A&D webinar series: General purpose amplifiers

Just another WordPress site

A&D webinar series: General purpose amplifiers

Hi, my name is Sanjeev Manandhar, I am a systems engineer for General Purpose Amplifier product line I’ll go over the GPAMP roadmap for aerospace and defense My colleague, Pete Semig, who is the apps manager for GPAMPs, will go through the technical portion of this presentation So I’ll cover the roadmap And as I said, Pete Semig will cover tools, resources, and the technical portion of this presentation So, GPAMPs is a part of our Linear Amplifier BU with a portfolio of op amps that have gain bandwidth of less than 50 megahertz and offset in range of a millivolt or higher We have the industry’s largest cost-optimized, competitive, and easy-to-use amplifier portfolio We sell op amps in almost all sectors of market, including aerospace and defense The focus today will be on aerospace and defense– A&D– roadmap devices that we have released in the past and already in pipeline This slide shows GPAMPs investment area We have three categories of product– low-voltage, high-voltage, and standard amps So low-voltage op amps are op amps that have supplied voltage capability of 7 volts or lower These are high-performance, low-voltage op amps High-voltage op amps have supply capability of 7 volts and higher It can go up to 40 volts And these are high-performance, high-voltage products In the standard amps, we have commodity products, if you will These have cost-competitive op amps that goes in all applications and market sectors, and these are good performance op amps and sell in very large volume and are cost-optimized And if you look into this slide, we have chosen our low-voltage op amps– one of our low-voltage op amps, one of our high-voltage op amps, and two of our standard op amps that would fit in A&D space So, in low-voltage, we have TLV904x which is in quad-channel and dual-channel It’s 350 kilohertz, very low quiescent op amp, and that has very low offset We see this being useful in low-power applications that need decent bandwidth of about 350 kilohertz or so And the quiescent current runs lower than 10 micro amps or in single figures, which could be helpful for low-power power-sensitive applications Then, in the high-voltage area, we have taken OPAx991, which is in dual and quad versions, which is capable of 40V supply and is in 4.5 megahertz gain bandwidth range This is an op amp that has rail-to-rail input and output capability So if you go into 40-volt or our high voltage product portfolio, you run out of op amps that are capable of doing both rail-to-rail in and out So having this op amp with rail-to-rail input and output takes that constraint away and makes design simple And the input offset voltage of this op amp is about 1.5 microvolts, in typical In standard op amps, we have taken LM2904B and LM358B, which is the next-generation version of the products we have released in the market, which are [INAUDIBLE] LM358 and LM2904 These are dual-channel op amps These are industry-standard bipolar op amps and these are designed in new and more advanced

bipolar process than the older version And these op amps have found success in many market sectors, and I think these will be good for the A&D sector, as well So this slide shows what is already released from GPAMP in the market for both enhanced products and space products And these have been very successful in both market sectors If you see on the top, these are high-voltage op amps from TL074-EP and OPA2171-EP are 36-volt capable op amps The [? TLZ-2171-EP ?] is a 3-megahertz, CMOS-based, dual-channel op amp with 36-volt capability TL074-EP is a JFET-based topology with very low noise And if an application is sensitive to noise, this is the op amp that would be useful There are links to these products in this slide You can click on these links and get more detail about these products OPA2904-EP and LM2902-EP are dual- and quad-channel devices with similar features These are very high-gain single-supply op amps capable of gain bandwidth of about close to a megahertz And they are internally compensated to be stable at unity gain And in space, we have LM158QML-SP and LM124AQML-SP These are space-grade parts, GPAMPs They are rated for 100 kilorad total ionization dose Now, I’ll shift to that innovative technology that we have implemented in the op amps that I mentioned that’s coming in the pipeline, next-generation op amps What is EMIRR or EMI filter? So [? OPA991 ?] dual-channel and quad-channel has very good EMIRR– EMI rejection ratio, so what it is, is we are trying to provide a very compact solution We are putting components very close to each other, whether it be an analog, digital, our even switching circuit Maybe part of the power supply And sometimes these op amps are even part of the switching power supply EMI can have very serious effects, detrimental effects, if not taken care of properly So our next-generation op amps, like the one I mentioned OPA4991, -2991, the TLV90xx and LM358B and LM2904B– these have integrated EMI filter What it is a low-pass filter right at the input/output of the op amp with a cut-off frequency that’s outside the gain band of the op amp so that it doesn’t affect the performance of the op amp when it is outside its bandwidth So you can see the figures over here with the op amp that has the EMI filter and the one without EMI filter So performance can be seen The difference in performance can be seen If you look at the EMIRR rejection plot of two parts which are OPA– which operates similar to the op amp that I mentioned earlier And you can see EMI rejection ratio is pretty good So slew boost technology– what it is At very high level, the slew rate is limited by a compensation– a compensation capacitor and the current flowing into it from the first stage So what slew boost technology does

is it senses the input signal, and when it is large, it makes sure that there is extra current– another source of current is put in parallel with the first stage current and helps charge that compensation capacitor faster when the signal– the differential signal– at the input of the op amp becomes smaller, it turns off the extra current path and lets the internal frequency of the op amp take over and settle the large signal So if you look at the plot on the bottom-left side, you will see an op amp without slew boost technology, and on the right side next to it, you’ll see the op amp with slew boost technology, and the first part of it is when slew boost activates, and the second part is when the inherent op amp frequency takes over, or again, bandwidth takes over and settles it And one of the benefits of this is not having to use an op amp with a lot of IQ just to get the higher slew rate So OPA2991 and 4991 that I mentioned earlier implements this technology So MUX-friendly input– what is it? So our old high-voltage amplifiers have these back-to-back diodes connected between its inputs you see here These are put in place to protect the input stage of an op amp And if there is an application where an op amp takes multiple inputs from a MUX and those inputs have wide, varying voltages, and when the input is switched between different inputs by the MUX, you could have a condition where they diodes turn on, like shown in this figure, where one input is at 10 volts, another input is at -10 volts So what it does is it affects the settling time because of that current So in our new MUX-friendly input, what we have done is we have patented MUX-friendly input that are implemented without these back-to-back diodes What it does is it still protects the inputs but doesn’t cause any input current to flow through these diode kind of activities So if you look at the plot on bottom-left over here, the green one is the one with the MUX-friendly input and the red one is the one with the traditional back to back diode structure And you can clearly see how there is a difference between the settling time So I’ll pass the ball to– OK I think you’re good to go, Pete Thank you very much Thank you, Sanjeev Thank you, Conan I appreciate the opportunity to talk to everyone this afternoon As was mentioned earlier, my name is Pete Semig I am the general-purpose amplifiers applications manager located in Dallas, Texas And for the next 45 minutes or so, I’m going to present to you some tools and resources– about 10 minutes of tools and resources– that TI has available on ti.com that can certainly help with your next design And then also after that, for the remainder of 30 minutes or so, we’ll discuss some technical topics, namely offset voltage and input bias current And then also we have some slides on input and output limitations, which are common mode swing and output swing And then finally, if there’s sufficient time remaining, I have slides on CMRR and PSRR, which are common mode rejection ratio and power supply rejection ratio So for tools and resources, TI has a lot of resources on ti.com available for download The first one I’d like to highlight is the “Analog Engineer’s Op-Amp Cookbook.” This is a downloadable PDF that has at least 50 common op-amp circuits There is also an “Analog Engineer’s Circuit Cookbook for Data Converters” that can also be downloaded

And these cookbook circuits are fairly simplified They’re about 3 pages long And they start out with a set of design goals, input, output, supply voltage, et cetera They’ll give a basic circuit topology, along with some design notes on the front page The next page will have the equations and different mathematical relationships that you can use to size resistors and capacitors and such And then finally, the third page will have simulations which are in our simulator, TINA-TI, which will have AC or DC or transient simulations And then finally, they have the links inside of them that you can download the circuits So if you want to tweak the design a little bit, you can then easily get up and running very quickly with those designs by downloading these simulations and tweaking the circuit as you see fit We also have about 10 of them that have accompanying videos as shown at the bottom Those include circuits like NTC thermistor circuits, PTC thermistor circuits We have a low-side bidirectional current sensing one, and one of our most popular is a transimpedance amplifier video as well that, within about four to five minutes, will walk through the entire design As far as EVMs go, this is the DIY amp, or do-it-yourself amplifier EVM This here is a sheet of PCB that has scores on it where you can literally flex it and break them apart So on the left side, you see the sheet and it has the capability to be used with three different package types and a dozen different amplifier configurations So upon receiving this sheet, this EVM, you can solder down an SC70, a SOT23-5, or an SO-8 part onto the board You can also then put in your caps and resistors and header pins or SMA connectors and evaluate a bunch of different circuit configurations And I have some of those listed here– non-inverting op-amps, inverting op-amps, diff amps, filters– so I think there’s an MFP and a sample key filter on there, also a spot for evaluating comparators, et cetera You can use dual- and single-supply And, all in all, this is a very cool product that you can use to quickly evaluate different devices in common configurations We also have a version of this for dual packages, as well, as shown in the lower right-hand corner One of the nice things about the DIY amp EVM is that you can actually take multiple PCBs, you know, multiple EVMs, and chain them together And in this case here, we show an example of three second-order filters combining to give you a six-order filter And you see that we have our function generator, along– or, well, we have our Bode plot or Bode 100 gain phase analyzer that can connect up to these and actually show you what the performance would be of these three stages of filters, all connected in series And you can do that with anything– with your diff amps, non-inverting amps, or any of the other configurations So it’s kind of a plug and play sort of setup We also have what’s known as TI’s “Pocket Reference Guide,” and this is a very popular download This one is not only a download but you can also– there is also a mobile app for it, and you can also purchase a hard copy of this I literally have one on my desk right now I keep one with me at work, in my bag It is very handy It basically has all the information that you find yourself– or at least, I find myself looking up on the internet all the time So, for example, it’ll have equations for op-amp noise It’ll give you all of the information about different discrete components So, you know, when are you going to use which type of capacitor,

et cetera And it is a very, very handy PDF or document to have available And along with that is what we call the “Analog Engineer’s Calculator Tool.” The “Analog Engineer’s Calculator Tool” is a companion piece of software that you can go to ti.com and you can download this and install it on your computer And it contains most of the equations in the “Analog Engineer’s Pocket Reference Guide,” but they’re already implemented in different calculators So you can see right in the middle of your screen that there’s sections for data converters There’s a section for amplifiers and comparators, a bunch of calculators for passive devices One of my favorites, under passives, is for calculating the values for a resistor divider What’s neat is it will use standard 1% or 0.1% values to calculate the nearest value for a resistor divider There’s also all the equations on noise and stability, a bunch of different calculators on PCB layout, based on copper thickness and whatnot So a very handy tool to have that is a companion to the “Analog Engineer’s Pocket Reference Guide.” We also have a fairly new way to search for op amps This is what we call a quick search by circuit function, and this is actually a companion to the cookbook from earlier And the idea here is that many, many times you go on the web and you’re searching for an op amp, for example, and you plug in things like gain bandwidth, your power supply, and you know the end application in mind Let’s say it’s low-side current sensing, but you don’t necessarily translate– or it’s hard sometimes to translate the specifications of that design to actual parameters of an op-amp And that’s what this search by circuit function does Upon selecting a circuit function– and you’ll see that we have eight of them here, dual and single-supply, inverting and non-inverting amplifiers, we have a diff amp, low side, low-side current-sensing amplifier, transimpedance amp, as well By selecting that, it basically invokes the equations from the cookbook circuits shown earlier so that as you type in parameters for your design, it will automatically translate those to filter constraints in the online op-amp searching tool So, for example here, if we were to select an inverting amplifier circuit, as shown here– a single 5-volt supply and the input is 100 millivolts peak to peak, and it has a 0.1 megahertz frequency with a gain of 10 And let’s say you only wanted one channel– so you use a slider bar to change the channel count– what it will do is automatically calculate for you the parameters of the supply voltage, the required gain bandwidth, the required slew rate, and then the input common mode and output swing that are required of the op-amp to implement this function And the reason why I love this example here is that if you look at the gain bandwidth, it says the typical gain bandwidth has to be 1.1 to 5.1 megahertz Well, how would– you might say, well, wait a minute The input frequency is 0.1 megahertz and the gain is 10, why isn’t it just 1 megahertz? So one of the tricky things about this is that when you’re dealing with bandwidth, for example, you actually have to always use the non-inverting bandwidth– or the bandwidth of the gain rather, as seen by the non-inverting input pin of the amplifier So for this particular circuit, the gain– the non-inverting gain as seen by Vcm is 1+R2 over R1 So that would actually be a gain of 11, not a signal gain of 10 For more information on that, we certainly

have those lectures in our TI Precision Lab series on ti.com, and I’ll point that out shortly But it’s those sorts of nuances in a design that this can certainly help you with as you design your next circuit Another thing that I would like to announce today that has come out very recently is a simulator called PSpice for TI PSpice for TI is based on Cadence’s ubiquitous PSpice tool, as shown here This is a very powerful, fully-functional Spice engine And you can download it at ti.com/tool/pspice-for-ti All you have to do is request access for it Generally, that’s pretty immediate, and then you can download and install it There’s also a series of video trainings for it It has a curated library of parts that will synchronize with ti.com It’s also important to realize, perhaps for your end equipment, that the models for general-purpose amplifiers are all unencrypted, so you can definitely take a look at inside those models by right-clicking on it and say inspect model This has the– if you’re used to using OrCAD or Cadence, it’ll be the exact same sort of environment, so that should be an easy transition for many folks And then, as I mentioned earlier, it is a fully-functional SPICE simulator, which means you can do some of the more advanced analysis, such as Monte Carlo analysis, worst-case analysis, parametric sweeping, et cetera I should mention that the only restriction on PSpice for TI is the number of nodes that you can plot It is unlimited if you use TI parts or anything from the curated library within the tool If you import a model from elsewhere, the number of nodes that you can plot will be limited to 3 So that it is important to note that But as long as you use all TI devices or everything that’s been signed with by TI, you can use as many node probes as you desire And then there are many different use cases for it, from simple test benches to complex subsystem design So it is a fully-featured Cadence PSpice engine And then, finally, to segue into our technical topics today, I’d like to introduce you to TI Precision Labs, which is located at ti.com/precisionlabs And all of the information from now forward is available and at this URL I’d like to note that for offset voltage and input bias current, in particular, and then also common mode rejection ratio, I did remove some of the material in order to meet some time constraints But if you want additional detailed information, you can always go to ti.com/precisionlabs to watch the full video and also get all of the slides that I’m going to present today and more But Precision Labs is basically a whole bunch of curriculum, or curricula– one on op amps, one on ADCs, a curriculum on comparators and multiplexers with relatively short 15 to 20-minute videos that describe many different aspects of the devices So today, like I said, we will be discussing these in yellow, which is going to be input offset and input bias current and input-output limitations And then we will, if time allows, look at common mode rejection ratio and power supply rejection ratio OK, so moving forward with the technical portion of this discussion, let’s take a look at input offset voltage and input bias current of an op amp So input offset voltage or Vos is defined as the offset voltage required across the input pins of a device

to drive the output voltage to 0 That’s the technical, academic description or definition of input offset voltage It is very typically modeled as a DC voltage supply in series with a non-inverting input As a matter of fact, many errors– as we’ll see later, if we can get to CMRR and PSRR– are all modeled from a DC perspective as DC supplies in series with the non-inverting amplifier of the op amp And in this case here, we’re showing an op amp that has an offset voltage of 50 microvolts Now, you might ask, well, where does this offset voltage come from? In school, you learn that that offset voltage is ideally zero Similarly, you learn that the input current or input bias current flowing into the input pins of an op amp is also zero However, in the real world, op amps have an input bias current that is non-zero, and we’ll discuss those shortly So the input offset voltage or Vos manifests itself due to mismatches of the input stage transistors, as shown here Mismatches between Q1 and Q2 manifest themselves as what I call the initial input offset voltage And it’s that initial input offset voltage that has to be corrected for in order to set the output equal to zero Now, there are many different types of amplifiers In this case here, we’re showing a type of op amp that has some resistors in series with Q1 and Q2– those are Ros1 and Ros2 Those resistors can be trimmed to help match the characteristics of Q1 and Q2 better in order to minimize that offset voltage However, there are more advanced techniques, such as auto-zero techniques and chopper amplifiers that use digital correction in order to minimize those offset voltages This is a very important slide, perhaps one of the most important slides in this discussion because this slide describes basically how to read a datasheet, especially for offset voltage You’ll see at the top of the screen in the green box that there is a whole bunch of parameters, and it’s those parameters that are understood that must hold true in order for the device to meet the datasheet specifications in the table So in this case here, these parameters in the data sheet are valid when the total supply voltage is 1.8 to 5.5 volts, valid when the ambient temperature is 25 degrees C, also when the load resistor is 10 kilo ohms and connected to mid-supply, and then also that the common mode must be also equal to mid-supply, and the output is also connected to mid-supply Mid-supply is Vs divided by 2 So the idea here is that the device, in order that the device meets these specifications when in this configuration Now, sometimes there are some test conditions that must be specified that when the device– that whenever there might be things that are different than what is listed at the top of the table So, for example, in this case for the offset voltage of the TLV9062, the tested condition for offset voltage is with a 5-volt supply So even though at the top of the table it says 1.8 to 5.5 volts, the offset voltage specification is really only good for a supply voltage of 5 volts You’ll see over on the right-hand side that the typical value is 300 microvolts and the maximum value is 1.6 millivolts And those are both plus or minus So offset voltage can be either positive or negative, depending on the mismatch of the input stage transistors Now, what does it mean to have a typical or a maximum value? A typical value– and again, this is a general statement, it’s not the case for every single amplifier– but in general, a typical specification in an op amp datasheet means that we’re talking about a plus or minus one standard deviation,

or plus or minus 1 sigma of the devices that were tested fall within this typical value So that’s about 68% of the devices So the idea is that 68% of the devices that you receive from TI for the TLV9062 should have an offset voltage, initial input offset voltage, of 300 microvolts or better Now, it’s often the case that you need to use some statistics to expand this because you know this is a typical Gaussian distribution And if you look at plus or minus 3 sigma, so that would give you about 99.7% percent of the distribution So in this case, 3 sigma would be 0.9 or 900 microvolts, would yield 99.7% of the devices However, the maximum value– and depending on the specification of the minimum value– is the tested limit– or sometimes ensured by design limit– that upon receiving advice from TI and meeting all of the test conditions and connected as shown at the top of the table, that is the maximum value that we would allow this device to be shipped with So in other words, if you connect up the device, given the test conditions and the conditions at the top of the table, you should never receive a device– a 9062– with greater than plus or minus 1.6 millivolts of offset voltage because that is the value that we actually test it at Our models simulate the typical values So what’s important about this slide here is that it shows the connections in accordance with the top of the table So you’ll see that the device 9062 is connected with a single 5-volt supply The input is connected to a common mode voltage or Vcm of 2.5 volts The output, R1, is connected– the 10 kilo ohm resistor load– is connected to a mid-supply voltage of 2.5 volts And you’ll notice that the offset voltage is equal to 300 microvolts, which corresponds to the datasheet Now if you use a different power supply, if you have a different common mode voltage, if your load is different, these things will start changing So if we take a look at a quick example here, we have a non-inverting amplifier circuit with a 1 millivolt input signal And in series with that is the initial input offset voltage of 0.1 millivolts The gain, as seen by the non-inverting input is 1 plus 99k divided by 1k, which is equal to 100 And then if you were to take the 1-millivolt input and multiply it by 100, you would expect to see 100 millivolts on the output However, that offset voltage– initial input offset voltage of a tenth of a millivolt– is an error source So you could actually get, in this case here, 1 millivolt plus 0.1 millivolts times 100, which is 110 millivolts So in this case here, Vos actually introduces a 10% error Now, of course, you could also just look at the ratio of the initial input offset voltage to your signal and determine that 0.1 is 10% of 1 millivolt. That works as well Now let’s switch gears to input bias current Input bias current is the dual of input offset voltage– input offset voltage was a voltage source in series with the input– input bias current is modeled as current sources from each input to ground And in this case here, we see that we have an input bias current for the inverting terminal of 150 nano amps, and for the non-inverting terminal, 210 nano amps The difference of those two is what is known as the input bias offset current In this case, 210 minus 150 is 60 nano amps So where does input bias current come from? We saw input offset voltage came from mismatch in the input stage For input bias current, it depends on the topology of the device In this case here, we’re showing the input stage of a simple bipolar device that has no IB cancellation– we’ll talk about that in a second And what you see here is that Ib1 and Ib2 are

the base currents of Q1 and Q2 And these are typically of the nanoamp magnitude So you see, for example, input bias current at the bottom red box is typically 80 nano amps up to 500 nano amps max And the offset current is from 20 to 200 nano lamps However, what we could do is, let’s try to– if we knew the amount of current that’s flowing into the base and we measured it, and this is the effective idea of IB cancellation– where you could inject an equal but opposite current into that node, kind of like Kirchhoff’s current law, you could effectively cancel out the input bias current That’s what IB cancellation does So in this case here, you see that we have an input bias current cancellation circuit injected into the base nodes of Q1 and Q2, and that greatly reduces the input bias current and input offset current So the two red boxes on the lower part of the screen you’ll see are now 0.5 and 1 nano amp Now, it’s important to also realize that these are now either plus or minus because if you inject a little too much current or not quite enough current, you can have current either flowing into or out of those input pins So that’s what input bias current cancellation is Now, for CMOS devices, it turns out that the input bias current is actually dominated by the ESD protection diodes of the amplifier, as shown here So Q1 and Q2, while they do have a gate current, it’s very, very small compared to the ESD diodes shown on Vin1 and Vin2 And it’s those leakage currents for those diodes that actually create the input bias current for the device Now, you’ll see for CMOS devices the magnitude is pico amps, as opposed to bipolar devices, we were looking at nano amps So you see that CMOS devices, in general, have much lower input bias current than bipolar devices OK, let’s move on to input/output limitations So I/O limitations are violations of input and output limitations, or perhaps the most common mistake made when developing or designing an application circuit And the simple fact of the matter is that you need to ensure that the input signal to an op amp and the output signal of the op amp are both within linear operating range Otherwise, the device is not going to act like an op amp, OK? It will be a non-linear operation If either the input or the output are outside of linear operating region then, obviously, you will not have a linear operating op amp So let’s take a look at an example And this is a fairly common thing We have a device, the OPA735 with dual +/- 2.5-volt supplies And we get a question that says, hey, we input a 3V peak to peak– or 1.5-volt peak– triangle wave, and our output, unfortunately, seems to clip at 1 volt. There’s something wrong with this op amp Both the input and the output signal are clearly within the supply voltage– 2.5 volts– and it’s just operating as a buffer What is wrong with the OPA735? Well, it turns out there’s nothing wrong with the OPA735 We need to delve a little bit deeper into the common mode and output swing limitations of the device Before we do that, let’s define common mode voltage, or Vcm Common mode voltage is defined as the average voltage as seen by the input pins of the device So Vcm equals Vin+ plus Vin- divided by 2 Now, if they differ a little bit, as we already learned, that’s due to the offset voltage And for example here, that’s in– depends on the device– could be single microvolts up to a handful of millivolts, depending on the device In this case here, we’re looking at an offset voltage of approximately 30 microvolts But by and large, Vin+ and Vin-, if you add them together and divide by 2, that is your common mode voltage And the input and output swing voltages–

or the input and output voltage swings are both defined based on the supply So in this case here, the OPA140 has a single 10-volt supply, and therefore the common mode voltage and the output swing must be within the range as defined by the OPA140 datasheet, which is dependent upon V+ and V- So let’s take a look at an example Here’s the OPA140 In red, we have the equation for the common mode voltage, and in purple, we have the equation for the output swing So the common mode voltage ranges from V- minus 100 millivolts to V+ minus 3.5 volts So in this case, if we have a single 5-volt supply, the range of Vin must be between -100 millivolts and 1.5 volts The output has a different equation The output is V- plus 200 millivolts to V+ minus 200 millivolts So therefore, given a single 5-volt supply, the output swing must be between 200 millivolts and 4.8 volts If the common mode, or the output swing tries to go beyond that, the op amp will be non-linear Now, one way to sort of circumvent these potential issues is to look at the configuration of the op amp On the left hand side, you have inverting configuration, where the non-inverting pin of the device is connected to ground The inverting pin, of course, is the summing node And the way that an op amp works is that it’s going to output whatever voltage is required to make the inverting pin equal to the non inverting pin In this case here, that would be 0 So if you calculate the common mode voltage of the inverting configuration on the left, you would have 0 plus 0 divided by 2, which is 0 So the common mode voltage of an inverting configuration is always equal to the voltage placed at the non-inverting node, and that’s because, by definition, the output will try to get the two input pins equal So even though the input voltage, which is 1 volt on the left hand side– even if the input voltage of an inverted configuration changes, the op amp will always change the output voltage to try to make the two input voltages, in this case, equal to 0 On the right-hand side, in a non-inverting configuration, the input is connected to the non-inverting pin So as the input voltage changes– as that 1 volt changes up and down– the output is going to try to make the inverting pin equal to the input So now your common mode is constantly changing In this case here, you have 1 volt on the non-inverting pin plus 1 volt on the inverting pin, is 2 volts divided by 2, is 1 volt. So since they’re both approximately equal to each other, as the input moves around so while the common mode So if you wanted to make sure that you don’t have to worry about common mode issues, you could use an inverting configuration and just make sure that whatever voltage is present at the non-inverting pin is within the common mode range of the device So let’s take a look at a quick example Here’s the OPA140 and it is connected with a single 5-volt supply The input is 0 volts, connected to ground, and the output should also be 0 volts However, upon looking at this, you see that the output is actually 171.4 millivolts So when you look at the datasheet, the offset voltage is only 120 microvolts maximum So how in the world can the output be so much larger than the offset voltage? Well, what’s happening here is we’re violating something The op amp is not in a linear operating range So hat’s going on? Are we violating the common mode range or the output swing So first, if we calculate the common mode range, given a single 5-volt supply, the input should be from -0.1 volts to 1.5 volts So our input is 0, and 0 falls within this range Therefore, this is not a common mode violation

However, if we look at the output, we’re trying to output 0 volts, right, because our input is 0 We want the output to be 0 But if you calculate the output swing, it’s V- plus 200 millivolts to V+ minus 200 millivolts, which is 0.2 to 4.8 volts We’re trying to output 0 volts, but we can’t We’re running into the output limitation So one potential solution is to just simply shift the supplies So if we shift the supplies– instead of using a single 5-volt supply we use a dual +/- 2 and 1/2-volt supply, well, that should fix it, right? Because our input is right in the middle and our output is right in the middle, right, because our input is 0, which is halfway between +/- 2 and 1/2 volts, and our output should be 0, which is halfway between +/- 2 and 1/2 volts Upon doing that, however, notice our output is even worse than it was before Our output is now negative 260 millivolts, and the previous one was about 170 millivolts So this actually got worse So what’s going on here? Is this a common mode issue or an output swing issue? If we take a look at the output and calculate its range according to the datasheet, the output is valid when the voltage is -2.3 volts to +2.3 volts And we’re trying to output 0, so that should be fine However on the input side, when we use the equation in the datasheet, the valid or linear operating region of the common mode voltage is -2.6 to -1, and we have connected our common mode voltage to 0 So 0 is outside of that range, therefore we have a common mode range issue That manifests itself as non-linear operation So if we go ahead and take a look at the original problem that we were discussing earlier, common mode or output voltage problem, and we calculate the common mode range and the output swing, you’ll see that the output swing is valid from -2.48 to +2.48, and we’re trying to output a 3-volt peak to peak triangle wave that goes everywhere from +1.5 to -1.5, so the output is perfectly fine The common mode, however, goes from -2.6 volts to +1 volt, and we see that manifest itself as clipping in the simulation Now I do want to take a second to note that in this simulation, we see it as clipping In the real world, it could be anything It could be a sort of, you know, slightly nonlinear curve that almost gets up to 1.5 volts It could be a harsh clipping like this It could be anything But whatever it is, the idea is that when you see a non-linear behavior, definitely look at your common mode and your output swing limits in the datasheet to ensure that you are within the range So I see that we only have a couple of minutes left, and I’m not sure that we’ll have enough time to go through CMMR and certainly not PSRR But I did want to introduce one quick concept, and that is the RTI or refer to input concept And this slide here in the CMMR section shows what I mentioned earlier about offset voltage And you see that what we have here is a stack of a bunch of different voltage supplies in series with a non-inverting input And the idea is that when we characterize an op amp, we refer all the errors to the input, or as many of them as we can, because we don’t know how the device is going to be used Even though it’s a 5-volt device, it might be used at 3 volts We don’t know many things about the configuration Is it inverting or non-inverting? There are many different questions as far as how the device is going to be operated So by referring all these errors to the input, that enables you to separate the error terms and then add them back up based on your configuration So in this case here, you see that we have six different error terms– error due to EMIRR, PSRR, CMRR, AOL, CMRR with respect to temperature, and then also offset voltage, and Vos is the initial offset voltage

And what you do is once you’ve put your device in a particular configuration, you can then calculate the individual contributions of each of these error terms and then ultimately add them together I generally recommend that you add them together as an RSS, or root sum square, because that’s more realistic as opposed to adding them directly So that is all I have for today I want to make sure that there’s enough time for the next discussion Thank you very much