SemiWiki.com - Podcast EP345: The Impact of the New proteanTecs PVT Plus Sensors with Nir Sever

Episode Date: May 8, 2026

Daniel is joined by Nir Sever, Senior Director of Business Development at proteanTecs. Nir has over 30 years of experience in advanced VLSI engineering. Before joining proteanTecs, he served for 10 years as the COO of Tehuti Networks, a pioneer in high-speed networking semiconductors. Prior to that, he served for 9 years as Senior… Read More

Hello, my name is Daniel Nenny, founder of semi-wiki, the open forum for semiconductor professionals. Welcome to the Semiconductor Insider's podcast series. My guest today is NIRS server, senior director business development at Proteontex. NIR has over 30 years experience in advanced VLSI engineering. Before joining Proteontex, NIRServe for 10 years as the COO of TAHO of TAHOD networks, a pioneer in high-speed networking semiconductors. Prior to that, he served for nine years as Senior Director of VLSI Design, and technologies for Zaran Corporation. Welcome to the podcast, Near.

To be here, Daniel. So first, can you tell us what brought you to Proteontex? Yeah, it's a nice question, Daniel. So, yes, I joined Proteontex about six years ago. It was after I finished my tenure at Tootene Networks. And to be honest, I was about thinking of retiring. And then I was called by Shai Kohn, or a CEO and he told me about this exciting new company that they just launched out of stealth mode and he told me about this story about in cheap monitoring and I immediately got hooked because it's something that as a chief designer for 30 years that was something that I was always being keen of having

and suddenly there is a company that I can get involved with actually introducing this novel technology to other semiconductor companies. Yeah I have to tell you we've been working together for quite a few years Proteintechs and I'm hooked as well so let's talk about pvt plus you tell us what that is yeah so people know what pvt is actually pvt sensor and was was there always okay for many years people use pvt p for process monitoring v for voltage monitoring and t for temperature monitoring that's something that was not you and again as i hinted before a proteomics is more than uh than pvt It's actually a comprehensive, complete hardware software software solution for in-chip monitoring. But PVT by itself was something that some customers needed,

and we believe that the current PVT offering were not sufficient, and we could definitely add value. So if you ask me what is the plus in the PVT, I think that's the interesting point. And I see the PVT plus with the plus represent, I would say, three major angles. The first one is quite trivial because I said PVT stands for process voltage and temperature. PVT plus offering adds two additional sensor components. One is a V-Drup sensor, and the second one is a duty cycle dystopin sensor. So adding more functionality than traditional PVTs.

The second angle is that unlike traditional PVTs that are delivered as a single, IP and IP only solution, we provide a complete solution from hardware and software, which means that it is containing control logic, it containing interface logic, it contains firmware, and it contains processing hardware, and all that with a software, a test and verification solution, all very comprehensive to make it easier for integration. by the castor. And that I would say is the second plus. And the third plus is that in every of those components, we actually got some novelty. And maybe I will explain a bit about the voltage and temperature sensor that we developed by using core level transistors

and not using traditional BJT devices or thermal diodes, because we wanted this to be a future ready for a two nanometer gate all around, backside power, and the most, again, the most advanced process features that do not rely any more of the oxide, analog gyo, or ABJT devices. So this represents why we call it the PVT plus. Okay. And what makes this system different than other PBT offerings? So other than what I already said, I think, first of all, it's because it is a complete

solution. It's not a standalone IP. It contains everything. So again, hardware and software, as I already mentioned, you don't need to build your own logic. You don't need to connect it to your own firmware. You don't need to connect it to develop a state machine to operate it. It all comes out of the box. It is also part of our what we call hardware monitoring IP subsystem. So it can connect and work with our entire outdoor monitoring IP subsystem, meaning that if you have a monitoring system integrated into a chip, it's trivial to integrate the PVT plus into that, but it can also be a working as standalone. Again, software and hardware is already included. I think it's also a different because, again, it is future ready. It's the only

solution that was developed with the most advanced nodes in mind. We are partnering with the most advanced companies in the world in the most advanced processes to verify it so we get something that it's already already deployed in the field by big customers and it is silicon proven and I think we have also proven that it's very scalable and configurable with very wide-scale deployment of big customers. And what are the specific PVT plus portfolio sensors and monitors? Yeah, so let me now break it down a little bit further.

So again, the piece is sense for the process monitoring. Our process monitoring is very unique. It's actually based on our portfolio or what we call our core agent portfolio. Our process monitors are unique that they are not just giving you one number that maps you the process, but they actually break down the process classification into the different behaviors of different standard cells and the different transistor types. So we can separate, for example, PNN, we can separate, for example, buffers from nose and nance. So the process classification is very granular and can get you to the root cause of what process element is the one that is different or change. between areas in your chip or between chips in your population.

So that's about the uniqueness of the P. Our voltage and temperature sensor are unique and very wide offering. So we have the LVTS. As I mentioned before, it's our local voltage and temperature sensor. It is the industry first to be based only on digital VD and digital transistors. So no dependency of any analog IOs, no dependency on any thick oxide devices or special devices. It has additional benefits that it can sense

temperature locally as well as remote. It can connect to remote sensors in dense areas where you cannot place the main hub. It can also sense a voltage locally, meaning the supply voltage at the placement area. But also it can sense arbitrary voltage signals, DC signals that can actually be any reference signal. That doesn't need to be a voltage

supply down to zero volts so we can sense arbitrary every voltage domain. It has additional benefits that you can set thresholds and get immediate alerts if that threshold is crossed so you don't need to continuously probe the value. Once a value is crossed, you'll get an immediate interrupt from the sensor. In addition, it can be operated as a monitor. The difference being that the monitor continuously work in the background. It has its own memory so it can store a number of last readings. So you can then go a little bit like back in time and to understand how voltage and how temperature was varying over the last couple of measurements, or actually more than a couple of measurements.

And it has a, again, full subsystem either by hardware or by femur to translate the raw reading into a, absolute values in meaning in millivolt and in 80s centigrade. So that's about the LVTS. Our VDS is a VDrup sensor. We actually have two variants of the VDrup sensor. One is a more digital one that is intended primarily for analytics, meaning when you are characterizing your design, you can use the VDS along with our software analytics

to characterize the VDrup behavior of your chip. of your chip. It is very small. It is good for a widespread location across the die. But on the contrary, we have the analog VDS or as a complementary, sorry. The analog VDS, the main advantage is that it is very, very fast. It has very low latency. And therefore, it is more geared into the real time emission mode usage. So to give you immediate alert, if you have a crossing of a critical threshold of the V-Drupp in order to get immediate response and then take a corrective remedy, for example, immediately do a clock circling or cycle skipping or whatever to avoid a cheap failure.

So these are two V-group sensors. And then the last one is the duty cycle sensor. It senses the duty cycle distortion, so meaning the time difference between the high and low of the voltage that is primarily used for where both edges of the clock are being used in high-speed interfaces, or actually when customers are using very high-speed clocks in multi-g gigahertz, they can use it to see if actually one of the edges is actually being skipped completely just because of some glitches. So the duty cycle distortion sensor is again unique and special for very high-speed designs. So these are the monitors and sensors included in our PVT Plus.

Oh, interesting. So what's the benefit for customers with the PBT plus system? Can you give us some examples? Yeah. So PVT is a standalone. It can serve different purposes. For example, process grading allow customers to tune into a dynamic voltage and frequency scaling. They can use it to a tune analog circuit.

behavior to trim a high-speed digital circuits and there are multiple customers using that voltage and temperature obviously are needed for a power and performance management is a thermal runaway protection again voltage input for a DVFS tables these are the typical usages of the V&T a VDrup again as I'm mentioned before, to avoid failures due to excess voltage supply and usages of the duty cycle is sorting again for managing high-speed clocks and their integrity. But here I want to say something very important. I already said our PVT plus offering is complementary to our in-chip monitoring system. It is very important for me to emphasize that the fundamental

of VIN chip monitoring is by looking into the cheap behavior from the, mostly from the timing standpoint, in very high coverage. So our philosophy, first of all, is to say, your chip is going to fail first of all because there is a timing violation. And then you are trying to understand why the timing violation happens. And those things that I mentioned could explain a timing violation or timing change between, again, between different units, between different areas in the same unit, or in the same area in the same unit

across time due to changes of workload or due to changes, again, of temperature, voltage, and noise. So the PVT is very important for customers, and it is especially and more useful when it is used with our hardware monitoring system, our health and performance monitoring system completed, because it gives them the full picture,

for real-time usage as well as for offline analytics. And why is this important now? Why it is important now, again, PVT has been around, it has always been important. Obviously, as semiconductor process scales and the advanced nodes, the way we implement it is already geared into the most advanced nodes. Again, gate all around, already is not enabling or not implementing thick oxide devices that were used for the traditional thermal diode.

So it is more important to look into advanced ways of implementing those sensors, especially even Dexite power has even more challenges of implementation, but it is also more important, not just because of the advanced nodes, but because the cheap scales are becoming so big. And one thing that does not scale is the power consumption. In fact, power consumption continues to increase and therefore thermal density is one of the major problems in the advanced semiconductor, especially AI, HPC. They are suffering a lot and they are dealing a lot with failure prevention due to access thermal and noise variation.

So PVT is becoming an invaluable tool to prevent. cheap failures to prevent accelerated aging, accelerated stress. And if we are even looking into the multi-dye projects, advanced packaging, die-to-di again, 3D integration, 2.5D integration creates even more severe hot spots for both voltage and thermal. And you just need those sensors to help you need. mitigate or prevent failures due to excessive, again, excessive temperature and excessive voltage noise. And are there specific markets that these are applicable to?

I would say no. I think that every chip nowadays need to take into consideration those phenomena. And we see customers all the way from the biggest cost, biggest chips customers, HPCs, AI and all the way down to consumer to mobile. Everybody needs those PVT sensors along with our complete monitoring solution. They work together and they best work together. But all our customers definitely need those capabilities. And again, I cannot see even one market segment that will not require today or at some point in the near future. these capabilities.

And final question, near, how do customers normally engage with your company? Okay, so again, we have already established ourselves as a leader in the cheap health and performance monitoring so they know to reach us. That's kind of the obvious. We are very active in conferences. We issue papers. We present at seminars. We do webinars, we do podcasts such as this one. We have a very engaged website. Actually, we are launching a new website quite shortly, and it has great content that customers can learn about what we do.

We are also active in some standardization committees, so companies learn about us and what we do by seeing our discussions and our offering those standardization committees. So I would say combined, we already have a very nice reputation for being the one-stop shot for anything to do with cheap health and performance monitoring solution, again, including PVT Plus. Yeah, I agree with that completely. I see you at conferences all the time. In fact, we have a couple conferences coming up.

I'll see you there. And thank again for your time, NIR, and hopefully we can have you back soon. Okay, thank you very much. You're welcome. That concludes our podcast. Thank you all for listening and have a great day.