#67 | Interview, Victor Kalinin, Chief Scientist at Transense Technologies

[Music] Hello and welcome to the ETEG podcast with me, your host, Ryan Morn. I have been involved in the development of electrified vehicles and machines since 2005 as an engineer and a business leader. This podcast is the product of my passion for electric and autonomous vehicle technology. I'm here to share knowledge from some of the world's leading experts as well as my own insights. Join me as we accelerate the transition to cleaner, safer and smarter vehicles and grow the industry around the world. [Music] So I'm here today at Transense Technologies with Victor Kalenian and we're going to talk about Surface Acoustic Wave Sensors and Talk Measurement. The first thing though that I do need to say is, although this podcast hasn't been sponsored by Transense, I do work with Transense and I'm paid by them in an advisory capacity to help them with their business. But just for today, Victor's agreed to give me some of his time and we're going to talk about him and his history and background and the kind of really interesting things that they do at Transense with this unique sensor technology. So welcome Victor. Thank you. Thank you for the introduction. Yeah no problem. I wonder if we could just start with your background so tell us where you're from. Right. Well I've been working as a chief scientist and technology director at Transense Technologies for 20 years. Before that I was a university lecturer and a researcher in the field of RF engineering, radio physics and physics of Surface Acoustic Waves for 22 years, first in Moscow and then in Oxford. I have to say that I always knew that I would become a physicist, a scientist and an engineer always starting probably from the time when I was six years old kid. And what was it that kind of triggered that? Yes, I can remember that time very well, very well. And the reason for that is quite simple. This is mostly the influence of my family because both of my parents were engineers and my elder brother was also a scientist and an engineer. Besides I probably read too much science fiction when I was a boy. Okay. Definitely I preferred to read science fiction than go to music lessons. Right. Okay. Yeah. Yeah. I think a few people, a few of our listeners can probably relate to that. Yes. So but I read a proper hard science fiction, not fantasy about dragons, wizards and warriors with swords. And as a result I became an engineer and I started my engineering career in 1978 when I got my master's degree from the university and then I became a university lecturer in 1983 after getting my PhD. And that was you were in Russia still at that point? Yes. I was in Moscow. I was working and studying at one of Moscow technical universities. Right. The biggest question and I probably can't answer it for myself even now is why I decided to move from academia to industry after 22 years of my academic career. Yes. Yeah. It's not an easy question. Perhaps I just wanted to be a bit closer to real world. Yeah. And do some things that could be a bit more practical and more useful to people. And when you moved from Russia to the UK was that still, you were still in academia at that point? Yes. Yes. I basically what happened. I got a research position at the university of Oxford. But that was quite a short employment. But during that I was persuaded to take part in, well, basically to try and apply for permanent position at Oxford Brooks. Yeah. So I got it. And this result I moved from Moscow to Oxford. Wow. Okay. That's what happened. So after that I worked as a senior lecturer at Oxford Brooks for several years for six years in general. And at the same time I was a consultant for Transnist Technologies for a couple of years. Okay. And at some point they offered me a position of a chief scientist. Okay. So they lured you out of academia into the real world. Yes. That's right. Yeah. After some resistance I accepted the decision. I think while you're down after two years they eventually convinced you. Yes. Yes. So was your academic sort of life? Was that focused around RF devices or sort of sort of technologies? Was that something you were deep into? Well, not quite. Because I for one year from 1986 to 1987 I worked at the University of Oxford with Professor Lasso Sonima. And we did some research on photorefractive materials in application to optical signal processing. Face-front conjugation things like that. Okay. So that I continued doing this research when I went back to Moscow and then when I came to Oxford as well. Okay. So we published a number of papers on this subject on physics of photorefractive materials. But at the same time somehow transcends people learned that I was working on so devices in the past. Right. I did my PhD on sophisticated way physics and design of the devices. So they asked me to do some design for them. And that's what I did. That's how I became consultant. Yeah. Yeah. Gradually I moved from the field of optical signal processing back to sophisticated devices. Yeah. And I also did some work on just design of RF circuits back in Moscow. So it was quite natural to me to move to so sensing. Because it's very close linkage there with the RF side and yeah, absolutely. I mean, I could probably a good segue into, you know, what is it that transcends to? So as a company, what are you engaged with? Well, transcends was quite an attractive place to work actually. Because for me, it wasn't much different from the university research group because it was a small company with a very special creative and inspiring atmosphere, which is quite characteristic for a research and development enterprises. The nice being sort of working right in amongst the technology, not to kind of abstract. Yes. Yes. That's that's right. The company itself was founded in 1991 by two engineers, entrepreneurs. Okay. And one financial journalist. Okay. And the company became especially active in 1999 when it went public. Right. Yeah. Yeah. So transcends is a is a PLC. Yes. For anyone interested, I'll put some links in the show notes for you to sort of check out, you know, where you can find out more about it. But it is a publicly traded company here. Okay. And the main idea of the company was to develop IP in the field of subsequent equate census for automotive industry. In particular for electrical power assisted steering, e-pass and tie pressure monitoring system, the PMS. Right. Well, the question one can immediately ask is why sophisticated wave census? Yeah. Yeah. What is special about sophisticated waves? Yeah. Well, actually intensive development. Actually, there's a first of a question of what is the sophisticated wave? Yes. I'll try to answer it in one sentence. Okay. The intensive development of surfs acoustic wave devices began in early 1970s after invention of a so-called inter-digital transducer. It's a thin film microelectronic device that could convert electrical radio frequency signals into surfs acoustic waves in a very, very efficient way. What is the surfs acoustic wave? It basically looks like a seawave, but it propagates not on the surface of a sea, but on the surface of a polished solid body. Okay. Yeah. So we've got these the inter-digital transducer, which is like two coms basically. Yeah. Inter-digits like your fingers. Yeah. And We're creating a wave which. Yeah, on that, okay. Yes, due to the psilectic effect on some studies. Because the solid body which is used in cell devices is usually a psilectic crystal. Yeah, yeah, okay. Okay, so that this wave runs along the surface of this crystal and at this point we should bear in mind the velocity of acoustic wave in general and the psilectic wave as well is five orders of magnitude smaller than the velocity of light. Okay, yeah. As a result, the wavelength is over micron size. Right. And as a result, the dimensions of solid devices are quite small. Yes, yeah. Within one millimeter structure or two millimeter structure, you can achieve a very nice, very unique frequency response. Yeah. And you will never be able to achieve a similar frequency response using traditional electronic components like capacitors and inductors. Yeah. So, this miniature, very stable and inexpensive devices quickly became a dominating technology in intermediate frequency filters, delay lines and resonators. Okay. And you could find them in every TV set, in every transceiver. Okay. And in every radar. Yeah. That's what started in the '70s. Okay. And later in the '90s, psilectic wave technology became a dominating technology for RF radio frequency, front end filters and duplexes for mobile phones and bass stations. Yeah. Yeah. And so using it as a sort of electronic component like that as a filter device, that it's essentially the same technology as you've got in your sensor devices. Exactly. Yes. It's exactly. But it has become obvious in back in '70s that the resonators based on surfacistic waves and delay lines can also be used as temperature and strain sensors. It was known for quite a long time. Possibly by accident of noticing they're not quite doing the job you expect when they're under different physical conditions. Yes. Quite often, this sensitivity to temperature and strain was regarded as a parasitic effect that people tried to avoid. But also, quite a number of researchers did investigate the possibility of using them for sensing. Yeah. Well, this is why they can be used as sensors, just simply because the physical dimensions of the source of strain, as well as the source of velocity, happens to depend both on temperature and on strain. However, so strain and temperature sensors didn't find any noticeable applications in the real world till the middle of the '90s. Why do you think that was? Well, I think, and I'm sure I'm right. It was very difficult for them to compete with significantly less expensive traditional, temperature and strain sensors. OK. In their wild form, when a surfacoustic wave sensing element is hardwired to the active electronic circuitry that is necessary for the sensor, the advantages of using saw devices as sensors due to their high stability or better resolution often just don't overway their increased cost. Right. OK. That's the situation with wired source sensors. And is that because you've got the sort of, I guess, fairly inexpensive interdigital sensing unit, but then to make that work, you've got quite a bit of electronics sitting behind it. Exactly. You need to make it a part of an oscillator, for instance. Yeah. You need to include this device being at a delay line or a resonator into a feedback loop of the oscillator. And then you also need to measure the frequency of this oscillator. Yeah. And so that there is quite a bit of electronics, which is involved in the saw sensor, even if it's just a wired device. And I can imagine back in the day, actually doing that kind of thing at the right frequency was pretty difficult, you know? Yes. It wasn't straightforward. It wasn't. Yeah. Yeah. Whereas more modern microprocessors and signal processing technology probably has made that much more straightforward. You are absolutely right. Only by the middle of '90s, the progress in CMOS R/Circuit integrated circuits, I mean. And in microcontrollers and in digital signal processes, allowed to reduce the cost of intelligent transceivers to the level that sort of renewed the interest in self-sacoustic wave strain and temperature sensors. Yeah. Why did it happen? Well, first of all, everything went wireless in the '90s. Yeah. Wireless phones, wireless keyboards or modern computer, even wireless toothbrushes. Yeah. Absolutely everything went wireless. So temperature and strain sensors also tried to become wireless. But yes, you can connect a simple, a telemetry transmitter to a traditional temperature of strain sensing element like strain gauge. But in this case, you will have to provide a DC power for the transmitter. And that's sort of important, I guess, point that the common method of providing a wireless torque sensor on a shaft involves putting quite a lot of electronics on the shaft with normally some sort of battery power or inductive coupling to transmit power on the shaft. What energy harvesting? Yeah. And you've got then electronics on the shaft that are transmitting a signal off the shaft in the same way that a phone would transmit a signal. But the sort of differentiator with the SOAR technology is actually all the things that are on the shaft are passive, like-- Yes. Absolutely. The electronics off the shaft. Yes, exactly. If you use a strain gauge connected to the telemetry transmitter, then the whole device will probably weigh 10th of ground-- sorry, not even only 100th of ground. While the SOAR device itself weighs only milligrams. And besides its totally passive device, it doesn't require any DC source for its operation. And it's very well-suited for wireless interrogation, especially if it works within the frequency range from 400 megahertz up to 5 gigahertz. I want to. OK. Yeah. One of the challenges with the traditional-- I know this from the past, from my experience of trying to make strain gauges work on electric motors and things-- they don't really like background noise because of how conventional strain gauge works. How does a SOAR sensor react to background noise? I'm kind of thinking, is it a challenge with a SOAR sensor with background? Well, to much less extent than for the strain gauge. Right. The strain-- when you use a strain gauge, you rely on measuring its resistance, basically. Because in this case, you need to perform amplitude measurement. And obviously, any noise at any frequency range will contribute to the random errors of the measured strain. The SOAR device is a bit different from that. How do we interrogate passive SOAR devices, wirelessly? Let's consider a sensing element that is based on a SOAR resonator. We actually prefer to use SOAR resonators rather than delay lines because they have low insertion loss at radio frequencies and higher Q factor. So OK. So you lost me now. What do you mean by insertion loss? Insertion loss. Yeah. Well, you can apply an RF signal to excite the SOAR device. And then you can measure the response of the SOAR device. So then you divide the amplitude of the SOAR response by the amplitude of the excitation signal. And that ratio is obviously less than unity. Yeah. And characterizes the insertion loss. Usually, it's expressed in decibels. OK. So does that mean you basically don't. need a lot of power because you're getting quite a good signal back. Compared to the delay lines, that's true. Resonators have low insertion loss and at the same time they have higher quality factor, Q factor. And it turns out that the Q factor determines the potentially achievable resolution in case of wireless interrogation. So the higher the Q factor, the better the resolution of the sensing system. And what is the quality factor? What does that actually mean? Well, the quality factor determines the length of the response of the device after its excitation by a very, very short pulse. We call it delta function, the Iraq's function. So that how long the response is after you kick the device and excited. The longer the response, the better the resolution. I guess it's easier to measure and things like that. So to measure the so resonant frequency in a remote way, we use an electronic interrogation device that we call a reader. And this device sends periodically a very short RF interrogation pulse. This pulse is picked up by the sensor antenna, the simple antenna connected to the passive sensing element with the resonator. It excites the resonator and the resonator starts ringing. And this ringing continues even after the end of the interrogation pulse for a long time which is determined by the Q factor. And then this electric oscillation is being retransmitted by the same sensor antenna, back to the reader antenna. The reader's receiver picks up this signal and analyzes it, finds the frequency of the ringing, which is the same as the resonant frequency. And then it calculates torque, temperature, strain. That's how it works. So the difference from the strain gauge is that we measure the frequency rather than measuring amplitude. So you're less, basically background noise doesn't matter? Well, at least it still matters, but to a lesser extent than for purely amplitude measurements. Okay. Easier to fill, right? Yeah. What about things like electromagnetic fields? Well, it's a good question. If, well, obviously the reader's receiver has certain front and circuits, front and filter, that limits the bandwidth so that if there is a signal interfering signal outside this bandwidth, then it's suppressed by the front and filter and sub-sup-sequence circuitry of the receiver. But if the interference happens to be within the working, operating frequency range of the sensor, then we are in trouble. Okay. So the only way how to sort of mitigate this is to perform the measurement of the noise before launching the interrogation pulse. Okay. And then listen, wait and take, listen, listen, before interrogate. Yeah. And I guess you can do that cycle many, many times a second. Yes. Yes. That's right. That's right. Well, I just, another question is just popped in my head. Sorry. To, okay, hit your flow. But what about, I guess, like if you held like a regular magnet, you know, magnetic field, because it's the actual sensor, element, the interdigital sensor, that's, you know, there's no ferrous materials in that or anything, it's not going to see that magnetic field. Will it? That shouldn't affect it at all. No, it's not sensitive to magnetic fields at all. Yeah. Unlike magnetic, restrictive sensors. Right. Okay. Yeah. Yeah. It's much, well, it's not much, it's totally insensitive to stray magnetic fields. And for that reason, one of the advantages of source sensors is that they can be used quite safely near powerful electric motors, power electronics circuitry, things like that. Yeah. Yeah. Because it's always a challenge in a design of a motor and electronic system, because you do have strong magnetic fields, I mean, all over the place in the in the cabling in the motor itself, in the inverter unit. And you're going quite often you're measuring those magnetic fields to get, you know, estimate or calculate the performance of the machine, but, or the system. But, you know, the components you're using have to be able to withstand exposure to those fields, which can be quite a challenge for some, with some sensors. Yes, that's right. Yeah. The reader is not a simple electronic device. Back in the 80s, it would probably occupy the size of a good desk drawer. A desk drawer. Desk drawer. And it would cost something like several, well, maybe even tens of thousands of pounds. Okay. Right. Well, it's basically a reader. It's a very specialized spectrum analyzer. Wow. So it's not, it's nothing trivial. No, no, no, no. But due to the advances in the microelectronics, the situation is absolutely different now. Yeah. At Transense, we have developed an application-specific radio frequency integrated circuit, RF-ASIC. Right. That together with the of the shelf digital signal processor does all the work of the reader. Okay. And the reader now can fit within a matchbox. Okay. And its cost is quite acceptable for high volume automotive applications. Well, so that M is quite, obviously electronics development, it's driven so many things, but it's quite a quite a sort of graphic illustration that you've gone from something that was physically the size of a desk drawer. So maybe 50 centimeters by 30 by 15 or something. And that basically doing exactly the same job, but through the application in an in an ASIC has been able to be shrunk down with a few other components to something the size of a matchbox. So it is, that's quite phenomenal, really. Did it take a long time to develop that? Several years. Okay. Several years. And I guess so your background at that point and mixed experience with the RF electronics and with the ASICs must have come in really been very valuable to that development. Yes. What happened? We, first of all, we did quite an extensive simulation at a system level of the entire reader. After that, we designed and built a prototype on the discrete components of the shelf components. And only after that, we could specify every single component of this reader and then do the development together with a company specializing in automotive electronic integrated circuits. And now they're producing this RF ASIC cheap for us and our licenses. Okay. And it's quite interesting. I'd be honest, I hadn't realized, I probably should have. The fact that you start it from the start, it's always been about going into the automotive sector. So, because the requirements for automotive are quite demanding in terms of reliability and robustness and things like that. So because you started there, actually it probably made life a bit more difficult initially to make something that was going to do what it needed to do. But now you've got something which is very robust little product set. Yes. Yes. I hope so. I think so. And you mentioned about making it volumes because actually one of the really interesting things about Transcences, you have commercialized already in some applications. So. Yes, that's right. I will mention these applications a bit later. I just want to say now that we can now do the wireless measurement of the resonant frequencies of the SO resonator with the resolution better than 400 hertz at the frequencies above 400 megahertz. And we can do this measurement within 150 microseconds. So this is not bad at all. But being able to measure this resonant frequency accurately, unfortunately is not enough to measure strain and temperature. And the point is that the frequency of the source is native depends both on torque, both on strain and temperature. Right, okay. But what we want is to be able to measure independently, strain and temperature. So how do you sort of separate the two out? We have developed the method of that allows us to do the temperature compensated strain measurement and independent temperature measurement by using several different source resonators within one single sensing element. Ah, okay. Right. And we also developed calibration methods for this sensing element. Right. And this is a part of the intellectual property portfolio of our company. So you're able to look, presumably you look at the temperature and then you're able to sort of take that value out of the combined measurement to compensate back over for that. That's, that's, yes, that's right. That's the basic, basically this is correct. Okay. Yeah. Like, though, since we use resonators with a different design, yeah, they are affected by temperature and strain differently. Yeah. And this allows us to resolve both physical quantities independently. And that's why, because when you look at the saw element, you know, initially we describe this inter-digital sensor, but when you look at the sensor, that's not what you see. You kind of see this kind of quite, it looks like a, when you, when you blow it up, someone said to me, it looked like a video character, but it's sort of a scattering of elements inside the sensor. There's quite a lot going on in there. And that's these multiple different, so, you know, yes. Well, obviously we can't compare complexity of our sensing elements with modern processes. Yeah. Yeah. The number of elements is smaller. Yeah. We're talking about hundreds of elements, which you can find in the layout of our sensing elements. Yeah. Yeah. Still not trivial design. Yeah. And it's, it's, it's still totally, it's just passive. We're creating those elements. Actually, with a similar production method, as you would make an ASIC or a semiconductor device, but you, you're effectively fabricating those elements to micro scale in very, very thin materials in that sort of fab type process. Yes. The sizes of the elements are about one micron. Wow. And the number of elements, as I said, hundreds of various stripes, rectangles made of thin aluminum film. Yeah. And the thickness of this film should be very, very carefully controlled, because we need to achieve good tolerances. Yeah. Rather tight tolerances on the resonant frequencies. Yeah. Yeah. Okay. Yeah, I got an interesting. That's the sensing element. Another part of the intellectual property portfolio of transcends is the patterns that allow us to design sensor antennas and reader antennas that can be used on the shafts with diameters varying from 20 millimeters up to 420 millimeters. That's a big shaft. Yes, it's a big shaft. Yeah. So it's a covering, I know you mentioned initially the steering device and you probably talk about that later, but 400 mil shaft could be a very large propulsion system, you know, a ship or something. Yes. Well, yes, it can be, for instance, the output shaft of an industrial, big industrial gearbox. Yeah. Yeah. Well, one of the examples is the gearbox used on big wind turbines. Ah, okay. Yeah. And these antennas are quite, I mean, they're sort of quite clever, but in that, they're quite simple. They are pretty simple and not expensive at all. Yeah. They are based on, well, there are different designs that we used in the past, but many of them are made of a standard printed circuit board material. Yeah. So called FR4 or FR5. Yeah. And you simply have an antenna on the rotating part and then another antenna is fairly close to it, held static. For torque-up applications, we do use antennas that what we call "neafield antennas" that are positioned relatively close to each other. The gap between the rotor part and the state part is probably, well, between one millimeter and 20 millimeters. Okay. To quite a, it doesn't have to be like a half mill gap. And I guess it doesn't really matter if that gap moves particularly. Well, to some extent, well, definitely it's much less sensitive to movement than the pickup coils of magnetostrictive sensors. Right. Okay. We still need to make sure that the reader antenna doesn't get too close to the sensor antenna. Right. Because in this case, we will face quite noticeable rotational errors. The measured frequency will start depending on the rotation angle, which is an undesirable thing. Yeah. So it actually pays. So we want them further apart. Yes. We'll see close together. Yes, but to a certain extent. Okay. Because since it's a near field antenna, if you move it too far away, the signal strength will be too small. It loses its area. So there is a certain range for the sort of axial movement of those antennas. That's really interesting. Because a lot of times you think of, you know, when you're designing a system with a rotating shaft, you want to allow for movement on the shaft and, you know, vibration and things like that. But it sounds like that shouldn't really be a problem here. Well, it's much less of a problem for so senses. Because as I said before, we rely not on amplitude measurements, but on frequency measurements. And the fact that the signal becomes weaker doesn't. Well, it doesn't matter to that extent. Well, clearly, when the signal is too weak, the random errors in the measured frequency will increase. But it doesn't affect the measured value itself. See that? So it's always self-compensating. Well, since our sensing element is well suited for strain measurement, we can also measure any physical quantity that can be converted into strain in the sole substrate. And that's where you come into talks. That's right. But we also design. Well, how do we convert? The physical quantity into strain, well, it depends on the physical quantity we measure. For instance, to measure pressure, we developed a special metal-can, metal package that deforms the sole sensing element, depending on the pressure applied. Yeah. This way, we designed a pressure and temperature sensor for type-resure monitoring system. And we also developed what we call an all-quarts package for the torque and temperature sensing element, which is particularly well-suited for measuring shear strain that is generated on the surface of the shaft when the torque is applied. So it reminds you of that. So the pressure sensor actually is kind of like using a mechanical element to put a load onto the sole sensor to create the strain. Yes. So a little diaphragm or something? Absolutely. That's right. We have a diaphragm which presses upon the sole device deforms it and changes the frequency of one of the resonators. Yeah. There are in total three resonators inside. Well, this is not one possible approach. Other approaches can also be used, for instance, to achieve a good high pressure sensitivity, we could have etched the thin diaphragm in the sole substrate. I am. But at the time when we did all this development, etching diaphragm in quartz was not a standard operation. So to keep the cost of the sole sensing element low enough and acceptable for automotive applications, we decided to use mechanical transducer that converts pressure into deformation of the sole sensing element. And I guess something is just thinking about that. Probably the advantage of doing that is in most pressure sensors, you would have a diaphragm again, but typically, to be sort of guaranteed that it's completely himetic is quite difficult. And if the diaphragm ever breaks and so. service, you might have a leak and sort of issues can come from that. But with the SOAR device, you can have the full SOAR unit on one side in the pressure area and then all of the electronics and the reading stuff outside in the non-pressure area. And if the diagram fails or there's an issue, it doesn't matter. There's never ever any possibility of leakage or failure from that unit. Well, I wouldn't say so. Okay. Well, the point is that the SOAR device in our TPMS sensor is enclosed inside the fully hermetic metal package. So the hermiticity of that metal package is still important. Yeah. Because we measure actually the difference between the outside pressure and the pressure which of the air inside that cavity. Right. So that the hermiticity is still important. But providing hermetic seal is not that difficult because we just lays the weld, the diaphragm, to the metal base, holding the SOAR device so that it's all hermetic sealed. Yeah. Very nice. Okay. And as a tire pressure sensor, what benefit does that bring? Well, the main benefit is that it's a battery less sensor. It doesn't require battery. But the current, currently dominating technology in TPMS is battery powered telemetry transmitters. So, and it's, it does dominate everything. And at the moment, it's not feasible to sort of commercialize TPMS sensor for passenger cars. For some time, one of our licenses manufactured this kind of sensing systems for motorsport. I guess on a passenger car with the batteries lasts, you know, as long as the tire lasts sort of thing. So people, yes, there's an opportunity to replace the batteries when tires are changed and things. So the durability, life cycles, yes, not all that. Yes. That's right. For motorsport, it's critical to have as small weight of the sensor as possible. And also our sensor allows to read the pressure every millisecond, or every 10 milliseconds to be more exact. Yeah. While the passenger car, TPMS sensor, transmits the pressure information and temperature information once, I don't know, every 10 seconds. So you get very high, very high frequency measurement, a lot of data. So in some applications, I can imagine that would be really important. And having that extra level of data and durability would be an advantage, but not to mention the passenger car feels. No, no. Okay. Interesting. Yeah. Well, that's the current situation with TPMS, but truncence as a company is more focused now on torque measurement, torque and temperature measurement. Yeah. And well, originally we developed this sensing element for electrical power assisted steering system for e-pass. Moving on then to torque sensing. Yes. Yeah. So as I said, the torque sensing element was originally developed for e-pass for electrical power assisted system. But obviously we can measure torque applied not only to steering shafts, but also to any other shafts starting from relatively small input gearbox shafts for sports cars to quite big shafts of gearbox, industrial gearboxes or wind turbines or marine prop shafts. Yeah. Yeah. It's interesting because you know, torque quite often in a development environment. We spend a lot of time and effort very accurately or as accurate as we think we can do measuring torque to validate a product and sort of validate a design for a system. But there aren't many applications where we're able to embed a torque sensor into the product when it's gone into production using traditional technology. Like it's quite difficult to do and not all that common, but it seems like the transcense technology allows you to do that potentially. Yes. From the very beginning, from the point when transcense was founded as a company, the aim was to develop something, some torque sensor that could be used in high volume applications in production units rather than just for distinct purposes. Yeah. And I guess power steering the E-PASS unit is one example where today there's a few different ways of doing it, but measuring the torque is necessary. So that is an active field of development. One thing that always sticks out on the E-PASS, obviously it's very safety critical. So you want that steering system to be completely robust, reliable. Was that a challenge? Yes, definitely. It is a challenge. Currently, the torque sensor which is used for E-PASS is often based on a so-called torsion bar that makes the shaft torsionally compliant. To increase the twist angle, because traditional sensors based on hall effects, hall effect, or based on optical measurements, they measure the twist angle. In order to make this device safe, the quite a tricky mechanical design of the shaft is used. Well, and actually even sometimes, so that twist is something that you need a enough twist that you can measure, but not so much that the driver can feel it, or it hinders any kind of vagueness into the steering system. Yes, it's very challenging. Yes, it is. What advantage does the torque sensor give us? The torque sensor based on so reasonated happens to have the resolution, strain resolution, and order of magnitude, higher than the typical strain resolution of a traditional thin film strain gauge. As a result, we can build a torque sensor for E-PASS on a stiff shaft. Right. Our torque sensor for E-PASS is non-compliant at all, which is an advantage for certain types of vehicles, for instance, of the road vehicles. Where they're seeing very high impact loads on the steering. I think heavy-duty commercial vehicles as well, that kind of application. That's interesting. Yes, and the fact that we can measure pretty small values of strain helps us to achieve sufficient safety factor for the steering shafts. And obviously, the shaft design itself is quite simple. So that's one advantage. Obviously, another advantage is that it's a batteryless sensor. It doesn't require any active electronics on the shaft, so that there is no need for changing the batteries or providing expensive devices for power transfer to the shaft. It doesn't require any rebalancing after instrumentation. Since this device is very, very light, it can work at speeds, at rotation speeds, up to and above 20,000 RPM. One of our licenses uses so-senses for torque sensors on the shafts of IVE Unique turbine engines, where the rotation speed is pretty high. And the sensor that we developed can work within the temperature range from -40 to +150 degrees centigrade, and provides the accuracy better than 1% within the whole range. So, not only are you able to get a very high sampling frequency out of it, so you can get a lot of data resolution, which if I'm steering or propulsion, I would think would be essential. But you're also able to do very accurately as well. So, you know that that number that you're getting is a solid number. You can really rely on that. So, from a system control point of view, that's going to be critical. Yes, yes, that's right. Well, as we discussed before, it's totally insensitive to magnetic fields. And well, you mentioned the high update rate. Yes, it is. Our standard reader provides the update rate of 6.7 kilohertz, which is sufficient for most of industrial and automotive applications. And It's just about sufficient. Yes. Well, it allows us to monitor quite fast torsional vibrations as well. Yeah. Well, you know, one of the challenges that you-- I've certainly seen in the past, designing control systems in vehicles, is actually you end up having to do so much signal processing, filtering to take out noise that you really-- you actually-- there's a fine balance between seeing noise in the system and getting false readings and getting valid useful data out. And you end up at quite a slow measurement rate because of that. So you have to be quite careful. But you're not going to hit that problem with the sort of ice. You're just able to get such a quality stream of data coming out that you really could-- you could use that in an active control loop, for example, which-- That's right. Well, basically, the EPUA sensors are used in the feedback control loop indeed. And also they can be used for condition monitoring of various systems. And they provide lots of information. It's a big question how to extract the most informative parameters from the torque signal. Because obviously, we can't keep transmitting this stream of torque values. So that-- but that's more about the condition monitoring algorithms. We can provide enough information for condition monitoring. Yeah, I should-- Because you'd be able to detect-- well, quite a lot of things going on in there. So like, premature bearing failures, because you'd have a vibration signature off that. But you'd also potentially have things going on with system torque because of the extra friction that you could detect as well. Yes. OK. Although transences mainly involved in torque measurements, our wireless source-trained sensing elements can also be used for non-contact measurement of axial load. OK. And bending and bending vibrations if needed. Yeah. So that's the situation. What do you think about-- so my particular area of interest is around electric motors. And I've looked at this. I think, wow, really in the past, getting like a temperature reading off the rotor of a motor would have been quite helpful. It's quite difficult to get into the rotor normally. Do you think it would be possible to use a source sensor to get a rotor temperature reading? Definitely. As I mentioned before, the temperature reading comes as a byproduct of the temperature independent measurement of torque. So that our reader provides information about the torque and temperature within one bucket. And if we had a motor-- you know, if you think of it like the-- normally you have the rotor on a shaft, could the sensor be quite close to the rotor? Or does it-- you have to space it away to get a good torque measurement? Well, the only limitation is that the position of the sensor should be in the place where the strain does represent applied torque. So that's the only limitation. How close we can position the sensor to winding, doesn't matter? Yeah. It can be as close as the mechanical design allows. But we need to place the sensor where the strain linearly depends on the applied torque. OK. Or proportional to torque. You'd have to examine the shaft and make sure you correctly placed it. But as long as it was on the shaft, and that all agreed, then-- so you could definitely measure temperature on the shaft at the very least, which would give you a good indication of the rotor temperature. Yes. From a torque measurement point of view, you're doing that with always with differential strain. So do you always have to look at what's happening streamwise on either side of the shaft? Or not always. Usually, we install two sensing elements on two opposite sides of the shaft. The main reason for doing this is to cancel the influence of potential bending moments that can be applied to the shaft together with torque. If we don't want those bending moments to sort of distort the value of strain, then we need to cancel this influence by averaging the readings coming from the two sensing elements. OK. So that's the only reason you have two. Another potential reason could be some sort of degree of redundancy. Right. Yeah. OK. Actually, I mean, a shaft sag in an electric motor is a known phenomenon. I guess that would be that kind of thing that bending strain would be those sort of loads on the shafts. OK. That's it. One other thing-- so in the motor system, we always have a speed and position sensor, because to control the motor, we need to know where it is. So we know when to fire the phases to make it go. And is it possible with the RF coupler to do something along the lines of speed position measurement in there as well? Yes. It is possible. And it's actually being done for the E-pass sensor. The coupler that is used in that system is made only standard PCB material, as I mentioned before. And the same rotor coupler can also have a copper target, which is working together with the coils positioned on the state RF couple. Yeah. So this arrangement allows us to add an inductive absolute angular position sensor into the torque sensor, so torque sensor. So that two devices will work together quite happily. Yeah. It's a bit potentially, because the normal is some sort of hall effect optical, I guess, like the E-pass system. So a hall effect rotary encoder is quite a common thing, or optical less common. So we potentially could replace that rotary encoder with-- With the inductive angular position sensor. Right. OK. The accuracy will be around one degree. Another, just again, wondering when an electric motor-- when the phases-- you're firing the phases on the motor at quite high frequency. You've got a three phase motor. You sort of-- shove, shove, shove, shove. Not quite like that, but it's an impulse reaction. Would you be able to sample the torque to the extent where, let's say, a phase was off or a phase was not quite right. Do you think you'd be able to measure that kind of torque imbalance in the system, or-- Well, the only thing I can say is that when we developed a torque sensor for kinetic energy recovery system for Formula I-- Oh, OK. That's an interesting example. The torque sensor was installed on a short shaft, connecting the electric motor to the crankshaft of the IC engine. Yeah. And we were told by our customer, what sort of maximum torque we can expect in the system. After the first testing, it was absolutely obvious that the torque was at least doubled. Right. Because of a very, very short sharp pulses, torque pulses that our sensor measured. Right. We decided that these pulses actually come from the electric motor. OK. Because there is no other way. They were not firing pulses of the IC engine. Yeah. So we'd be able-- we'd have a data stream where we could actually basically verify down to a phase firing level, potentially, that the motor was working properly. I hope, but it all depends on the frequency of those pulses. Yeah. If it's in the right kind of domain, I guess-- I mean, you've got some quite high-speed motors being used at 20,000, 25,000 RPM machines. So you can imagine, depending on the pole, the pair combination, you'd be-- of firing those fairly quickly. But yes. Well, in many cases, I think we should be able to resolve pulses. We can also resolve individual firing pulses of internal combustion engines. Well, OK. Right. So I think I just noticed we're kind of running out of time quite rapidly. So just to start to wrap up then, it's been really interesting. And actually, I mentioned the start. I do do some work with transcents. But we've not really had an opportunity to properly sit down and talk about what the sensor can do or can't do. So it's been absolutely fascinating. Thank you for your time. And for explaining it so wonderfully, it's helped me. And I'm sure help the listeners understand a bit more about saw, sensor technology, and what you can do, and what you can't do with it. Just to close up a question that I basically always ask people is looking at the business and the market and your technology. What do you think the trends are going to be over the next few years? And what are you excited about? What's coming that's got you excited? Well, I very much hope that we'll come to the point when we can see, yes, our sensor has found applications in the products which are being produced in not in hundreds, but in at least tens of thousands units per year, if not millions units per year. That would really excite me. Just the scale of application, that would be really great to come to that point. Yeah. And it must feel-- does it feel like we're kind of getting there in after this long journey? Well, I hope we are pretty close to it. Yeah, yeah, brilliant. OK. Thank you. Thank you, Ryan, for the nice opportunity. Well, close there for the day. So thank you very much for taking the time out to listen to the podcast. I really hope you enjoyed that. We've got loads of more exciting episodes coming on electric vehicle and autonomous vehicle technology, powertrain components and systems. So a number of episodes ready. Back to the usual formats. So don't forget, if you're working in a really exciting field in electric and autonomous vehicles and you want to talk about what you're doing, you can get in touch with me. You can find out how to do that down the show notes. This is one of the first episodes that we've recorded under the new brand. So you'll notice that as well. It's really exciting to be kind of relaunching under the new brand. So it's fantastic. So if you can make sure, if you've enjoyed this, hit like or hit subscribe, depending on the platform you're on, and leave us a rating, it really helps to get the show in front of some more people as we build that audience and help to share the knowledge and experience from the people that I'm talking to. So thanks again. And I really look forward to speaking to you again soon.