The Next Generation of Magnetic Sensing: XtremeSense TMR Technology: Tech Chats | Mouser Electronics
(sparkling music) Smart watches, smart cars, smart homes, modern life and industry are increasingly reliant on accurate and efficient sensing technology. For decades, designers have sought solutions that are smaller, more sensitive and energy efficient. Today, from massive wind turbines to tiny digital pills, tunneling magnetoresistance sensors have emerged as the answer. Compared to traditional magnetic field sensors, TMR sensors offer several key advantages. They are more accurate, consume minimal power, and are highly resistant to temperature fluctuations. But what makes TMR sensors truly remarkable is their broad range of applications, from automotive and wearables to space exploration and medical devices.
Coming up in our "Tech Chat," we'll discuss the next generation of magnetic sensing, XtremeSense TMR Technology. Welcome to "Tech Chat," sponsored by Mouser Electronics. On "Tech Chat," we meet with engineering experts to learn about the latest technical innovations that are shaping and reshaping our world.
Today, I'm happy to welcome Ali Alaoui, a product line manager at Allegro MicroSystems. Welcome to "Tech Chat," Ali. Hey, Dale, thank you for the time. Happy to be here.
Well, we're glad to have you here. What are we going to be talking about today? So today I wanna give you an introduction to XtremeSense TMR Technology and talk you through this new magnetic sensing technology. Well, that's great. Let's jump right in. So I wanna hit three topics today. The first is why TMR, why we are, as a company, excited about TMR and where we think it's interesting. The second is, what is TMR? I wanna give you a high-level introduction to this technology and spend a little bit more time on XtremeSense TMR specifically which is our own implementation of TMR.
And then lastly, I wanna talk about products, you know, give you a high-level overview of how TMR is leveraged in these products and how they impact real-world applications. So, why TMR? This is at the high level. TMR, just like Hall effect, is just another tool in our toolbox. So first of all, I know we're gonna talk about TMR, but just to say Hall is not going anywhere. It has proven itself as a technology over the past decade, and we see it still growing. However, as we look forward, we're convinced high-performance magnetic sensing is gonna go to TMR.
Just to walk through some examples here, I've listed, you know, three. If we look at current sensors, for example, you know, we see 20 times higher resolution with TMR while achieving high bandwidth, five megahertz and beyond. And this is critical, as you know, in many power applications, especially as we go and transition to higher switching speeds to GaN and silicon carbide. The second, if we look at angle sensors, we see up to eight times higher resolution at high RPM. And again, this is critical in ADAB systems, especially, again, as we move to precision machines, autonomous machines, like cars, autonomous cars, robots, et cetera.
And lastly, I just wanna, you know, highlight our 1D and switch portfolio here, because with TMR, we are able to go below the one microwatt threshold, power consumption threshold. This just opens new possibilities for industrial IOT for consumer applications and medical. So if someone has been using Hall effect sensors, and they're thinking about making the change to TMR, is it pretty easy? Is it like a drop-in replacement part with better performance? How does that work? That's exactly right.
It's a drop-in replacement, especially in our switch products, in our current sensors, in our angle sensors. We try to make it as easy and seamless for our customers to, you know, transition between our portfolio. Well, let's learn some more about TMRs. All right, so I'll start with the smallest element, which we refer to as a TMR stack. A TMR stack, and we call it a stack because it's really just a stacking of different materials when we manufacture it. A TMR stack is basically three main components.
It's a reference layer, it's a barrier, and it's a sensing layer. The reference and the barrier are ferromagnetic. That just means that they have magnetic properties as materials. The barrier is a little special in TMR.
That's where the T of tunnel comes from. It's a thin insulating layer that is non-conductive, non-magnetic. The MR piece, which is magnetoresistance, refers to the ability of this stack to change its resistance, electrical resistance that is, when a magnetic field is applied to it.
At a fundamental level, this magnetoresistance, this ability to change the resistance under a magnetic field when exposed to a magnetic field. This is basically what is leveraged in all of our TMR sensors today. And I put a picture on the left, I'm sorry, on the right side with the red circle. And that's just to show to the viewers what it actually looks like inside one of our wafers.
So once we have a stack, we move into a TMR resistor. And the TMR resistor is basically using multiple TMR stacks in series and parallel to form one TMR resistor. The number of stacks that form one resistor, again, will vary from hundreds to tens of thousands of TMR stacks per one resistor. And again, that depends on the application and the specs we're trying to target.
Once we have a resistor, we then move into a transducer. And this is really the building block, let's say, that will go into our final TMR sensor. So TMR transducer in our case is typically made up of four TMR resistors that form a full bridge or a Wheatstone bridge.
We do this for mainly two reasons. Let's say the first is higher sensitivity, and the second is temperature performance or temperature drift performance. And I put a picture there on the left side just to show what it looks like inside of an ASIC.
The rectangles there are basically TMR resistors. So we have eight resistors in that picture, for example. So is the TMR then fabricated directly on top of the ASIC as part of the CMOS process? Or is it bonded on after the fact? So that's a very good question, and that's something that differentiates Allegro from, let's say, other TMR suppliers where our technology is what we call back end of line compatible POL. That means that we deposit our TMR during the standard CMOS manufacturing process. And so to answer more specifically, the transducer is actually between metal layers inside of the, you know, our wafer.
So then as I look at this TMR transducer, you say this kind of a basic element, do all of the TMR sensors respond the same way then? They don't, and we try to optimize the transducer to the application we're shooting for. And so here I'm gonna take an angle sensor as an example. So in an angle sensor, ideally, we wanna look at the angle or direction of the external magnetic field. And we do not wanna measure the amplitude of that field. Right, because the pure 2D angle does not look at amplitude. Only looks at the angle of the magnet relative to the sensor.
This is very fundamental to TMR. This is something that TMR can do which is very different to other technologies. And so just to show what this looks like. On the top right graph, it's a little bit technical, but I'll try to walk you through it. So the y-axis is resistance. The x-axis is the field. And as we increase the field, you'll see that there's this high slope at the beginning where the sensor saturates very quickly.
Once it's saturated, as we keep increasing the field, obviously the sensor is saturated so we don't have any more, call it sensitivity. However, if you look at the green areas of that graph that we highlighted, as the angle changes, the resistance of the stack changes. This is what we use to make our 2D sensors here. And so if you look at the bottom left graph, you'll see how at the top of it you'll see a TSSOP, let's say, a package with a magnet on top. And as we rotate the magnet, we create a sine and a cosine output out of the sensor. And here we're talking about the transducer only.
We can achieve very high sensitivity of two volts and beyond. And the operating region goes from 20 millitesla to 90 millitesla or 200 to 900 gauss. And this allows us a very wide operating range. It helps with the applications. And we're gonna talk about this later on.
So this is the angle sensor. I'm gonna move to linear sensor now where we're trying to do the complete opposite here, right. Because now we wanna measure the amplitude of the field, and we especially don't wanna look at the angle or not be sensitive to the angle. So if there is a cross field or an orthogonal field to the direction of sensing, we don't wanna capture that field.
And so here I'm showing a magnet that moves up and down. It can also be current. So we can sense, you know, the magnetic field generated by a conductor. The graph I'm showing on the bottom left shows a linear region. So that shows you a response of the TMR stack.
The y-axis, again, is normalized signal out of the transducer. The bottom, the x-axis, that's the external magnetic field. And you'll see this linear region in the middle that has the very linear (indistinct).
This is what we leverage in our linear sensors. And one thing I need to mention because here we're showing, like, plus or minus 400 gauss. Now this, again, can be tuned. So we can make this slope very sharp, very high in sensors where we know in the application we're not gonna go to very high fields. And of course we can also make it, you know, wider and so capture more magnetic field. So let me see if I understood this correctly.
So this linear center is operating in a linear region. You're not saturating. But the angle one was operating saturated, was that correct? That's correct. So this one operates in the linear region, and then it saturates. While the previous sensor we looked at, the angle sensor, operates in this saturation region. Saturation just means that increasing the amplitude of the magnetic field will not change the output.
However, by changing the angle, so this is why we have this animation here that's showing in the sensor layer how the magnetization of the sensor layer is changing, and that angle, so between the sensor layer and the reference layer, that angle is what will change the resistance of the stack. So now I'd like to transition to XtremeSense TMR. I'll spend here just one minute on XtremeSense TMR. And the reason we call it XtremeSense is just because we wanna differentiate our own implementation of TMR from what's out there, what's on the field. And the reason is because we believe that to make a good high-performing TMR sensor, we need to have at least four aspects covered. So the first is the TMR stack.
The TMR stack includes the material characterization. So at Allegro we have a team of R and D physicists who develop all of our TMR stacks. It includes micromagnetic modeling. This involves actually some in-house tools. Process development is a big one. So even if Allegro does not own and operate its own foundry, we have processed some of the best talent in process development to make sure that the TMR process is suitable and it is to our spec and our design.
The last point there is we talked about earlier, the back end of line compatibility. This allows us to put our sensor, you know, within the middle layers, which makes our sensor monolithic. Then the transducer design. So here we have to develop, you know, specific design tools.
So you can think of them as a TMR PDK. That is our own IP, our own ownership here. And that again allows us to from, you know, day one, design a TMR transducer that we know is gonna perform once manufactured. Moving to TMR test, as you know, at Allegro, we've been doing magnetic sensing for a while now.
We have all this capability in-house and that's really great and helpful for TMR also in terms of magnetic test. Pinning optimization, this is new. This is something that is needed for TMR that we don't typically obviously see in Hall. And so pinning is the ability to design a transducer, which means all the four resistive elements in that transducer need to have different sensitivity sine.
So when the magnetic field is applied, one resistor goes up in resistance, the other one down in resistance. And that ability comes from the pinning optimization. Finally, the TMR sensor is putting all this together with a circuit, with an ASIC.
And so here three things come to mind. The ability to make monolithic sensor gives us this really great leverage in terms of die size, in terms of ease of packaging. So we don't typically have dual dies in our sensors, etc. There's some TMR specific design IP here also. So in terms of how do we compensate for, you know, temperature drift and other drifts, let's say, in the TMR sensor.
And finally packaging. As you know, Allegro has been an innovator in this area with our current sensors, and that's something that we bring also to the TMR part. So this is why, you know, just a high level description here of XtremeSense. Well, it really helps to see everything together and how it all fits together to make a quality product because, you know, each of these portions of it are really important in fitting together. So thanks for that overview.
Absolutely. So I would like to spend some time here just going over four different product categories and hopefully giving you just an understanding of how we see TMR improving the applications in these domains. So I'll take the first one here for current sensors as they are applied in power conversion devices. So what we're seeing is an acceleration in the pursuit of efficiency, going for low weight, less cooling, small packaging, lower cost ultimately. And one way to get there is with faster switching speeds, as you know, to shrink the passives, which, you know, means moving to GaN, to silicon carbide for those low switching losses.
So with this now the application needs high-speed and high-precision current sensors. There's really no way around it. We need to have these high-speed, high-precision current sensors because there's, you know, small passives and a current sensor feedback basically, like, mimics the large passive.
The other critical aspect also is protection. We sometimes skip on this. So protection means that, you know, when we have a high-speed switching device, we need that high-speed current sensor to be able to protect it, not only for reliability questions but also to avoid oversizing the component for that, you know, rare overcurrent event. This also helps in terms of cost. I tried to illustrate this in the picture on the bottom left.
And so here I'm just showing two TMR-based current sensors called, you know, two families of current sensor, CT43x and 2x. These are one megahertz bandwidth devices, and we're comparing them here to one of our, you know, own Hall sensors. And just to show you how the signal looks like, the raw signal looks, like, out of an oscilloscope measuring a 50-hertz signal. You know, you can see how the TMR sensor, when we say, you know, has high resolution, less noise, this is basically a picture hopefully that shows that. Ali, is there a frequency where designers should really think about making that switch from Hall to TMR sensor, kind of a rule-of-thumb when the switching frequency because you're switching to GaN and silicon carbide, they're, like okay, you really need to think about moving on from Hall to TMR? That's a good question.
So I'd say it becomes easier as we cross the one megahertz threshold. So at one megahertz or above, I would say the designer needs to look at TMR. I would say number one TMR in my opinion. Below that I'd say above, you know, the 400 kilohertz threshold TMR can start to become very interesting. We looked at, you know, current sensors.
This is also, we call them integrated current sensors because current goes inside of, you know, the package. Now, if we look at high current now. So if we look at what we call field current sensors. And so here, you know, historically, field current sensors always used a magnetic field concentrator, either a C-core or a U-core, right, to concentrate the field on the sensor. Ultimately, a no-core, a coreless sensor would be better.
Historically, again, SNR, or signal to noise ratio, was the biggest problem because once we remove the concentrator, the signal to noise ratio just becomes not too good, let's say. And these coreless sensors were always a little bit, you know, limited in their applications because of this. We think TMR removes this SNR compromise.
We see 15 to 75x higher SNR between a coreless Hall or a coreless TMR. And so here I really wanna highlight the CT455 and 456. These are products that we just released, I think recently, a couple weeks ago. These are super small, really low-cost way of adding current sensing capability at that 100, you know, and above amps on a PCB. And the way we think about these sensors is for control and protection.
This is probably by far the most efficient way for protecting hardware from overcurrent and improving reliability because for two reasons. So we get the high SNR, and we get also very fast response times. And I'm putting here an oscilloscope picture just to show the below one microsecond response time that we claim.
I kinda wanna highlight that because it's a little small on the screen. But that's like a really crazy fast ramp to go from zero to a 100 amps in less than one microsecond. Oh, absolutely. That's not something that would be normal during an operating time, let's say. But in a short circuit event, I think that is something that we will see. Okay, so switching gears completely, moving to angle sensors.
Again, here we see at a very high level, TMR brings two really essential benefits to electrical motors. Number one is reducing constraints. And number two, it improves control. So I'll talk to each one. So in terms of reducing constraints, here we talked about it before during the, you know, the angle TMR slide where we talked about how we are insensitive to the amplitude.
Well, in an application, that basically means air gap, right, in an angle sensor application. So this means mechanical tolerances are less constrained in the TMR-based angle sensing system. So that includes not only air gap, includes tilt, misalignment.
It also includes any magnetic tolerances from the magnet manufacturing. These are all problems that we see on the field that TMR is able to solve. And these tolerances are also, as you know, some applications, and we deal a lot in automotive.
And so having, you know, tolerances over lifetime in NVH, this is very critical to us. So number two, in terms of control, and I tried to put some items here on on the slide. So TMR brings a high resolution over high RPM, low latency and angle accuracy over temperature lifetime and the mechanical tolerances we just talked about. These are all elements that basically together allow for, in our view, state-of-the-art, you know, position, torque and speed control.
And again, the reason for having a more efficient control at the end of the day is really to improve the efficiency of the motor, you know, reduce audible noises and torque ripple vibrations. And at the end of the day, reducing just the volume, the size and the cost of the whole system in the application. And I would think that there's a downstream benefit, you know, thinking about the lifetime operating cost of these devices. If you have smoother torque, you have lower vibrations, then that's likely to improve the lifetime of the system, reduced maintenance from, you know, vibration is just hard on things. That's absolutely correct. Angle sensors is area where we see some of the highest and fastest adoption of TMR sensors, and that basically proves your point.
Good sense. The last product category I wanna talk about is switches. This is also something where TMR brings a lot of value, and it brings a lot of value because, again, it removes an old compromise that we used to make with Hall. Typically, if you wanna make a low-power device, you typically go for a low-power, you know, process at any foundry. Usually, low-voltage and low-power processes are not the best for a good Hall transducer. And this is where TMR is interesting because TMR is basically process agnostic.
It goes on top of the process, right, as we discussed before in metal layers. So it's not really sitting on the silicon. So that allows us to use the best process for the application. And so if we wanna go for a low-power switch, we can just go for a low-power process. So that's number one.
Number two is, again, the high signal amplitude, let's say, that we get, or the high sensitivity that we get from TMR. So TMR, as you know, in these systems typically, we would put the system to sleep and wake up for a brief period to just sample output and then go back to sleep. And TMR does not require a lot of circuitry for this to happen.
And so the sample time, the wake up time is really very small. We're usually talking microsecond or way smaller below that. And this compared to Hall, as you probably know, needs a little bit more circuitry, a little bit longer time to sample, you know, correct, compensate, et cetera. And the result here is basically an extremely low-power switch. And this is ideal for battery-powered applications, and I put some of them here in pictures, especially with the CT8132. And again, we see traction here in industrial IOT, so, you know, robots, warehouse applications, autonomous robots, as I said, consumer and medical applications.
And I would think because this application is so much different than the other two that you just talked about, that some of the capabilities you mentioned before about doing the materials characterization, the modeling, designing the transducer in-house, the testing, all that really becomes important because you're doing something, I'm guessing these sensors are quite a bit different than the ones for power supplies or motor control. Absolutely, yeah. So the process, the TMR transducer, everything is different from these three sensors, for example, we talked about today. Current sensors, right, because we want to have very high bandwidth. For example, we would try to, you know, have a high SNR, and we can operate on a higher power level that allows us to, for example, reduce, you know, our TMR transducer resistance, for example. So instead of, you know, we just reduce the resistance to enable higher bandwidth.
Whereas in a switch, we can also, I talked about process, but we can also increase the bridge resistance. That's another way to simply reduce power consumption. But the drawback is lower bandwidth. But we typically do not see or require inventory need for a one megahertz switch. Just goes back to your point you correctly stated.
These three sensors are very different, very specific to the applications they're going after. Well, that's very interesting. Before we end today's discussion, can you provide some kind of key points for our audience to consider as they're thinking about their next system design that would require magnetic sensing? Absolutely. So when designers are looking into their next project, we really wanna encourage them to look at TMR and how it can benefit their next designs.
We talk today about why TMR, why we think this is a really interesting technology, and we talked about resolution for current sensing and how it enables faster switching in the next generation of power conversion devices. We talked about resolution in angle sensing and how it can improve, you know, resolution and angle accuracy in the motor space, in the electrical motor space. We talked also about switches and 1D linear sensors and how we can reduce power consumption and enable battery-powered applications.
The second topic we discussed also is what is TMR. And so I tried to give a brief introduction to TMR and more specifically, spend a little bit more time on the differences between, you know, an angle sensor, a linear sensor, and how at Allegro with XtremeSense TMR we are specializing, you know, these sensors to the application. You know, we have more than a decade of experience in TMR today, and hopefully that's something we can benefit our customers with. Lastly, the product. So we touched on different product portfolios and few product references for high bandwidth integrated current sensors, like the CT34X and 2X. We talked about field sensors.
We just, again, released these products a couple of weeks ago. They're available on Mouser already, I believe today. And so these can give customers the ability to go and enable a high current. We're talking 100 amps and and beyond, a very fast control and protection ability.
And then we talked about switches. Here, the benefit is gonna be around power consumption and precision for magnetic field detection. Well Ali, thanks so much. That was really educational to me 'cause I did not know much about TMR sensors. And so I learned a lot, and I would think our audience did too. All right, thank you, Dale. Thank you.
It's been great. Thank you so much. Yeah, and before we sign off here, we want to thank our friends at Mouser Electronics. If you're looking to purchase any of these Allegro MicroSystems products or to learn more, head over to the link you see on your screen now and then help them continue to support educational presentations like this one. And join us again next time on "Tech Chat" where we chat with experts like Ali from industry leading innovators like Allegro MicroSystems, who are changing our world every day. (sparkling music)
2024-11-17 16:13
