RF Performance in IoT End Devices - Is Good Enough Really Good Enough?

Welcome to our Tech Talk today. I was inspired to talk about this for one reason. When I often talk with customers, they often tell me that they just need RF performance that's good enough. 
  
And what does that really mean, that statement? RF performances really can directly impact the customer's user experience. And when you talk to customers, they may not tell you, "Yeah, RF performance is the most important thing," because they expect the product to work. 
  
But if they have that negative user experience because of RF performance, they're going to end up blaming a number of reasons. Maybe it's the internet provider, they just think, "Well, something's wrong with the internet." Or it could just be the hub they have, or maybe even the direct device. But whoever gets the blame, it's going to be a losing situation for everybody involved. 
  
We don't want technologies to be viewed bad, so RF performance is really critical. And as you're designing products and making products, defining products, these are areas that are often overlooked. So today, we're going to actually take a step at that and try to talk about some key areas for RF performance, why they're important, and what Silicon Labs is doing around that. 
  
So my name is Matt Maupin. I've been a product manager here at Silicon Labs. I'm currently in product marketing, where I have a team that's responsible for 802.15.4 Matter, as well as Wi-Fi. 
  
And then I'm really excited today to have Chris Calvo. One of the reasons I joined this company 14 years ago was because of the designers. Their experience, their knowledge, and their passion around RF was really inspiring to me. 
  
So I'm really excited that Chris is on today. He's going to be able to present some and then be on for the Q&A as well. Chris has a master's degree in electrical engineering. 
  
And he leads some of our RFIP development for our latest Series 3 platform. If you're not familiar with that, we've had Series 1, Series 2, and Series 3 over the years. We just recently launched our first Series 3 end of last year, the SiMG301. 
  
And if you look at the RF performance on this part, it really leads out there. We've got exceptional output power, best-in-class sensitivity, and then we also have other areas around selectivity, et cetera, that I'm going to talk about. So excited to have Chris on today. 
  
This is really a benefit for everybody. From an agenda standpoint, we're going to sort of tag team this between Chris and myself. I'm going to talk about what are the key RF performance specifications that you should be concerned about to make sure you have best-in-class products from an RF standpoint. 
  
And then we'll go into details on those. We'll talk a little bit about link budget. Chris is going to jump in and talk about coexistence and blocking performance. 
  
We'll have sort of a Q&A after him, and then I'll wrap it up with antenna diversity, and then we'll have another Q&A. So, when talking about RF performance, there's really several specs that are important, and some of these are commonly known to people, and then others they just may not be aware of, or they may not understand the impact. So today, if we talk about it, sensitivity, I think is one, and output power both. 
  
Pretty much everybody understands what those are. As I go through these, I will try to relate them more on a marketing standpoint on how to make it a little easier to understand. For me, I've actually got a genetic hearing impairment, so I like to sort of relate things to hearing and how it affects that for me. 
  
And so I'm going to sort of put a lot of these on how it would relate to your hearing. So sensitivity here is really the minimum signal strength the receiver can detect and decode accurately. I say accurately because imagine you're sitting somewhere and somebody's talking. 
  
You may be able to hear them, but you can't understand them. That's sort of where that sensitivity and that minimum decode comes in. Sensitivity and output power are both in DBMs. 
  
That's a measurement. And if you look at that, sensitivity is a little weird because it is a negative number. So think of it like this, the bigger the negative number, the weaker the signal can be that it can detect. 
  
So for sensitivity, you want a bigger negative number. And obviously, higher sensitivity improves range, link reliability, and that's true especially in low signal or noisy environments. So really sensitivity, how well do you hear? 
  
Output power, obviously this is from a hearing standpoint or communication, how loud do you talk? I'm a marketing person, so I tend to talk much louder, so I have a greater output power when it comes to conversations. Same thing with receivers is how loud does it transmit? 
  
What's the power level there? And obviously, higher output power sort of increases that range as well as penetration through obstacles, et cetera. So sensitivity and output power are two of the critical factors. 
  
We're going to refer to these as link budget, but I will talk a little bit that's often overlooked here on the next slide about how these really need to be balanced, and I'll talk more about that. When we get to selectivity and channel rejection, that gets a little more complicated, in my opinion, to really fully understand. These things are the receiver's ability to hear that sort of wanted signal when there's other signals that are close by that are in the same realm. 
  
For example, Bluetooth. I could have a lot of Bluetooth activity around me, and while it may not be on my channel, it may be near me, and that's affecting the receiver's ability to get the signal it wants. And so-I think a good example here for me is, you're in a conference room and you're listening to a speaker or presenter, and next to you there's sort of a side conversation going on. 
  
Selectivity is the ability to sort of ignore that side conversation and actually comprehend and hear what that presenter is talking about. And again, these are really essential when you're in environments that have a lot of traffic. Even in my house, I have a lot of home automation devices. 
  
I think I've got close to 80. Obviously, I've got Wi-Fi, Bluetooth, 802.15.4, even some Sub-GHz. But for 802.15.4, I actually have multiple networks. 
  
So these networks could be close to each other and interfering with each other. So that's really where selectivity or an 802.15.4 channel rejection come in. And then next, you have blocking. 
  
This is one that I think a common thing here is obviously Wi-Fi. Wi-Fi transmits at a very high power. With Wi-Fi Mesh, it's much stronger throughout an environment. 
  
And when it comes to blocking, it's the receiver's ability to sort of still maintain performance when there's a strong, unwanted signal. So something completely out of band or in band 2.4, but not what its modulation is, and it's present and nearby. So here, I think, if you're trying to understand that, imagine you're at a venue, whether it be a restaurant or a bar, and you're trying to have a conversation and away from you, there's a loud band playing. 
  
It's very hard to sort of have that conversation when there's all that noise. So that sort of comes into where blocking performance is, and this is critical, especially in all environments with Wi-Fi. One of the things Chris will talk about, he'll show some different blocking performance testing that we've done, and there's a couple of different examples he'll show. 
  
One is -10dBm blocker, one is -40. To give you an idea of what you may experience in the home, I do have Wi-Fi mesh throughout my house, and so I went around with a spectrum analyzer and I looked at where my different Zigbee devices were and my Matter over Thread devices were, and I measured my Wi-Fi signal at those devices. And in some areas it was in, -20 or so. 
  
I actually have a credenza that unfortunately, I go against what they say, and I have my access point and a couple of hubs all located very close. In that instance, I was getting very strong blockers at those devices. But even throughout my house, most of my devices or my light switches or my light bulbs, I was seeing Wi-Fi signals in the strength of -40dB, which is pretty strong. 
  
And Chris will talk about why blocking is important in those environments. And then finally, we'll talk about antenna diversity. I think this comes down to the old adage, two ears is better than one. 
  
Same thing with antenna diversity. What you're basically doing is because of things like multipath, reflection, it can affect signals at an antenna. But by having two antennas that are spaced out, one of those antennas is going to have better performance than the other. 
  
And the whole idea here is the device can select between the better of the two antennas and sort of reduce that multipath and deep fades. So really that's going to give you the best performance. I will say most of the people I see implementing antenna diversity are really more on the industrial and commercial side, where they may have buildings that have a lot of metal reflections and they're more concerned with it. 
  
There are a few on more of the consumer side that are really worried about performance and want the absolute best, but this is something I typically see implemented more in industrial and commercial environments. So let's talk a little bit about link budget, which is the combination of output power and sensitivity. And why is this important? 
  
One of the things I talked about was these sort of need to be balanced, and I'll talk about why. So, higher link budget does improve better range and reliability, but often what you see is vendors will have high output power, but their sensitivity will stay where it's at. And really what happens there is you get into what I call a shouting node syndrome, especially in a mixed environment. 
  
I'm able to transmit very high, speak very loudly, but maybe the transmitter coming back to me is at a lower output power. Well, they're going to be able to hear me, and I'm not going to be able to hear them. So really balancing these two is important, and I'll give you some examples here. 
  
So, if we talk about symmetric communication, this is when the link budget is the same on both sides. So for example, I've got a couple examples here. We have a competitor out there that actually has a plus 20dBm part, and their sensitivity is -102. 
  
And if you look at the range they can get, that's about 156 meters. That's a calculated range that I used, and there's something called a propagation factor of N equal four, which is indoors. So that's just a calculated range. 
  
What we have is we have similar output power, but we have a -105.4 sensitivity on the device that I'm talking about here. And what that means is we do get a better range, obviously with that, just because we have a higher link budget. But what's more likely is you're going to sort of have this asymmetric communication because you're not going to be communicating just with your devices. 
  
In my home, I have devices from all different vendors, and so what you're going to see is one vendor may perform better than the other. So here I'm showing an example of a typical node and then transmitting to them. So in this purpose for this typical node, it's a 10dBm device with -101 sensitivity. 
  
And again, this is 802.15.4, but that's a little more typical if you look at a lot of the chipsets out there. So, this competitor A to this typical node, really the range goes down to 83.2. But in our case, even though they have slightly better output power, we have much better sensitivity. 
  
We balance those two better, and we get about 29% better improvement. So my whole point here is just don't look at one. You need to look at two. 
  
Both of these need to be balanced and work together to give you the absolute best range and performance.And then I often hear, especially for 802.15.4, whether it's Zigbee or Matter over Thread, "This is a mesh network. I'm really not worried about range because all I have to do is get to that next node." Yes, that's true, but there's two things there. Obviously, one is you don't know where that next node's going to be. 
  
But more importantly is a lot of these networks are optimized to reduce the number of hops. So the farther I could transmit reliably, I will have less hops in the network, and that reduces traffic latency and obviously improves battery life as well if I'm not having to do retries, et cetera. So even in environments where you may not think range is important, mesh range is still a very important aspect. 
  
And then obviously, why doesn't everybody do this? Well, one of the things is with this higher link budget, while it does improve range and reliability, it does come at a cost. There's increased current with both receive and transmit because of these. 
  
Obviously, on the transmit side, I've got PAs that are putting out more output power. On the receive side, I have an LNA. I'm having to crank up the current on that usually to get that higher sensitivity. 
  
So there is a battery cost on this. And then obviously there is some increased die area as well. So it is a trade-off, designers and product managers have to sort of make as they're defining and building these products. 
  
So with that, I'm going to go ahead and turn this over to Chris, let him come online. Again, if you have questions, go ahead and put them in, and we'll answer those on the way. So Chris, go ahead and take it away. 
  
All righty. Let's go on to the next slide then. We're going to talk about coexistence and blocking performance. 
  
This is a topic that's pretty interesting to me. This is definitely one of the biggest challenges when it comes to receiver design, and how to pull off in such a crowded environment, being able to perform well even though you have all of these blockers in that environment. If you go to the next tab here. 
  
So on the right side of this graphic, you can see here it is kind of a breakdown of the 2.4 gigahertz band. And then this is going to give you-- There's basically many protocols operating all within the same band. You can see Zigbee, Bluetooth, Wi-Fi, and all of these operate in the same frequency band at the same time. 
  
There's no coordination between them. So at any time you could be trying to send something on Zigbee, and at the same time, streaming traffic on Wi-Fi is going to be coming across. And this makes for a really tough design process. 
  
We have to be able to filter out signals, make sure that we have large dynamic range capability so that even though you're trying to listen to this small signal, you're able to pull that off even with a Wi-Fi transmitter sitting right near you. If you could go on to the next bullet point there, that'd be great, Matt. And so, this whole thing gets worse and worse as we talk about this proliferation that we're going through of more devices in the home. 
  
And then all of these devices increasing their transmit power, more of these nodes and networks. This whole problem just becomes worse and worse and worse as we're starting to move along in technology and things like that. And then in general, this problem is getting worse because even our radios alone are starting to have multiple radios in one device. 
  
You have your smartwatch or anything like that. You're running Bluetooth and Wi-Fi simultaneously. Sometimes you're trying to do GPS simultaneously, and these end up being really tough problems to solve for a radio designer to be able to pull that off without losing performance. 
  
You don't want to necessarily have to move your watch closer to your phone or your Wi-Fi router just because they're both running at the same time. And this is why it's a very challenging and interesting thing to me personally. And we try our best here to produce the best performance we can, even with these blockers in place. 
  
Can you go to the next slide, Matt? Yep. Thank you very much. 
  
Okay, so in general, you can imagine that this coexistence scenario gets really tough. You're going to have all these multiple technologies, and then really, like I alluded to, there's really no coordination between them. Each one of these is going to act independently. 
  
They're going to operate on the band that they were predefined or pre-selected to. Wi-Fi is a good example of that, right? If you want to go to the next bullet point, that would probably be helpful here. 
  
Wi-Fi here, in this example here on the right figure. Maybe go to the next bullet point. It helps to have the figures up there. 
  
Thank you. When you're having Wi-Fi transmission, there's a pre-selected channel. So in this case, they selected channel one. 
  
And you're going to have no ability to know, predetermine that what channels are going to be available or not. And so you have to have very unique technologies in place to be able to mitigate these things. So in order to operate between us, we have to have technologies like, "Hey, we need to make sure our radios are separated in space." This gives us more ability to handle larger power levels from the Wi-Fi transmitter while still pulling off our receive that we want. 
  
We got to separate in frequency, and so the graphic here gives a good example of that. So where you wouldn't want to select the left channel here. For Zigbee, you'd want to select the rightmost channel, give yourself the largest frequency separation you can get here. 
  
And then, in general, the protocols will have some level of management of this. CCA timeouts, retries, and things like that. But these are things that we would like to avoid. We would like to inherently design the radio so that we don't need to use these things, because as you can imagine, these timeouts and retries all require time. 
  
So you're going to lower your throughput. You're going to say, "Hey, I tried to send this. I didn't get an ACK," which means that you're going to have to resend, and then you're lowering your throughput of your data. 
  
And then that's going to lead to customers saying, "Man, I can't get this over-the-air update. It's taking forever." And that's not what we want to provide to customers. And so what we really try to go for is how can we make the radio perform to as good performance as if there was no blockers in the area? 
  
And then that comes down to these three things are kind of overlapping topics, selectivity, channel rejection, and blocking. And there's more things involved in this, but they kind of all together work to create the capability of your receiver. Selectivity is basically how much filtering do we have in the lineup? 
  
How much can we reject this yellow Wi-Fi 20 MHz band away so that it doesn't affect our channel 25 transmission on Zigbee? And then channel rejection is kind of similar vein, right? And blocking is how much of this blocker power can we handle before we start having an issue where our sensitivity starts to degrade? 
  
And so really, a lot of our discussion here is going to go into the radio performance and things that we choose or make trade-offs to improve that, because all the other stuff is great, but they're kind of outside of our control, right? These are users or software at higher levels, but the radio performance is what we can control to make it so that you don't have to do a CCA retry or timeout. And you can keep up the data throughput that you were hoping for. 
  
Matt, if we go to the next slide. Okay. So here's a slide where we're going to get into some testing that we've done for one of our products, and then also comparing it against a few competitors and one of our Series 2 parts. 
  
And so we're going to discuss here a test that we did where, hey, we have a Wi-Fi blocker, right? And we're trying to transmit that Wi-Fi blocker. It's kind of the mid-band, so this will be like channel 6. 
  
And then we're going to sweep our 15.4 receiver across through that band. And we're going to see what sensitivity level can we pull off given that there's this really large Wi-Fi blocker. And in this case, it's a -10dBm blocker. 
  
And so on the right side figure, you can see here that as you traverse across the 2.4 GHz band, you'll see right in the middle, which is approximately where the channel six of the Wi-Fi transmission is happening, or channel seven in this case for this plot, you see that the sensitivity really degrades. It comes up to -15dBm sensitivity for the 15.4 transmission for the receiver. And that's kind of the case across all parts, Series 2, Series 3, and our competitor parts that we have here on this plot. 
  
In general, there's not much you can do when all of this power is right on channel. But really the benefits and where you care about it is when you're going outside of that. So that's like beyond on the low side, so that's about 24, let's say 25, and above 2460, right? 
  
This is where the Wi-Fi signal level starts to drop off, and then this is where you're going to see the cream of the crop, who's got the best blocking performance amongst all of these products, right? And in this scenario that we're looking at, this is a -10dBm blocker from Wi-Fi. And so this is a pretty high power level. 
  
So you're talking something that's going to be like one to maybe two feet away from the router. And so an example of this would be like a wearable watch, a glucose meter, or something like that would be a really good example of this because you'd have your phone, which is not coordinating with this device, and it could be sitting right next to you, and it's transmitting Wi-Fi back to the router at max power. And so even in this scenario, you need to perform and get that information, whether it's notifications, music streaming, things like that. 
  
You need to get that stuff across without having an issue. And so down here is where we do a basically apples to apples comparison across XG24, so this is one of our Series 2 parts, XG301, which is one of our new Series 3 parts, and then two competitors, A and B. And then we pick out a couple points, we call them deltas from the band edge. 
  
So you could think of it like, on this case, we're looking at the lower band edge, so that's approximately like the 2425 MHz delta from there. So you're going to say 12 and a half MHz away, and we select off what are the sensitivity numbers of the XG24, the 301, and the two competitor parts. And you can see that at this high power level blocker from Wi-Fi, there's a pretty significant delta between us and competitors. 
  
And what does this delta mean, right? How can you put these numbers to meaning and understanding, right? If you look at these numbers, when you have, for example, the XG24 versus the XG301, and you have a 4dBm or 4dB difference, technically, this is a dB difference, right? 
  
What does that mean? And what it means is that in the case of the XG24, its sensitivity is 4dB better than our Series 3 part at the same delta And then at the 22.5 MHz delta, it's 1.5 dB better. And so what do these dB's mean to you as a user or a designer for your customer, right? 
  
When you're looking at a 3 dB delta in sensitivity, that's approximately 1.4 times more range, 1.4x more range. And then what does that mean for coverage? That means for the same area of your network, you've doubled the coverage comparatively from one chip to another. 
  
And then you can imagine that when we're talking about our products that are, in the case of the 22.5 MHz away, we have a 1.5 dB delta between them, which is small. But then you look at these two competitor parts that have a 28 dB and a 7.3 dB delta for the 22 and a half. We're talking, this is a pretty big delta for blocking performance, where, for example, for competitor B, right, that's four times less coverage area when a blocker, a 20 MHz wide Wi-Fi channel seven blocker is happening 22.5 MHz away from what the channel they're operating at. 
  
So these are real problems for our customers, right? You're going to say, your customer is going to now have to get that much closer from their watch to their phone, and then now it's not going to work. They leave their watch on their wrist, and they leave their phone on their desk, and now they have to get them closer in order to manage that link, right? 
  
And so these are the things that very careful receiver design and AGC-related things get us these types of performance numbers and the deltas that we get between us and competitors. Let's go to the next slide, Matt, because I think this next one is an even better example of this. So we'd kind of talked about the -10 dBm blocker case. 
  
And so that I kind of equated to being something like, hey, you got a wearable and your phone, right? Those are the 1 to 2 feet distances. And then in this case, this is a -40 dBm case, right? 
  
So this is more realistic. You're talking something in the 20 to 40 feet range, depending on a lot of environmental factors, right? But a significant amount further away than the scenario we were talking about before, where your coverage area is a lot larger. 
  
And then so this is probably a very good apples to apples comparison where you could say, hey, this is a very normal use case. You have one of our receivers operating in an environment where you have a router sitting somewhere in the house, right? And you're getting inundated with this blocker power channel seven Wi-Fi streaming some video, and how do we compare? 
  
How do we stack up, right? And so here you can see that for the 12.5 MHz blocker, right? Our Series 3 versus Series 2, we have a small delta here for that particular one. 
  
It's about 3 dB delta. But nonetheless, the Series 3 and the Series 2 both perform better than the competitor A and B parts. But as you progress further and further away, for example, the 22.5 MHz, our two product lines basically get closer and closer to each other as far as blocking performance goes, but the delta stays between us and the competitor. 
  
So back to what we were talking about before, a 3 dB delta, for example, between us and competitor A, right? What does that mean to you? If you have a 20-foot radius or a 40-foot radius, let's say it's 40 feet. 
  
A 40-foot radius, that means that your area has in essence doubled your coverage area for that 15.4 connection in this case. So you can, in essence, have your nodes or your 15.4 network be spread across twice as big of an area compared to one of these competitor parts. And that's a pretty big delta as far as for our customers. 
  
And what matters here is that you're just giving the customer more reliability. You're going to have less problems in a very cluttered environment that's only getting more cluttered. And so these are things that we have to take a lot of careful design work towards to give the right amount of blocking, the right amount of when do we back off gains in the presence of these blockers, and then balance that with current consumption, because that also matters to your customers. 
  
They don't want their watch to only last for eight hours. They need to wear it for an entire day. So we are trying to balance all of these things to give best-in-class performance in these scenarios while still being able to be best in class in power consumption. 
  
We go to the next slide. And so we could sum this up on this last slide and kind of end for most of this blocking related stuff, where really at the end of the day, it's a breakdown of all of these things, and I kind of already alluded to these bullet points, where we're trying to give the best blocking performance, which really comes down to filtering, gain selection given a certain amount of power level that we're observing and things like that. And that's really what's going to drive most of our blocking performance here. 
  
But it all comes at a trade-off, and I kind of already alluded to that. It's current consumption and area/cost. And really area/cost, you can understand, we design a chip. 
  
If it takes more area, we have to pass that cost on to you guys, the consumer of our chips, and then that goes on to the customer. And overall, we want to drive a balance of area, cost, and current. To still give best-in-class performance in this coexistence scenario or any blocking scenario, really. These coexistence scenarios are just really important, and it's kind of a big interest of mine because it's starting to become more and more and more prevalent as we start trying to take multiples of these radios, get them closer and closer together, and in some cases, even co-locate them. 
  
If we look at antenna diversity, I'm not going to go into details on all the background of why you need this. But in general, what we're talking about here is two antennas that by spacing them out, you're going to get better performance than a single antenna. This is due to things like reflection, multi-path, etc. 
  
So when we talk about this, there's two antennas you use. Typically, these are at least a quarter wave apart, and I think I have some details on how far that is, and that's obviously based on the frequency. The lower the frequency, sub-gigahertz requires more separation than 2.4, which requires more separation than 5 gigahertz. 
  
And what this really means is it just allows one antenna to have a better signal than the next. And with antenna diversity, we're monitoring both of these signals during the preamble, and then we select the best antenna. And typically, you'll see about 6 dB improvement on average between the two antennas. 
  
Obviously, there's times where you're going to have a deep fade and it'll be much more, but on average, it gives you about a 6 dB improvement. The other benefit of antenna diversity is because you're using two antennas, you could actually orient them different to give you better polarization. So if you look at typical antenna radiation patterns, they're not completely omnidirectional 360. 
  
Usually, there's nulls. So by orienting those 90 degrees apart, you could sort of improve that coverage that you have because you don't know where the device you're transmitting to is going to be coming from in that null. It could be coming in a null. 
  
So again, by two antennas, I'm reducing those antenna nulls. So if this is a feature, why does chip design matter? Well, there's a couple things that are really critical for antenna diversity. 
  
One obviously is how quick can it switch antennas. So there's switching time, control latency, etc., that come in to effect. How quick does the RF settle on the front end? 
  
After it switches, how long does it take to be settled so it can start to receive a signal? And then the other important thing is how quick can it detect a preamble? And I'll go into more details on this, but it needs to be able to, if we're switching antennas and we're trying to select the best, I need to be able to listen to on one antenna, quickly switch over to the other antenna, listen, and then make a decision on which one to continue with. 
  
And then, of course, how reliable is that measurement? Because I'm measuring for a shorter period of time, I'm listening for shorter periods of time, am I getting an accurate measurement of signal strength, etc.? And then, there's things like even what's the insertion loss if I have an external switch versus an internal switch? 
  
So these are all things that designers have to think about. I will say that not every device out there supports antenna diversity. It is just usually a limited number of devices. 
  
And again, it's because like everything else, there's cost associated with this, both on the IC side as well as the implementation. So I'll try to step through this fairly quickly in the interest of time. But if we look at what's happening here, the red line sort of is that minimum line for reception. 
  
We talked about what's the sensitivity line. That's how long or how low a signal can be. And over time, signals vary due to reflections, people walking through, etc. 
  
And what you'll see with a single antenna is you'll have these areas where you have these nulls or deep fades or marginal areas, and it may not be able to receive. And in that situation, the device would have to do a retransmit on the other side. With diversity, like I said, this is going to result in a loss. 
  
But what diversity brings in is with that second antenna, because it is located a quarter wavelength apart, it's not going to have the exact same reception. It'll be slightly different. So here we see the second antenna, and we still have nulls on it, but there's different peaks and nulls between the two antennas. 
  
And by selecting between them, you can see this yellow line now represents that I'm able to have more consistent performance throughout over time because I have those two antennas. So there's a couple different ways of doing this. One of the easiest ones is select first good, and what that means is I look at the preamble of one antenna, and as soon as I get a good signal, I select it. 
  
This is easier to implement. I don't have to be as fast, switching, detecting the preamble, but there is trade-offs that that signal might not be the best, so I may be picking something that is just marginal, and then during that transmission, something happens, and I lose the packet. The select best is the better method, and really what's happening there is we're monitoring both antennas and selecting the better of the two. 
  
And there's a couple of different ways you can do this. RSSI, received strength signal indicator. Basically, that's what you'd want to choose for the most simple one. 
  
It's the fastest, and you can tolerate a little bit more incorrect assumptions or incorrect detection. We also support correlation, and this is if you're more in a noisy environment, and you need a more stable connection. This is a better way, more reliable way of doing it. And then from a device standpoint, we have a number of devices on here that support antenna diversity. 
  
We've got the Series 2 devices, which I'm focusing on. We've got the 21, 23, 24, 25, and 28. These are different products. 
  
You can see some are for sub-gigahertz and some are just 802.15.4, and then the method that we select. So depending on what you need, if it's sub-gigahertz, it's the 25, 23, and 28. And if it's 2.4, the 21, 24, and actually, I should have the 26 on here. 
  
I left that off this graph. So in either case, something we support today, we've put a lot of effort around this to do that. So let me show you an example of how we do this. 
  
And when you talk about antenna diversity, you do have to have a certain length of preamble in order to do this. So it doesn't work for every protocol out there, but for 802.15.4 is a good example because it's very popular, and the preamble of four bytes typically gives you enough time to do it. So in this example, I've got two antennas, and these antennas are just switching between themselves all the time. 
  
This device is always in a receive mode, and it's switching between antenna 1 listening, antenna 0 listening, and then it'll continue to do that until it picks up a preamble. So in this case, antenna 1 actually detected the preamble, and it's a good signal. But again, we're doing select best. 
  
So next, we're going to go to antenna 0 anyways. In this case, it did detect a preamble, but it's got a marginal signal. So the signal strength's not as strong. 
  
We look at both of these and say, "Okay. Based on that, we're going to select antenna 1 for the remainder of the packet." So we'll select antenna 1, stay on that, receive the packet, and then in general, if there's something like an acknowledgement, it'll do that on the same channel that it received on. So again, what are the implications and trade-offs of this? 
  
Why don't we just do it across the board? Well, obviously, there is trade-offs around BOM cost and product size. There's a lot of IC design complexity here that this is where I really appreciate our designers and what they're able to accomplish. 
  
There is additional components needed, obviously. In all cases, you're going to need two antennas. And in some cases, you may need some extra components, maybe a switch or something that you want to use on that. 
  
And I talked about the separation required. So you're going to have a larger product as well because you do need that quarter-wave separation. With 2.4 gigahertz, that's basically 1.23 inches, and with 915, it's 3.22 inches. 
  
So you do have some separation there. There may be some things you can do in the product design to try to reduce that, but in general, you need to get that separation for reliable diversity implementation. And then, as I mentioned, it is limited to protocols with longer preambles, and it really depends on how long it is. 
  
802.15.4 is a good example, and depending on how long that is, maybe you don't have time to do select best, so you have to do select first. Still gives you antenna diversity, but not quite as good performance. The other thing you'll typically see, too, is there will be a slight reduction in baseline sensitivity due to shorter time on the preamble. 
  
Because I'm not listening for that whole preamble, you've got to make that decision quicker, and sometimes there is a loss in baseline sensitivity as well. But again, you're getting that improvement overall about that 6 dB average, so you're better off implementing this than the loss you're going to experience by doing it. And then obviously, antenna diversity can also be mitigated by things that are good enough. 
  
Obviously, mesh networking helps like that. Things like frequency hopping, for example, BLE. BLE has a very short preamble. 
  
I haven't really seen anybody do antenna diversity with BLE. But again, because of that hopping, that sort of helps some of that. There's things like forward error correction, where if I just get a bad bit or two, I could correct, and then, of course, protocol retries. 
  
These all come at some cost, but they can help mitigate it. That's one reason I think I don't see this used as much in consumer applications because these other things could really provide that performance they're looking for. And then enabling this for our solutions, we have this enabled in our studio, our Simplicity Studio. 
  
It's really simple. For things like 802.15.4, you basically turn it on. On our sub-gigahertz, we give you a lot more flexibility because typically those are proprietary protocols. 
  
So you could tell him, "How do I want it implemented? Do I want select first? Do I want select best? 
  
How do I want the retry? Is that always on the same antenna? Is it on the last antenna used?" Et cetera. 
  
So very easy to implement these in our software. We do all the heavy lifting behind the scenes. You're basically turning it on. 
  
So a quick summary on here. Obviously, as I've talked about, antenna diversity can provide a better link margin. We see about 6 dB on my average, but you can have deep fades of 30 dB. 
  
And again, this is more important from my standpoint from where I'm seeing customers, commercial, industrial settings. We talked about the cost that it's going to add, different size, and it is limited to protocols with longer preambles. And again, BLE is typically not supported due to this very short preamble. 
  
And then, of course, product developers need to understand the trade-offs with everything. If it was free, everybody would do it. So how important is this to you? 
  
Do you want to implement it? Can you support that higher cost?