Sensing Solutions In Factory Automation

TOM FLOYD: Hello everyone, and welcome to today's webinar. I'm Tom Floyd, I'm the moderator for today's session. And with me is Harold Joseph, Senior Marketing Manager. Welcome, Harold.

HAROLD JOSEPH: Hi, Tom, how are you?

TOM FLOYD: And Jason Seitz, Staff Application Engineer. Welcome.

JASON SEITZ: Hi, nice to be here.

TOM FLOYD: Good to have both of you was well. As usual, one quick housekeeping tip to keep in mind before we jump into our conversation today. We really want our webinar today to be interactive. So to make sure, for those of you who are joining us live via Go To Meeting, make sure you type your questions to me directly and I'll make sure to ask those on your behalf. And with that said, I think we can get started.

HAROLD JOSEPH: Okay, thanks, Tom. So thanks for joining us today. We're going to be talking about Sensing Solutions in Factory Automation.

Panelists

So online here, the panel, are myself and Jason Seitz here to answer your questions.

Objectives

Today we're going to start off with a few minutes talking about National's Factory Automation Campaign. This is a campaign we're going to be running through the summer. We're going to have several different webinars dedicated to this campaign. The goal of this particular campaign is to get a better understanding of what sensors are used in Factory Automation. And we'd like to look at our solutions in those areas and talk about the products that fit those solutions.

FA System Overview

So Factory Automation covers a very wide range. This includes refining in chemical plants, pulp and paper, food and bottling, semiconductor processing, automotive and robotics, a very wide range of systems. And there's a lot of analog products that address these applications.

Overview

In fact, there's over a billion dollars worth of analog content in Factory Automation. The 16 largest suppliers address less than 50% of the segment sales. So things are really spread out. There's a lot of people in this area addressing a lot of different kinds of applications. The current products we have in development, and the existing products today, will address over $750 million of sale available market. So in a couple years we're talking about a major portion of this market. We're seeing a number of key trends in this market as well. A drive to more energy-efficient systems. We're also a seeing a real focus on getting better and finer control so that plans can stay more competitive worldwide. We're also seeing an increase in diagnostics to reduce maintenance and downtime. And also an increase in the adaption of wireless. And this is primarily occurring in the non-critical applications. There's some move to look at more of the critical loops, but this is still slow in Factory Automation.

TOM FLOYD: So Harold, which of these trends is most relevant to the applications that we'll be talking about in the webinar today?

HAROLD JOSEPH: Okay. Well, today we're really talking about sensing. And we're going to be focusing on getting the right performance for the right application. So looking at the better finer control and the focusing on performance is the key one. However, a lot of these products are also very power efficient and that's an issue as well. So it's really those two areas that we're going to be addressing today.

TOM FLOYD: Okay, great.

Industrial Factory Automation Overview

HAROLD JOSEPH: So over the next several months we're going to be addressing a number of different webinars. The first one circled in red is the one we're doing today in Sensing. Then on 7/9 we're going to be doing a Control Networks that tie in the control system, the I/O modules, back to the main control room. Then we'll be moving on 7/23 to a Motor Control webinar. And on 8/20 to an I/O Module webinar. And finally a Machine Vision webinar that'll probably be in early September, although we haven't scheduled that date yet.

Common Sensor Applications

Okay, the sensor applications address -- are found in a lot of areas in Factory Automation. You could have a pressure element, for example, where the same element might be used by one customer in a handheld calibrator where power consumption is very important. Then you could find that same element used in a pressure transmitter. So a transmitter in Factory Automation refers to a 4-20 milliamp output device, so this would be a direct analog output. So we see a lot of requirements for pressure and temperature, optical applications. We see control systems for HVAC and monitoring. Safety systems, weighing systems. Systems where somebody's sending a signal over a distance to a central control room. Or an application where everything's self-contained, like a weighing system is a good example of that, where the microprocessor is very close to the sensing element. So as I said, they're used in a wide range of areas within the plant.

TOM FLOYD: So the main point you're really trying to drive home here is that sensors are everywhere and they're in a variety of applications?

HAROLD JOSEPH: Yes, sensors are everywhere.

TOM FLOYD: Okay, gotcha.

Focus Areas

HAROLD JOSEPH: So we've actually divided the presentation into two areas. One is the Sensor Interface Solutions where we'll be covering a separate analog output, and we'll do that briefly. Then we'll get into some of the key sensing areas. We're focusing on pressure, temperature, optical, chemical and pH. And in those areas we're going to be talking about a solution that includes both the amplifier and the A-D. One of the big questions in this area when you're looking at a signal path is how do you determine signal path performance. Recently we released a Web 2.0 called the Webench Sensor Designer that actually calculates the performance for you. So in those areas we'll be focusing on a complete solution of an amplifier in an A-D.

Sensors Interface Solutions

So then on to the Sensing Interface Solutions.

Analog Output

Analog outputs fall into two basic classes. One of a 4-20mA. These are primarily found in a section called Process Control Industries within Factory Automation. So this is refining pulp and paper, steel, chemical and gas transmission. These are areas where you're sending a signal over a long distance. Voltage outputs can vary from a 5V rail, a 1-5V, 0-5V, up to plus or minus 15V. And these would typically be found on a factory floor in applications like food processing, automotive robotics, consumables and semiconductor. General trends we're seeing is a move to a lower voltage 5V applications. A couple of drivers for this are the fact that they're less costly than the higher voltage applications. And we're also seeing move to a distributed control. So as you're getting the processor close to the analog sensing element, there's less concern about noise in the environment, so a lower voltage sensor is fine for those applications. We are seeing a replacement of older high voltage applications with newer devices based on newer processes. Still more expensive than 5V products, but there is a turnover in the higher voltage area as well. We're also seeing a slow move towards digital on the plant floor. This wouldn't be like an SPI or I2C. It's more like the PROFINET or PROFIBUS field bus applications where these are bus protocols or digital protocols from a digital sensing element, digital output device to a digital control room. But that's happening very slowly. These digital interfaces provide easier calibration and reduce maintenance costs, but the real benefit is where you can change over a whole area at one point. So you only really see them done when you have major plant expansions or upgrades, and so that's a fairly expensive undertaking.

4-20mA Configuration

So going into the 4-20mA areas, we mentioned that they're using the process control areas. The major reason for that is the signals can be sent long distances, and they have a two-wire approach. So if you compare this to a voltage device where you save at least one wire, that's a major cost savings in a refinery or chemical plant. It costs about a $100 a linear foot to run wire, so that's a major savings for them. So we're going to be focusing on the left side, the sensor/transmitter side.

4-20mA Configuration

So these are used with all types of devices, pressure, temperature and optical. Typical performance for these applications are 0.25% to 1.0%. Right now we're seeing the analog signal path needs to be typically under 800 micro Amps. The sensor might consume about 1.5mA, maybe up to 2. You have 4mA to work with. You need a little bit of budget. So in general, anything under 1.0mA is fine, but the market's really right now about 800 micro Amps, with some even moving to 500 micro Amps.

TOM FLOYD: Now, why is less than 800 micro Amps important?

HAROLD JOSEPH: Well, you've got 4mA to work with. So that 4-20mA is actually the signal, but it also provides the power to run the device. And that is to include the sensor, the electronics and any overhead you might have. So that's a key factor in those designs.

TOM FLOYD: Okay.

HAROLD JOSEPH: So, Jason, do you have anything to add to this?

JASON SEITZ: Yes, so what's important with this application is typically the designer is going to want to maximize the loop length, and they can do that in a few ways. One is to increase the loop voltage supply, but often that's a constraint that's not easily flexible. Another way is to decrease the ohmic drop or the voltage drop across the loop by increasing the diameter of the actual wires. However, that's going to add cost. One of the easiest ways is to actually minimize the voltage necessary on the transmitter side. So you're going to want to do that with low power components, such as the amplifier here, which is the LMV951, can operate down to 0.9V.

Products

So here's some other low power parts in our portfolio. Again, starting off with the LMV951. You'll see most of these parts could go down to 0.9V, again, helping the designer minimize the operating voltage of the transmitter loop on the 4-20mA configuration.

TOM FLOYD: So these would be the parts you selected first as saying this would be a good starting point if you were doing this kind of application then, right?

JASON SEITZ: Correct, correct. Also here is the LP2951. It's a low drop-out regulator, again, with that thrust to minimize the voltage.

TOM FLOYD: Now, are all the products listed here available now?

JOSEPH SEITZ: They are.

TOM FLOYD: Okay.

Voltage Output Configuration

HAROLD JOSEPH: Okay, in terms of the voltage output configuration, as I mentioned before, these are used on the factory floor, food processing, automotive, consumables and semiconductor. The high voltage outputs, typically over 5V, provide better noise immunity than the lower voltage devices. The 1-5V, 0-5V, are the largest part of the market. That's an area we're focused on. As I mentioned before, we're seeing a growth in distributed or localized control driving a need for more 5V output applications.

Recommended Products

So we've got a list of recommended parts here for this, and these are 1-2 selected as well for this application.

JASON SEITZ: Yes, so you'll see up at the top is our higher voltage portfolio from the VIP3 process. It allows you to get up to 32V. However, if you want to address some of the 0-5V, 1V-5V that Harold was talking about, you could go to more precision such as our LMP2015. See the TcVos, which is really critical because this is an item that's hard for the designers to calibrate over temperature. Also, if you go to maybe the LMP7715, that'll reduce your voltage noise.

Key Sensing Areas

HAROLD JOSEPH: So now looking at some key sensing areas, and this is an area where we're going to look at specific types of sensors.

Sensor Overview - Signal Path Considerations

First thing we do is, and when we look at this, is we want to provide a suitable interface to the sensor. That means we need to look at scaling the sensor's output signal, source impedance. Need to convert the sensor output to voltage if necessary. Sensitivity, dynamic range are important. Also, the excitation source for the sensor. It might be voltage or it might be even a constant current supply. Then on the ADC side, we want to scale that output to match the ADC's input voltage. We want to look at the sample rate required for the application. And we also want to look at isolation, whether we would need to isolate the amplifier output and the ADC input.

TOM FLOYD: Okay, and tell us a little bit more about why sample rate is an important consideration.

HAROLD JOSEPH: Okay. So in terms of the application, there are really two issues going on here with this. One is we're seeing a need to measure a process, so you might have a pressure device, for example. And maybe the process requirement could be as low as a 100 samples a second or even a 150 samples a second. However, some of the data acquisition systems now want to look at signals more frequently. So they may put a requirement to say we would like to have a device sampled a thousand times a second or even 4,000 or 10,000 times a second, so that we can see if there's a change, and that will give us an indication if there's some kind of error situation occurring. So we're actually seeing a need to look at a higher sample rate for a device, even beyond what maybe the process might require. So it's important to understand that not only from the application standpoint, but from what drives that, and then that helps us pick the right products.

TOM FLOYD: Okay. We had a question come in and it sounds like, from what I'm reading, there's still some confusion about the offerings that were talked about in plus or minus 10V and plus or minus 15V range. Can you talk a little bit more about that?

Recommended Products

HAROLD JOSEPH: Okay, let's go back. Did it say what the particular question regarded? Was it the output voltage or...

TOM FLOYD: That's all I have.

HAROLD JOSEPH: Okay. So, Jason, I guess we really talked about two different devices here. One is, our group handles a 32V supply. And we were talking before the show that common voltages on the plant floor might be 12 or 24V, even 36V at some times. And then the outputs can be up to plus or minus 10 or plus or minus 15V. So I guess that's one difference. The other devices are more precise. But is there anything in particular you look at when you focus on these particular devices?

JASON SEITZ: Sure. So we're trying to really address several different applications with our portfolio here, right? So we have the plus or minus 10V, plus or minus 15V systems. If you look at the top half of this page, these are the amplifiers in our portfolio that are going to address that. They're rail-to-rail input and output amplifiers. And as you can see, they go up to 32V, so the plus or minus 15V will not be an issue. If you're operating instead on the 0-5V or 1-5V platforms, we have more parts in that portfolio, which will allow you to get even more precise. And that's what the bottom half of this picture is explaining.

TOM FLOYD: Okay.

HAROLD JOSEPH: Also we have our contact information at the back of the presentation. So you've got Jason's phone number and email address as well as mine if you have questions or if we haven't answered that particular question sufficiently.

TOM FLOYD: Okay, great.

Pressure

HAROLD JOSEPH: So let's see, talking about pressure sensors then, this is the first area we're going to be looking at. Typical output sensitivities from a sensing element for a pressure device are from about 5 mV/V to 15mV/V. 10mV/V is very common. Bridge impedance can run 3 to 10K. 4 to 6K is what we commonly see. The sensor drive can be either constant current or constant voltage, with performance ranging from a 0.1% to 1.0%. And actually the sensor drive, that is specified by the sensor supplier, and it's really tied to how they process their chip to get the kind of nominal output they want. So they will either decide to drive it as a constant voltage or constant current. And so we just need to know that when we look at the analog solution.

MEMs Pressure Sensors Most Common Type of Pressure Sensor

So we put in a slide here to talk about one of the most common types of pressure sensor, it's a MEMs device. I just wanted to illustrate how one particular sensor can go into a whole variety of different applications. So we divided this into three categories. The top one, a typical sensor cost is low. And this could be an integrated part or a very low-performance device. So we typically would not address those applications, but the other two applications we do, or areas we do. The next group has PC mountable products that could start at about $5. The center one says $10 to $12, but they're as high as $18 or $20. The one down there in the bottom of the middle bar is a stainless steel device, that says $25 is a typical cost, but it could be as high as $35 or $40. And so this is an application where someone would need an analog solution to interface that sensor to some microprocessor. Our content in there, it usually ranges in the $4 to $6 area. This is where the Webench Sensor Designer was primarily focused, to provide information about solutions for these kind of applications. Now interestingly, that same pressure sensor would be used for transducer companies where the sensor company would provide that product directly to the sensor companies or manufacturers, who would provide a complete solution. And here we would be working directly with those transducer or pressure transmitter manufacturers. And those prices can range, for the end product, from $100 to $1,000. Most of that is really focused on some of the housing and requirements. The one on the very bottom is an explosion-proof housing used in refineries and chemical plants. So interestingly, the performance across this range of that pressure sensor is really the same. What drives the difference in cost a lot is the end application, the requirements in that end application and the packaging concerns.

TOM FLOYD: Okay.

Pressure (cont.)

HAROLD JOSEPH: So continuing on, looking at the pressure sensor, if we want to pick a good starting point for this, the LMP7715 and 7716, along with the ADC121SO21, is really a good place to start. If you have a constant current application, the LMP7702. The amplifiers for that constant current drive are also a good choice for that. When we look at these applications, we're looking at the ADC in terms of performance. So we have an 8, 10 or 12-bit. We've outlined what we call the signal path performance for those devices to give you a good, rough starting point. On the opamp side, it really depends upon where that end application is used. So we might find an application in a handheld application where low power is very concerned. Or where maybe there's a noise concern in a high-end device. Or maybe something for one reason where they want to look at a higher bandwidth.

TOM FLOYD: So on this particular slide, are you saying that the majority, let's say 75% of applications, can be addressed with the 7715 or 7716?

HAROLD JOSEPH: Yes, I think so. Seventy-five percent is really a pretty good number. I think that's mostly where we'll climb for that.

TOM FLOYD: Okay.

JASON SEITZ: Harold, I think another nice thing to point out on this slide would be, you see the schematic, the voltage drive and the current drive, both of these configurations are handled by our Webench Sensor Designer. So it allows the flexibility for our designers to design in either format.

HAROLD JOSEPH: Okay.

Recommended Products

So in terms of the recommended products, this is something we will put with each of the sections we're going to be talking about. And in the ADC section we've added a column called signal accuracy. So here, typical ADCs are spec'ed in terms of INL, which is the worst case non-linearity from a transfer curve, so that's close to accuracy. Or ENOB, which is a dynamic performance called effective number of bits, that's really more on when you have like an AC signal. So what we've tried to do is help out a little bit here by looking at signal accuracy as a starting point. Typically this is focused around INL, but with a low-resolution product like an 8-bit device, you might have an INL that's actually specified much better than the actual LSB, the least significant bit value. And in this case the best thing to do, in terms of signal accuracy, is look at that least significant bit. So the signal accuracy is not something you typically find in our spec sheet. It is in the Webench Sensor Designer. But it is pretty much close to INL. It's a great place to look at that ADC in terms of performance. The other thing we've done here is typically we've looked at single channel parts. Two, four and eight channel parts have the same basic specifications. We've got both SPI and I2C devices here in 8, 10 and 12-bit configurations. For the opamps we've given a number of different parts. We mentioned that the 7715 and 16 are a very good starting point. We've given a couple of alternatives. If you look at 2011 for example, this has a much better TCVos. So if you have a wide temperature range, this would be a good alternative for something like that. The LMP7701 is another sort of mid-range device. So it gives you a couple of parts to choose from. The LMP2231 is a very low power device. So for some of the handheld applications you might want to look at that part.

Load / Force

So on to load and load sales and force sensors. So these are very similar to the pressure sensors, also a bridge configuration. Differences here, though, is typically the output of the sensors are lower. 1.5mV/V to 3mV/V is typical. Sensor drive's usually voltage 5V to 15V. And bridge impedance around 350-ohms. So what this means is you have a sensing element that requires a much higher gain. Also has very good performance at 25C. So you can see 0.02% to 1.0%, so a different focus on the pressure sensors in terms of the gain and performance. However, I will say that with a lot of load cells and force sensors, the performance over temperature is not that good. So it's really critical for us to understand what the overall temperature performance requirement of the application is. There are many combinations of mechanical structures that are used in load cells.

Load Cell Examples

We've given a few here as examples. And I wanted to point out that the structure doesn't always relate to performance of the load cell. What the structure's really designed to do is address the particular application. So the top sensor here is a tension or compression load cell. The middle one is used for compression. The bottom one, an S beam load cell, has a high output, 3mV/V. So you'll see these different kind of structures in load cells.

Load / Force (cont.)

In terms of product selection, we go typically with the 2021/22 for this particular device. And a 12-bit or 16-bit ADC. Now, we've shown a diagram here that has an instrumentation amp. The 16-bit product is a differential input. So the structure changes or the diagram changes a little bit. Jason, you mentioned before about the Webench Sensor Designer and how it handles both constant voltage / constant current. It also actually handles both of this situation as well. So if you have a 16-bit versus a 12-bit, you will get an output that is built around a differential ADC.

JASON SEITZ: That's correct.

HAROLD JOSEPH: Well, thank you.

JASON SEITZ: You're welcome.

HAROLD JOSEPH: Okay, so the ADC choices range from 8-bit to about 16-bit. We've listed the performance there. And the opamps, as with the bridge sensors, we're looking at things like bandwidth, noise and power.

TOM FLOYD: Now, how do you pick the signal path performance for a given sensor?

HAROLD JOSEPH: Okay, good question. One of the areas that -- the easiest way in this case is to just ask the customer. We were talking about the Webench Sensor Designer, and one of the difficulties we had with that is that we can't always ask the customer what overall signal path performance is needed? So in that case what we did was set sort of a guideline where we figured whatever the performance of that sensor is, you want the signal path not to contribute too much to that. So if it's a given value, maybe your signal path should be a quarter of that total value, and separate that between the amplifier and the ADC. So I think that's probably a good rule of thumb as a starting point. If you can ask the customer, though, that's the best thing to do.

TOM FLOYD: Okay, gotcha.

Recommended Products

HAROLD JOSEPH: So as far as the recommended products, we've listed the 8, 10 and 12-bit A-Ds again, as well as the 14 and 16-bit, showing the signal accuracy, all run in this range up to 200 Ksps. Load cells tend to be very low sample rate. So you might have, in systems where you need to run faster, up to a few thousand or even 10,000. But for the most part, a lot of these applications are near DC, so we end up clocking the ADC, but these should handle most of the ranges that you need. In terms of the amplifiers, again we're looking at a question of tradeoff in performance over the kind of bandwidth you need for the product and, say, performance over temperature. The 2021 has a very good performance over temperature. If you didn't need quite that performance, you could go with the 7715 or 16. The LMP2231 again is very good for low power applications.

Thermocouples

Okay, and Jason, you wanted to talk to us about thermocouples.

JASON SEITZ: I'd love to. Thermocouples. Thermocouples are great compared to other temperature types of sensors. The reason why is, they're extremely rugged, they're tough. They are bi-polar devices, which could go both negative and positive voltage, depending on what temperature you're measuring. They're also passive devices, meaning you don't have to have the excitation source as the pressure sensor for example. So in this signal path you'll see that there's an amplifier, an ADC, as well as a temperature sensor. And the reason for -- you might ask why is there a temperature sensor if you're also using a thermocouple, which is a temperature sensor itself? So the thermocouple has the advantages of operating over a very extreme temperature range. However, you need a silicon temperature sensor onboard near your actual measurement in order to do cold junction compensation. Basically the thermocouple is a relative temperature sensor. In order to get an absolute temperature, you're going to have some type of reference temp sensor. And this here is handled by the LM904022.

Thermocouple Types and Accuracy

Here we have a table of the different calibration types of thermocouples. Basically what we're trying to display here, what we're trying to communicate is, there are a couple standards. There's the standard accuracy. And then there's the actual special accuracy. Special accuracy becomes more precise. And you'll see that, depending on the different calibration types, you'll have different operating temperature ranges, and you'll also have different accuracies associated with them.

TOM FLOYD: Now what are some of the most common types of thermocouples that people will run into?

JASON SEITZ: The most common types are actually the top two, J and K. The reason for that is, they are generally linear. Of course thermocouples are a non-linear device over temperature. You're not going to have a straight temperature to voltage dependence. But J and K are relatively linear and they also operate over temperature ranges, which are applicable for many applications.

TOM FLOYD: Okay, great.

HAROLD JOSEPH: So when you were working on the Webench Sensor Designer for this, I noticed that for a lot of the vendors, everybody has to follow the same kind of type. So it's always the same type from different vendors. But a big issue I think you guys were dealing with was the different backend connector. So you have a whole variety of different connections. Is that really the difference between a lot of the different vendors?

JASON SEITZ: Yes, so it's kind of similar to the pressure sensor case, right? Really, all of the pressure sensors or all of the thermocouples kind of have the same behavior. But what the challenge is, is the connection to the actual circuit. So the Webench Sensor Designer actually has evaluation boards, which handle, for thermocouples at least, two very popular connections. It's the bare wire connection and also the plug-in connection.

TOM FLOYD: And a question that we had that just came in is, do thermocouple companies also make silicon temp sensors?

HAROLD JOSEPH: Mind if I take that? What you'll find within a family of sensor manufacturers is that you will find those companies providing the same product. You wouldn't typically make a silicon temp sensor or a pressure sensor in the same way that you'd make the process or a thermal couple. But a lot of the sensor companies tend to gravitate toward the common kinds of sensing applications. So you'll see a company representing five or six different kinds of technologies. Maybe these are companies that they've acquired over time, where those products are made in a variety of different plants. So it is common that you would see that kind of, you know, a variety of different products from sensor companies.

TOM FLOYD: Okay, great, thanks.

Thermocouple (cont.)

JASON SEITZ: So to our product selection page, some of the criteria for these parts is the thermocouple itself produces a very low voltage. For example the type K, the sensitivity is about 40mV per degree C. So you're going to want a relatively precise signal path. Starting off with the LMP7715, low noise, low VOS. Like I mentioned before, the LM94022 is used as the cold junction reference. And a 12-bit ADC will be more than precise enough for your signal path needs. But depending on what your actual precision is, you could go for 8, 10 or 12-bit ADCs. And some other important opamp criteria would be the common mode rejection ratio. For example, depending on what configuration you have, you want to minimize the common mode pickup, which could happen in a noisy environment over long thermocouple lines.

Sensor Overview - CMRR vs. Component Count

So coming back to the common mode rejection ratio, here we have three configurations. From left to right we're going to non-inverting difference to the instrumentation amplifier. So the benefits that going from left to right offer is an increase in common mode rejection ratio. Again Factory Automation, industrial applications, there could be a lot of noise in the environment. Thermocouples, very long lines that could add error to your system. So one way to minimize that error is by an instrumentation amplifier, which has an inherent very good CMRR. Of course, you're going to add components and you're going to add some complexity.

TOM FLOYD: So you've got three types of circuits here. How many types could you potentially have for a thermocouple?

JASON SEITZ: Yes, it could get up to 12 actually. And that depends whether or not what temperature you're actually sensing and how that relates to the actual system measurement temperature. So for example if you are measuring very close to your system temperature, you want to make sure the amplifier that you're measuring the temperature with doesn't go into saturation, because that's going to be a very low voltage. So you might want to include a device such as our LM7705, that's a negative bias generator, so to make sure you get to that bottom rail of the amplifier. Also, if you're measuring temperatures at both below and above the actual system temperature, if you recall, thermocouples are bi-polar devices, have the ability to produce both a positive and negative voltage, you're going to want to maybe add some level shifting. So in the end, it could be up to 12 different configurations. Luckily we took this into account when were designing the Webench Sensor Application Tool. And we've compensated for all of those different configurations.

HAROLD JOSEPH: So when you said system temperature, you meant that's what the temperature that the cold junction would measure. So it's really in reference to what that temperature is at the cold junction?

JASON SEITZ: Exactly.

HAROLD JOSEPH: Okay.

TOM FLOYD: A question that came in too was, would it make sense for Webench to support (inaudible) designs?

HAROLD JOSEPH: That's actually something we're looking at right now. So we do have a couple of different areas we'd like to expand. When we expand on the Webench Sensor Designer, what we actually do is go in, we look at where there's a bunch of common kinds of circuits that we can look at for a family of different products. Because what we actually want to do is be able to build and test and be able to ultimately supply a board to evaluate that sensor. So that takes a lot of work. It's not just a design on paper, it's an actual tested board that works that we can support. So we're looking at that. We're looking at some RTDs as well. But we don't really -- the team doesn't really have a date as to when those would be rolled out. If somebody has a particular concern, please send both of us an email about it. We're both on the Webench Sensor Designer team. So we are looking at those all the time, but unfortunately we don't have a release date for you.

TOM FLOYD: Okay, understood, thanks.

Recommended Products

JASON SEITZ: Moving on to product selection. Again, this follows the typical format on the top. We have the ADCs, whether or not you want 8, 10, 12-bit, depending on what your precision needs are. The bottom half are precision parts. Again, we're measuring very small voltages here. So anything from our LMP portfolio should suffice. For example the LMP7715 here has a very low TcVos for 1mV per degree C. You also want to keep an account of what your actual temperature range is, and pick a TcVos part accordingly.

RTDs

Moving on to another temperature sensor, the RTD, Resistance Temperature Device. This actually uses the ability of the characteristic of metal. As you change the temperature, it'll actually change resistance. So a common standard here is using platinum as the wire and putting it over glass. And you want to have a consistent or a known resistance value. Typically they come in either 100 or 1,000, or 1K, ohms. And this is the resistance that that RTD will provide at 0 degrees C. Or as you see here, in the range of negative 200 to 0 degrees C. So RTD, some of the benefits of the RTD versus the thermocouple is they're actually more precise at that 0 degree C range. If you recall back to the table that we had of the different calibration types of the thermocouple, you would see anywhere from 1 degree to 2 degree accuracy. Where here we get well below 1 degree accuracy. Again, you want to keep in mind what your temperature range is. Thermocouples keep that consistency over the complete temperature range, but the RTD will actually vary over that temperature and you're going to want to compensate accordingly. So there's different configurations. One of the most popular is the 4-wire configuration and that's what you see in the schematic here. That allows for force and sensing Kelvin type measurements.

RTDs (cont.)

Product portfolio. Because you're getting into a more precise type of measurements with the RTD, you're going to want to go to our LMP2011/2012. And you can start at ADC121S021, 12-bit ADC will be sufficient for this type of application. For the current drive, you're going to want to start with the LM4140 regulator, and that'll provide an accurate stable current to drive the RTD.

TOM FLOYD: So in general, is it fair to say that the RTDs are generally more precise than thermocouples?

JASON SEITZ: Given a certain temperature range, yes.

TOM FLOYD: Okay.

JASON SEITZ: Again, you're going to want to look -- each of the RTDs and the thermocouples have specifications. They have standards which are commonly available, and manufacturers design to those standards. So you're going to want to make sure what the precision is based on that temperature range. But at the 0 degree Celsius level, yes, RTDs are generally more precise.

TOM FLOYD: Okay.

HAROLD JOSEPH: I guess just to add to that, so thermocouples cover such a wide range of temperatures. But if you really look at -- when we look at the end applications, if you're looking at sort of an environment where you're seeing a reasonable kind of temperature range, like we live in an environment where a lot of our instruments are spec minus 40-125C. So if you're talking about any kind of process within some kind of normal range temperature we're used to or those kind of things, then the RTD really has a performance advantage.

TOM FLOYD: Okay, thanks.

Recommended Products

JASON SEITZ: ADCs can vary, 8, 10, 12-bit. Ten if you really want a more precise application, you're going to lean more towards the 12-bit side. And we have our more precise amplifiers down at the bottom here; the LMP2011, for example, will provide you .015mV per degree C TcVos.

Optical Sensing

HAROLD JOSEPH: So next area we're going to look at is optical sensing. And this is an area that's covered by the Webench Sensor Designer as well. So we really look at two basic configurations. And we look at photoconductive configurations, which is typically used for a faster response. And a photovoltaic configuration, more linear, no leakage current. So it's really two slightly different approaches, and we address both in the tool. So the applications for these configurations can vary a lot. You could use an optical sensor for a ranging application, you could use it for intensity, you could use it for position, you could use it for proximity. So when we look at these applications, it's very important for us to find out from the customer what aspect of the optical sensor he really cares about, and how is he using that sensor. Similar signal paths support all of these applications. So it's easy for us to come up with a generic application, we just have to pick the right products for it.

TOM FLOYD: Okay.

Optical Sensing (cont.)

HAROLD JOSEPH: So we've looked at a couple of different products, LMV793 or an LMP7721, and an ADC121S021 as a good starting point. Again with the ADC, though, you're looking at really a performance requirement, so it depends upon the end application. Then for the opamps, we look at some signal path issues. What bandwidth is required? What kind of bias current? What is the known capacity of input? One key issue is that the low input current amps are usually a good place to start since a lot of photodiodes have very low photo currents, so that's an important issue.

Recommended Products

In terms of recommended parts, we've covered the A-Ds before. They're very similar to what we've talked about with the other devices. There are some applications where you have a high-speed requirement. And in that case you might want to look at some of the faster ADCs. So they're all pin and function compatible, so they drop-in. So I've given that note here about the higher-speed products. They have same signal performance as the 8, 10 and 12-bits listed here. In terms of the recommended opamps, it gets back again to the application requirement. So the LMV793 is a part that's got a great gain bandwidth, so it can handle very fast signals. It's got a good voltage noise and a pretty decent input bias current. The LMP7717 is a product that's similar to that, but it's got better offset and drift characteristics. We didn't have enough room to put all of the specifications here, but that's why those specs are similar, but the change would be what kind of drift you wanted to tolerate. And then the other two products we looked at would be alternatives if you didn't need the gain bandwidth, but you were looking maybe for lower bias current.

TOM FLOYD: What's the most important thing for people to keep in mind about optical interface solutions?

HAROLD JOSEPH: Okay.

Optical Sensing (cont.)

Well, back to that issue before, as I mentioned, there are a lot of different alternatives and applications. But one of the first things we really look at is this low input current, because of the current of the photodiodes. So that would probably be the first thing that you would look at.

TOM FLOYD: Okay, gotcha.

pH Sensing

HAROLD JOSEPH: And pH sensing.

JASON SEITZ: Okay, moving on to pH sensing. Basically these pH electrodes measure hydrogen activity. The use the characteristic of when you have two different liquids with two different pHs and you place them in contact across a thin membrane, a thin glass membrane, an actual electrical potential or voltage will be induced. So some of the characteristics of these pH sensors are the fact that they are bi-polar. Depending on what pH you're measuring, it'll be either positive or negative. And also some typical applications that you'll see them in are water treatment, chemical processing, and environmental test systems. Another factor is, and you'll see in the schematic here, is we have a temp sensor placed close to the pH sensor. You want to actually place that in the liquid itself because pH sensing electrodes have the characteristic that they're dependent on temperature. The voltage span will actually increase as the temperature increases. So you're going to want to make sure you compensate for the temperature there.

TOM FLOYD: What are some of the characteristics of a pH sensor that folks should be aware of?

JASON SEITZ: Besides the ones I mentioned, an important factor is that it's of high source impedance, anywhere from the 10 mega ohms to a 1,000 mega ohms, actually. So that's critical and any current entering that source impedance, that very high source impedance, can potentially produce an error voltage. So that'll lead kind of to some of the amplifiers that we're going to look at.

TOM FLOYD: Okay.

pH Sensing (cont.)

JASON SEITZ: So the amplifier LMP7721 is kind of the focus amplifier here because it's our lowest input bias current part. The LM4140 is going to be the regulator that you're going to want to use to provide to the buffer, the level shifting buffer, to actually make this, instead of a bi-polar-type of application, more of a uni-polar application when working in a single supply environment, which is common in Factory Automation. And the LM35 temp sensor, it's a very accurate temp sensor. ADC122S021, 12-bit ADC. So some of the amplifier characteristics beside the low input bias current is, again, depending on the accuracy you need, you're going to want to have low voltage noise, low offset voltage drift. If you want to maintain linearity across the complete gain range or high gain and maintain that gain accuracy, you're going to want to have a high open loop gain on your amplifier as well.

Recommended Products

So, similar ADC parts, and I won't go into that too specifically, but I'd like to focus a little bit more on the LMP7721. You see a very, very low input bias current here compared to our other products in our portfolio. So that's really -- if you're concerned about the impedance of your pH sensor, if it's in the hundreds of mega ohm range and you want to keep that error voltage down, you're going to want to go for the 7721.

Chemical Sensing

Moving on to chemical sensing, so here we have electro-chemical sensors. They're good at measuring toxic gases. So environments or applications where that's necessary are in a lot of portable applications for fireman and mine workers. Before entering a certain environment, a closed environment especially, you're going to want to measure particularly maybe the C02 level and make sure it's safe to enter, so this is where these devices come in handy. You'll see on the right here is a potentiostatic circuit. The main function here is to keep the reference electrode and the working electrode of the actual electro-chemical sensor stable. And converting the very small or potentially very small working electrode current into a voltage, which then can be processed by an ADC.

TOM FLOYD: Now, when you're pairing an amplifier with a chemical sensor, which feature of an amplifier is important?

JASON SEITZ: So this is actually similar to -- when talking about the amplifier, it's actually similar to the photodiode or even the pH sensor case. And that's because what you're measuring here is a very low current, and you're going to want to minimize any error current, input current or bias current that the amplifier might interject into that signal. So you're going to see a very similar product suggestion for this particular application.

TOM FLOYD: Okay, understood.

Chemical Sensing (cont.)

JASON SEITZ: So again, LMP7721, it's our low bias amplifier and it's similar to the 121S021. So some of the same characteristics, low offset voltage, low offset drift, and really a focus on that minimum input bias current.

Recommended Products

ADCs here range from 8-12-bit, again, depending on what your accuracy needs are. Pay attention to the actual specifications of the electro-chemical sensor and the accuracy needs in order to determine what the appropriate ADC would be.

HAROLD JOSEPH: Okay, so in summary, weíve looked at applications where we focus on the most common high performance sensor interface applications in Factory Automation. Weíve selected applications where we can look at a finite number of analog signal path circuits that address a wide range of end applications. In terms of the sensor interface approaches that weíve looked at, weíve looked at applications where weíve looked at solutions that provide a sensor drive and amplification with just an amplifier circuit. Weíve also looked at applications where we provided a sensor drive amplification and an ADC to convert the signal to a digital output. And the Webench Sensor Designer is a good tool that you can use for taking a look at that signal path performance and it addresses most of the kinds of applications we talked about today. For additional information, weíve provided some links to the Webench Sensor Designer Tool. Also to application notes on the 4-20mA loop transmitter, as well as a variety of the different sensors that weíve looked at today. These are covered in the individual application notes. Weíve also got some individual application notes that address a little bit broader range, and look at machine monitoring, and maximizing signal path performance. As I mentioned at the beginning of this session, we have upcoming webinars scheduled, the first one for 7/9, that would be Controlled Networks. Then Motor Control Sensing for 7/23. An I/O Module one 8/20. And Machine Vision, that is going to be in September some time, but is currently TBD. So youíve got our contact information there in case you have questions or want to ask about a particular item. And I guess that would be about it. TOM FLOYD: Yes, sounds good. Well, thanks to both of you for joining us today. And thank you, all of you, for watching todayís webinar as well. Two things weíll ask you to do as next steps. One if you can take a few minutes to complete the short ten question assessment thatís associated with this webinar, weíd appreciate that. Just the way you test, what you learned in todayís session. And as always weíd like your feedback as well. There is an evaluation thatís associated with this webinar too. You can take a few minutes to complete that and give us your feedback. We look at all the feedback that you provide, itís especially helpful when planning topics for future webinars, so thank you in advance for that. Thanks again for joining us and enjoy the rest of your day.

From -	Name:	Email:
To -	Name:	Email:
Subject:	Sensing Solutions In Factory Automation