 2016). This month, I am returning to this topic and to explain why knowing the sensitivity of a receiver is not enough. Specifically, I will explain why receivers are not equal when some electronic perturbances are around, and for sure there will be some. As usual, I will not use any complicated math. So, stay cool and enjoy!
First, let’s come back to sensitivity. Imagine that you have a wireless link between two devices—a transmitter and a receiver some distance away. It could be any kind of wireless link, such as Wi-Fi, Bluetooth, LoRa, cellular or anything else. The transmitter is sending a message, which is a given sequence of bits, and the receiver does its best to receive these bits. If the receiver is close to the transmitter, then the received signal power is high, and the bits are recovered nearly without any error (Figure 1, Case 1). The so-called “bit error rate” (BER), which is simply the number of bits in error divided by the total number of bits transmitted, will be close to 0%.
It’s important to know that the number of errors is never zero. It could be very close, but in any transmission, there will always be some probability for errors. If you are not convinced, please re-read another of my articles, “Shannon and Noise” (Circuit Cellar 331, February 2018).
Now, imagine that the receiver is moved further away. The received signal power will be reduced, and the number of errors in the received bits will increase. BER will climb to 0.1% (Figure 1, Case 2), then 1% (Case 3), then 3% (Case 4) and so on.
What is the maximum tolerated bit error rate? Well, it depends. For a given application, the designer will have to set a threshold. For example, a BER of 1% may be tolerated if the communication link is not mission critical, and if there are some ways to detect and recover a few errors through a proper protocol. When this threshold is set, then the so-called sensitivity of the receiver can be measured. It is simply the minimum received signal power that provides a bit error rate not higher than this threshold. Looking again at Figure 1, in this example and for BER=1%, it will be the level corresponding to Case 3. So, keep in mind that a sensitivity must always be specified for a given error rate; if not it is meaningless.
I explained that sensitivity is the received signal power that allows us just to get the desired bit error rate. This is when life is perfect, and in particular, this means that there are no perturbances at all. This is, unfortunately, rarely true in real life, as you may have guessed. Now what happens in case of perturbances? I will start with the case of “co-channel perturbances.” This means that the perturbance is supposed to be at the same frequency as the received signal, more exactly, in the same frequency channel. This is, of course, the worst situation. What will happen? The receiver will be less sensitive if a perturbance is present, but not all receivers will be affected equally. How to quantify their tolerance of a perturbance?
In Figure 2 (Case 1), there is no perturbance, and the received signal is just at the sensitivity level. Because we are right at the threshold, any perturbance will increase the BER above the threshold, and the reception will be considered as “bad” (Case 2). So, let’s increase the power of the desired signal a bit, usually by 3dB (Case 3). If the perturbance power is low enough, then this increase in signal power will bring the BER back under the threshold, and the reception will be good. Then imagine we slowly increase the power of the perturbance, still with a wanted signal 3dB above sensitivity level. At a given power of the perturbance, we will again have a BER equal to the desired threshold (Case 4). If we increase it, then the BER will too high (Case 5).
With this method, we actually have a way to measure the resistance of the receiver to perturbances in the same frequency channel. Just do this test, and measure the power difference between the maximum perturbance power and the received signal power, which is sensitivity + 3dB. The resulting value is called the “co-channel rejection,” and is one of the figures of merit of the receiver (Figure 3, right side).
Of course, the higher the co-channel rejection, the more tolerant the receiver will be to perturbances at the same frequency. To be complete, the kind of perturbance used for the measurement must also be specified (carrier wave or modulated signal, and so on), but you get the idea. This co-channel rejection is, in fact, mainly linked to the characteristics of the modulation. For example, nearly all frequency-shift keying (FSK) receivers have a co-channel rejection close to -10dB, meaning that the perturbance in the channel must stay 10dB below the received signal level.
Is this all? Unfortunately not. The reason is that no receiver is perfectly selective, meaning that it doesn’t listen only to a given frequency channel. To explain why, I drafted the typical architecture of a wireless receiver (Figure 3). This is a simplified view, but it is enough for this explanation. Here, the signal from the antenna is supposed to pass through a rough, front-end, band-pass filter, which rejects frequencies of other frequency bands, then is pre-amplified by a low-noise amplifier (LNA). The received signal is then translated to a lower intermediate frequency (IF), passed through a very selective channel filter to isolate only the channel of interest, then is amplified again and demodulated.
Life is hard, and no filter is perfect. In particular, the channel filter will need to attenuate the signal in the received channel as little as possible. Since a brick-wall filter doesn’t exist, this means that it will not completely block signals that are close in frequency—specifically, perturbances in one of the two channels immediately adjacent to the received channel (Figure 4).
As with the case of the co-channel rejection, receivers are not equal in terms of their ability to tolerate a perturbance in an adjacent channel. And as with the co-channel rejection, a figure of merit of the receiver can be measured, called the “adjacent channel rejection.” How? The received signal is simply set 3dB higher than the sensitivity level, and a perturbance is switched on at the frequency of one of the adjacent channels (Figure 5). Its power is slowly increased until the bit error rate reaches the desired threshold again, and the difference between the perturbance and received signal is measured. This gives the adjacent channel rejection.
This figure of merit is mainly linked to the performance of the channel filter of the receiver. Typical receivers tolerate perturbances in the adjacent channel with power levels significantly higher than the received level, meaning that the rejection is positive when expressed in dB. Values ranging from 30dB to 80dB or more are possible, depending on the performance of the receiver.
Last but not least, let’s talk about potential perturbances that are far away in frequency from the received channel—some MHz away. Look again at the typical receiver architecture shown in Figure 3, The channel filter will strongly attenuate such signals, so there will be no impact, right? Unfortunately, wrong. The reason is in the first stages of the receiver—here the front-end filter and LNA. Imagine that there is a perturbance about 10MHz away from the signal of interest, but with a strong power. Even if it were rejected by the channel filter, this perturbance could pass through the front-end filter, and could saturate the LNA. And a saturated LNA no longer amplifies the signal of interest, so the reception would be jeopardized.
Therefore, similar to the case of adjacent channel rejection, it is a good idea to quantify the rejection of perturbances quite far in frequency. This is exactly the same measurement, but the perturbance is here set at a given frequency difference from the received channel, usually 2MHz or 10MHz away (Figure 6). In a nutshell, a receiver usually has three figures of merit in addition to its sensitivity, as shown in Table 1.
EXAMPLE OF THE ADF7030-1
As an example, let’s look at the figures of merit of an integrated wireless transceiver. Since my company regularly uses it, I have chosen the ADF7030-1 , a high-performance, low-power, sub-GHz transceiver from Analog Devices. (Figure 7). The datasheet includes several tables, depending on the frequency and modulation, so let’s look at one of them—the one for 868MHz operation (Figure 8).
The first value is the sensitivity, itself, which depends on the selected bit rate. For a 4.8kbps speed, it is stated at -118.4dBm. In the far-right column of the table, you can see that the manufacturer indicates that this value assumes that the maximum packet error rate (PER) is 5%. PER is closely linked to BER, but is measured on full radio messages and not at the bit level.
The second and third sections of Figure 8 give the adjacent-channel and blocking rejections, measured either for a BER of 0.1% or for a PER of 5%. For example, in the case of PER = 5%, a speed of 4.8kbps and a channel width of 12.5kHz, this chip provides an adjacent channel rejection of 43dB. This is great achievement, but it still means that any perturbance in an adjacent channel and stronger than -118dBm + 43dB, which is -75dBm, will reduce the sensitivity of the chip. And -75dBm is a quite low perturbance power, so this may happen.
Similarly, the blocking rejection at 2MHz is 79dB, and at 10MHz it is 87dB. These are very impressive figures. For example, 87dB of rejection means that a blocking perturbance may have a power 87dB higher than the signal to be received, so this power could be 108.7 = 501,187,233 times higher! Finally, the bottom section of Figure 8 gives the co-channel rejection. Here it is -11dB, which means that the perturbance must stay 11dB lower that the signal to be received.
A REAL-LIFE EXAMPLE
At this stage, you may wonder if such rejections are enough or not for a given application. Of course, it depends on the application, but I would like to give you some highlights on real-life situations where such an impressive 87dB blocking rejection at 10MHz simply may not be high enough.
The example I will use is a well-known potential issue in Europe, between 4G cellular networks and so-called ISM networks working in the 868MHz band. I’m sure there are similar situations in other geographical regions, but I know this one, because it is documented in an official report (CEPT ECC Report 207) .
Imagine that in your home, you install a device with a long-range, wireless connection to a gateway that is miles away, using the 868MHz frequency band (Figure 9, right side). Your device will send messages and may also receive some messages back, using a very sensitive receiver, since the gateway is far away, say with a sensitivity of -118dBm, as for the ADF7030-1 at a bit rate of 4.8kbps.
Now, imagine that you use your latest smartphone in the same room, using a 4G network. 4G, also called LTE, supports plenty of frequency bands, depending on the countries and operators. One of them (band 20, sub-band C) uses the band 791MHz to 821MHz for transmissions from the network to the smartphone, and 832MHz to 862MHz from the smartphone to the network. And unfortunately, 862MHz is not far from 868MHz! Your smartphone could transmit 1W or so (+30dBm), and the received signal power can easily be -20dBm, even a few meters away. And now you understand the issue. The blocking rejection of the 868MHz device would need to be -20dBm – (-118dBm) = 98dB, and it is “only” 87dB.
This situation is extreme, because your smartphone may not use this band, may not be so close to a high-sensitivity ISM device and may not transmit when the ISM device is receiving a message. But this really could happen. And, as stated by Murphy’s Law, if anything can go wrong, it will, and usually at the worst moment!
Here we are. Even if few of you develop wireless receivers, I hope that you were interested. Interferences and perturbances are, and will increasingly be, the main cause of disappointments in wireless communications. Understanding how to read rejection characteristics of the receiver you intend to use is a mandatory step in assessing if it is good enough or not. Because it is a very good exercise, I strongly encourage you to download and read the datasheet for your preferred chip. And then just try to find out and understand its rejection characteristics!
 ADF7030-1 integrated transceiver (Analog Devices)
 CEPT ECC report 207
Adjacent band co-existence of SRDs in the band 863-870 MHz in light of the LTE usage below 862MHz
Analog Devices | www.analog.com
PUBLISHED IN CIRCUIT CELLAR MAGAZINE • OCTOBER 2021 #375 – Get a PDF of the issueSponsor this Article
Robert Lacoste lives in France, between Paris and Versailles. He has more than 30 years of experience in RF systems, analog designs and high-speed electronics. Robert has won prizes in more than 15 international design contests. In 2003 he started a consulting company, ALCIOM, to share his passion for innovative mixed-signal designs. Robert is now an R&D consultant, mentor and trainer. Robert’s bimonthly Darker Side column has been published in Circuit Cellar since 2007. You can reach him at firstname.lastname@example.org.