Hi Welcome to Fundamentals Friday This is a follow-up video from my previous one about measuring input offset voltage and some issues I had with an Analog Devices part. If you haven't seen it, click down below and you'll be able to watch that it's a 30- minute video of me McKing around measuring some input offset voltages. Great fun, But today we're going to do something related which we'll get back to on the breadboard and solve our problem from last time we've got Opamp input bias currents today. What is an Opamp input Bias Current and why is it important? Well, let's take a look at it.

You've got your basic opamp here. Doesn't matter what Opamp it is, they're all going to have what's called an input bias current now. Uh, you're used to dealing with an ideal Opamp now an ideal opamp. Of course.

One of the rules for the ideal opamp is that no current flows into the input pins. But of course in practice that's complete because you're going to have some current flowing into or out of. We'll get into that these input pins on this opamp and that's called the input bias current and I've labeled that here IB plus and IB minus. Now a data sheet for a typical Op amp will only usually only show the input bias current as one figure I B they'll typically call it.

They won't actually specify different input bias currents for the positive and negative, and that's where we'll get into later is the input. There's there's two parameters to do with input bias current. One is the input bias current itself. IB The other is IOS which is the input offset current and that is actually the difference between these two.

but we'll get into that in a minute now. This input current in practice can range anywhere from Fpto amps Really? Schmick Precision Instrumentation grade Op Amp might have you know a few tens of of Fto amps input current or for a really fast, wide bandwidth opamp. They don't care about the input current, so it's going to be a couple of microamps. They're going to optimize the internal architecture for that opamp.

They don't care about the input bias. Currently, they're going to optimize that for bandwidth and slew rate and all sorts of other things. So there's a pretty big range there in the input bias current. So how does it work? Well, let's take a look at the input circuitry or the basic topology of the input circuit for a Bipolar operational amplifier.

So this is not a fed input, it's a bipolar input opamp. And you notice that we have two input transistors here, both Npm going into a current Source it's a current mirror configuration and I won't go into all the details of, well, any significant detail of input opamp architecture that requires a whole separate video. But suffice it to say look, these inputs are connected directly to the base of these transistors. Here, your positive negative input on your amp.

Op Amp over here goes into these transistors. So naturally you're going to get some base current flowing in there and in there as well. and there's ways to lower that, of course. Um, if you use a Fet input Op amp.

well, Fets aren't effectively right. They're a Field Effect transistor. No input current, but in practice there is. In theory, there's not in practice.

but this is a bipolar Um operational amplifier. You'll probably the most common Uh one on the market, although Fet ones are pretty darn common as well Seos ones. and there's other ways to lower this input bias current as well. They can actually use a Darlington Uh configuration here.

Uh, so that your input bias current is lower it still using the Bipolar manufacturing process, but uh, there's issues with that in terms of bandwidth and all sorts of things which I won't go into. and the other way to do it is what's called a biased Uh bias offset operational amplifier or a bias offset input. And what they do is they actually put in here a current Source in there and in here as well so that that biases the transistor instead of the input bias current So In theory, it cancels it out and you can get very low input bias currents from the input, but not zero. They almost never.

Well, in practice, you can never get down to zero bias current, so every practical opamp it regardless. Even if it's input bias compensated is going to have some form of input bias current or IB and it'll be mentioned in the data sheet. Now these bias currents. They can be a real pain in the ass in Precision applications and there is quite a few uh different ways it can screw you up.

Let's take a look at the first problem is the your uh Source impedance of uh well, your Source impedance. Okay, that's the series resistance of your Source Let's say it's 10K might be a typical value for um, some systems and if our input bias current is a very low 100 perram, right? that's a pretty good opamp. Very low input bias current at 100 P amps. Okay, then our error that we're going to introduce in our system without ignoring everything else, offset errors, and all the rest is going to be 10K times that 100 pamps because the 100 pamps is flowing through that 10K resistor.

We've got a one m molt offset already. one microvolt error and doesn't sound like much, does it one microvolt. But if you're dealing with a Precision system then that can like a you know strain gauge or something else you're dealing with. That can be a real big deal cuz that these errors get multiplied by the gain of your Op amp so it can screw you up right there.

Just be careful. And 100 Peak amps? that's a very low one. So imagine if your Source impedances like a Meg or something like that and you're uh, bias currents even higher, you can get molts of error. Watch out.

Now, the second problem we've got is errors introduced into our Uh gain Network and our feedback resistor Network for this thing. Now what happens? Let's take this uh, invert in opamp configuration and here's our input voltage. That's our output of course. and let's ground this input here and see what happens Now, because the input the uh, non-inverting input down here is also ground.

And once again, we go back to our ideal opamp properties. We're assuming our input offset voltage Remember, we're not talking about offset voltage yet is zero. So the difference between here and here is going to be zero. Remember, due to opamp action, the Opamp does whatever it needs on the output to ensure that the two inputs are exactly the same voltage.

If you're input, offset voltage is zero, which we'll assume. So this point here is also zero volts. It's also grounded, so there's no current flowing through R1 there at all. So, but we still have an input bias current.

Remember, we always have that it can't get away in this practical configuration. So where does it come from? Well It can't come from through this resistor here because Ohms law says Zer volts across that resistor doesn't matter what value it is this. R1 No current flows through it, so all of the current must come from r F Here in this case, it flows around like that. So all of our input bius current is flowing through RF and you might see the problem already usually to you want on an Op amp.

you're going to have some sort of gain times 5, 10, 50, 100, even a th000 depending on your Um. you know your circuit application well. that means RF has your feedback resistor has to be a large value. What do you get when you have a large large value resistor with a current flowing through it IB Even a very small one.

Aha, you're going to get an offset error introduced over and above your Vos your which is another figure on the data sheet. your Uh input offset voltage. This is in addition to whatever that is. So this is where input bias current can be a real pain in the ass for these Precision applications.

All of that bias current flows through RF So how do we solve these input bias current issues? They're a pain in the ass for these Precision applications. How do we get rid of it? How do we cancel them out? Well, let's uh, have a look at what we got here. IB plus and IB minus one goes into the Uh inverting input. one goes to the nonin inverting input.

so they're essentially opposite currents so you can actually cancel them out. Now let's go back to our original non-inverting configuration here and we've got a simple voltage follower. Okay with our Source impedance RS in here. Whatever value, it is causing the error due to the input bias current flowing through it.

Well, we know that the other input has an equal and opposite, or should have an equal and opposite input bias current. How do we cancel it out? Very easy. We just put another a feedback resistor in there equal to, let's call it Rs2 equal to the value of RS. So if IB plus and IB minus are equal values except opposite because they're going into opposing inputs of the opamp, I won't go into the internal details of it.

but then you can cancel them out by having a a resistor in the feedback loop like that. And how do we solve this inverting opamp configuration here? Well, essentially exactly the same thing. Instead of this going to ground down here, we have a resistor in there going to ground. So we're going to get our input bias current and we can actually cancel them out like that.

Now let's call this RB and what value should RB be? Well, it's a standard formula. It's actually RF in parallel with R1. Now, I won't go into how you actually derive that. It's not actually that hard if you want to go into the details, but I won't.

Uh, waste the time doing that. But that is a basic formula and this is a very common thing that you'll see in a lot of circuits and you might have wondered why. Well, why does it have this resistor in here? Is it protecting the input somehow? No, You know it's to do with the input bias currents in these Precision circuits. And of course, if you got a very large gain here an RF is much greater than R1, then you can just say RB equals R1 because the parallel combination of RF on top of that H.

To a rule of thumb, you know, to an order, if you basically got a gain of 10 or more and RF is 10 * R1, you can generally say RB equals R1 near enough. So that was too easy, right? We fixed our input bias current. Eh, No big deal, just whacking a resistor in there And you're right. Unfortunately, what doesn't work like that In practice Because the input bias currents are not matched, these transistors are never a matched pair.

You're never going to get precisely the same value. There's some opamps that get really, really close, but in practice, it's not going to be zero. Just like the input bias current itself, regardless of the topology used, is not going to be zero either. And that's where this iOS comes in.

The input offset current is the difference between IB plus and IB minus. so there's actually two parameters there that you have to look at. not not only the input bias current itself, like the input bias current could be really low. it might be 10 Pico amps for example.

But then if your Uh iOS is 100 Pico amps for example, then well, jeez, it's all over the shop. There's no way that you can choose a resistor buyer's value that's going to work. You might fluke it. You might be able to tweak it for an individual uh chip or an individual circuit, but just as choosing a generic value to do it.

No, you're not going to be able to do it. Unfortunately, it is just a problem with these: Ops If you're really, if you're doing Ultra Precision applications. Yeah, you'll have to trim these things to take into account the input bias currents. and on top of that, then we've got our voltage offset as well.

Oh, so you could get to a point depending on your particular Op app you've chosen that iOS could actually swamp your effective IB or vice versa. Or there's a combination of the two and it can get really ugly. But on top of that, you wouldn't know. It's even worse.

Well, you might have a Railto rail opamp now I mentioned right back at the start. I Think that the input bias current can go either direction. Now on a standard bipolar opamp uh, like this, or on a Fet input opamp for example. you're typically going only going to get your input bias current going one way, but on railto rail opamps, they're internally biased and they've got different configurations and I won't go into the transistor configuration of an of a railto rail input opamp because there's lots of Tricks They can actually uh uh, vary and stuff like that.

But basically it means that current can also flow out of these pins as well depending on the common mode voltage you're operating at. So if this is zero, for example, you might have a curve like that where your current can be bias, current can be positive and negative, so this can be IB here positive and negative and at some transition point where you pass through it, you might end up going and the current May flow out of the these pins, back out and ruin your day. It can get really ugly, so there's all sorts of problems you got IB you got iOS you got V offset You've got whether or not your input, uh, configuration, your topology, your circuit, and all sorts of stuff. Is there an easy solution? No afraid, not.

So in practice, this can be a really big issue, even for circuits that you might not think are that critical. so you got to keep an open mind. Watch out for it, just in case you get caught. I mean depending on whe what? Supply this can change with Supply voltage as well.

The common mode here, which is, that's V common mode here uh, depending on what input configuration, whether or not you've got an input bias uh, compensated or railto rail opamp, uh, the supply voltage, all sorts of stuff, and which configuration you're using your Source impedance? Oh, and we haven't even mentioned temperature. so how do you know if your Opamp he has one of these. uh, input bias? uh, compensator? VAR Well, you're typically going to get a plus minus IB value like that, they're usually not going to give it for your input offset, but plus minus IB That's a dead giveaway that the opamp you're using, even though they may not tell you, is input bias compensated, so you can see how complicated this is already getting. And really, you know we're not even throwing Vos into the mix much.

The input offset voltage which the gain is going to get multiplied which is going to get multiplied by the gain of your opamp here and it can get really, really nasty. So if you see any input offset issues, they will typically be a combination of the real input offset voltage which is a separate parameter Vos entirely separate parameter on your data sheet up here. but it can also include your input bias currents and that's what we saw in the previous video. If You haven't seen it.

Link it down below. So let's now go back to the breadboard from that previous video and see if we can reduce or eliminate or compensate for these input bias currents and see how it affects our final output offset error and just a quick background to our previous video. and you really should watch the Uh previous video. It'll be linked down below if you haven't seen this.

Otherwise, it may be a little bit confusing for you. But basically we had an Analog Devices Ad 862 8 Opap with time 100 gain in the non-inverting configuration and uh, it were and we were getting an output voltage which was higher than just the expected offset voltage. Here you here's the actual circuit we' got. we've got 1K going to ground here.

We got 100K here. total gain of 101. We're going to call it 100 near enough and uh, the input offset voltage of this particular Op Amp is supposed to be around about Uh 1 microvolt. so we only expect sort of around about 100 microv volts on the output here, But this is our output voltage and we're getting over 300 microvolts here and this is our supply voltage.

We're using a split Supply here so the ground point is actually in the middle so it's plus - 2 1/2 volts. but our output voltage is higher than what we expect from just a typical device. And technically yes, it is within spec because the here it is input offset voltage V there at 5V Supply is a typical value of 1 microvolt, but it could be as high as 5 microvolts so that would translate to 500 microv volts on the output here. But as I said in the previous video, that's not actually the case.

This opamp is actually typically one microvolt or less and I've tried multiple opamps and the and that error is coming from somewhere else. Well guess where input bias currents and if we have a look at our data sheet sheet here, you'll notice that as I explained before, there's a figure for IB there input bias, current and input offset current right here. Here it is the 80 Uh 8628. it's going to it.

Actually, Uh gets higher with the Quad package, by the way. That's just to do with the Uh process technology that they're actually Uh using. They got four of those on the one die and it's different. So anyway, we've got the single one, the 8628 input bias current typically 30 Peak ramps.

It could be as high as 100 Pico amps there, but you know it. We are going to get the typical figure down in here. So aha, 30 Pico amps. Let's have a look.

Is it a coincidence that our output is just over 300 Uh microvolts here? Let's do the math. shall we? 30 If I be here? Let's ignore this input here. Okay, let's ignore the non-inverting input. Let's just look at the inverting input here.

If our input bias current let's assuming it's going into to the pin, then uh, if it's 30 PCO amps going in, You remember before, if we ignore the V offset, then there's no voltage drop across this resistor. So all the input bias current that whole 30 Pico amps a u you know, typical is coming through that 100K resistor. Aha. So let's do the math here.

30 pamps there it is times k equal 3 microvolts. So we're getting 3 microv volts error just due to the current going in there, the input bias current into that inverting input there. But the gain of the Op amp of course is 100 so we have to multiply that by 100. And of course you get 300 microvolts.

Uhuh, look, 300 microvolts are just over. Is that a CO coincidence? I Think not. Well, it's not going to be the entire story, but it is certainly going to be a lot of the reason for this. And by the way, um, if you haven't seen the previous video, this does change with Supply voltage.

So if we drop the supply voltage down, you can actually see it change significantly and even go negative like that. You remember what we were talking about with Oop, Sorry, it should only go to 2.7 There we go, 2.7 It's almost going negative there. So that does change with Supply voltage as I mentioned previously in the video earlier. Now of course that's not going to be the only reason, but it's going to have a significant effect on that.

Of course, our V offset is going to come into it and V offset's going to get multiplied by the gain of 100 as well. And then we haven't taken into account the Uh input offset on the other pin and stuff like that. but there you go. That is going to be a very significant reason for it.

So so the previous video yes I mean I've done this stuff before, but it's actually been. you know I don't know 7 years or something since I last uh touched this sort of stuff. So I forgot that the Um error term is going to be coming through the 100K resistor here. and it didn't help that The Analog Devices data sheet actually shows these exact same values in a typical Uh example test circuit.

So what you know: I just overlooked it. That can easily happen to even somebody who's done this stuff before and I um made the assumption that it was only going through the 1K which is incorrect. All of that bias current is going through there. That's why I I knew that I could have put an additional bias resistor in here like this.

but then I knew, you know I I Sort of did the math in my head as I was going along and I went. Oh, it couldn't make an effect. I could make that one 1K Of course, as we're talked about, it should be 100k in parallel with 1K but you know that's going to be pretty close to 1K so we can put that in there. but I think we'll find if we do that and we will do it in a second that, uh, it will change this value, this value will actually change now.

let's actually get a average of that shall We There We go. it's pretty close. Let's call it 300 spot on to 300 average. So what we'll do now is we'll just put a bias resistor in here and we'll find that it'll likely change um I'd be very surprised if it doesn't but it's not going to null out to zero cuz we still got the V offset plus other issues and then the supply voltage as well.

And this is a railto rail opamp as well. And as I mentioned and all that sort of stuff, uh, sort of comes into effect. and we haven't even mentioned Tada the elephant in the Room which is the input offset current iOS Look at this, a typical value of 50. The Pico amps.

look at this so it's actually higher than the input bias current. So even if both inputs are matched precisely. So even if we had Vos equal to zero which it's not. but let's assuming it was and both these input bias currents were identical and we put our 1K bias resistor in here, this iOS is still going to screw us up because we have an uncertainty there between the two inputs IB plus and IB minus of 50 P amp.

So you just don't know, you don't know what this Chip is going to do until you actually build it up and test it. All right? So I've now solded a 1K resistor in there Bias Resistor into the uh, non-inverting input there, so we should be able to compensate for that that claimed input. Uh, typical input bias current. So let's switch the supply in here.

5 Vols And look, it has actually dropped. There you go. it's dropped from 300. it's all over the shop.

There's a bit of noise on there. Let's get an average of that. There you go. Let's say it's dropped to 180.

Oh, I Think it's it's changing because it's probably still something's still a bit warm there. perhaps from my, uh, soldering eyon. You got to be careful of that. That can be a trap for young players when you're measuring critical stuff like this and you've just solded it.

Oh, the chip could be hot. The components could be hot. Whatever. Anyway, it's dropped from 300 down to 200.

so that is a change which I would have been urised if we didn't get any. There we go, it's 240, so it's not quite. You know it's changed by 70 micro, but it's still as you can see. Even with the input bias resistor in there properly designed, it's still not happening.

So how can we solve this? Well, of course we could put a pot in here and actually tweak that out and null it out, but that's not actually a Uh solution for the final circuit I want? So what we're going to have to do here is drop these values because as I said, all that bias current IB looks like it probably dominates. Um I I Basically, what I want to do is reduce. Uh, it. fix this circuit so that the input bias current really doesn't have an effect.

it's dominated by the V term there. So um, let's change. Let's lower these by an order of magnitude and we should see. Let's leave this 1K in.

Well, no, actually, I'll take out the 1K right? I'll take out the 1K I'll short it back down the ground. Let's lower these by an order of magnitude. So I put 100 ohms and 10K in there and we should find that 300 we got before drops down to. You know, in theory it should drop down to 30 right if Vos doesn't come into it.

But it could drop down to 100 or something like that. But we should find that we'll get better than what we're getting here now. and there you have it. Uh, it dropped.

but uh, not by a huge amount. You know we're only talking H What average? you know. 240. It's dropped like 60 uh, microvolts or something like that.

Not a huge amount. When we changed, dropped these by an order of magnitude. But then again, we don't have that uh bias resistor in there. So let me add 100 ohms in there.

and bingo There you go with 100 ohm bias resistor in here. That's look we're getting. It's a bit noisy I got a huge resistor on there. It's probably picking up something, but uh, let's get an average on that.

There we go, it's dropped down to an average of like 50 or 60 microvolt. So we've effectively, um, you know, really gotten down to just what we expect with the V offset voltage there. cuz I on a typical chip like this I expect it to be slightly under 100. Uh, one microvolt.

So like, which would, um, be this digit here would be one. So if we had one there, it would equate to a V offset of one microvolt because of the gain of 100. So the fact that we're getting around about 7 if in theory, we've nulled out both those input bias Curren and we're not sure if we have I Mean that's the thing with this, right? We don't actually know what we've Uh, well, we know we've achieved something right. We've trimmed this thing down so it's better, but we don't actually know what the exact culprit was cuz this is a railto rail.

Um, opamp here, so we don't exactly know the topology. Add on that that this is a chopper amp as well. so there's going to be input switching Uh current and all sorts of stuff to do with that. And this isn't just your regular Op amp by the way, it's an input.

It's I've done a whole video on Chopper amps which I'll link in down below as I did link on on the previous video. So you know we've tweaked it by lowering these values and adding in our proper Bias Resistor And you know we can probably say now that Vos dominates. but does it? Uh, we don't know because we've only got one sample of this chip. I'd have to do like, you know, 10 samp to get sort of meaningful, uh, data to see what the exact cut is.

We don't know, uh, whether the input bias is flowing in or out of either of those pins. You know it's just crazy and well, let's see what happens if we change the supply rail. Here we go. Oh sorry.

I Got to switch off my average mode. It's not a rolling average, it's a average over time there. So let's start back up at our maximum Supply rail of 5 volts. Let's say we're getting 100.

There we go. That's different to what. Anyway, it's jumping around a bit. It's a bit noisy, but we drop in and look, look at that folks, it's going negative.

Now there you go, which is okay, it's to be expected. There you go. Now we're down at its minimum Supply rail of 2.7 Vols and we're getting basically a negative. Uh, we've gone all away from positive up to oh, hold, hang on Bloody multimeter.

There we go. Average: We've gone all the way from positive to negative there because this is a railto rail up amp. As I said, the topology we've got allows currents to flow either way and in combination with V offset which also could change over change over the supply rail range. So our common mode range is different and our Um bypass decoupling as well which we saw made a in the previous video.

All these things combined make this a little tricky circuit to actually null this thing out completely over the Supply Range but I'm pretty happy with that by adding the bias resistor in there. we've done a pretty good job. I mean I Don't care if it fluctuates plus minus like that over the supply voltage range, you kind of expect it to do that really, as long as it's within your acceptable margin. I Mean before, when we're getting 300 microvolts in some chip celles measuring, we're getting 500 microvolts that clearly wasn't acceptable for my purposes that I wanted for so.

but this sort of range, you know, plusus 100 microvolts or 1 microvolt essential V offset combined, then hey, that's just fine. And then if we install a 500 ohm trim pot as the bias resistor here, you'll notice that uh, we can actually tweak it. Let's uh, get our average there watch. see it's stable now I Can uh, tweak this thing? Look at that.

Not a problem at all, so you can actually tweak your input bias currents. You can null them out, but then you've got to do that at a particular Uh Supply voltage of course and common modor range. But of course, if you start changing your voltage rail here, then you're going to find that that set in is W look no good anymore. So yeah, it's only going to be valid that particular bias current for a particular Supply and common mode input range.

Oh, and by the way, if uh, your bias resistor here is actually uh, too large in value, then uh, noise can become a problem thermal noise, so you may want to actually bypass that with a cap. And for those curious about the Uh bypass caps, I've removed both of these and I've put A47 mik directly across the rails like that. Once again, we're back to our 100 ohm input resistor there. and there's our offset Uh voltage or our equivalent offset voltage with our 5vt Rail.

And of course, if we take that down, it's just going to drop again and again and again and we'll probably find that one sucker there. Go a bit negative down at 2.7 and it makes no difference if we leave both those caps in and put A47 across there as well. it's still not going to fix in quote marks the issue. So um, what we've got here is a chopper amp.

Uh, its internal architecture is unknown to us. It uses some pingpong patented architecture that Analog Devices have come up with at Auto zeros and chops and does slices and dices and makes your bread for you and does all sorts of things. So um, yeah, it's going to. We're not going to get a consistent uh, well, Vos and or um, input bias currents that can go in either direction for both the inputs over the supply voltage range here.

So eh, what are you left to do? Not much. You either put up with it over your supply voltage range or you fix it at a particular voltage range, or you choose some other opam. It's up to you. And for those curious, if we're actually able to measure the input bias current on here, well, let's give it a go, shall we not? Uh, particularly easy.

but I've hooked up my Keithly P meter. There it is, and that at 5 vol Supply rail, it's the 1 Nano amp range. So we're talking Pico amps there. So 1620 odd Pico amps into that non-inverting input that's uh, through that 100 ohm uh Bias Resistor and if I that's at 5 volts.

and if I lower the voltage, that's 4 and a half. we'll probably see it change. It's a bit noisy there. This isn't the lowest noise setup I could uh, I'd have to dick around a lot more to get low noise.

that's 3.5 Vols There we go, we're getting almost down to zero, Will it actually go in the opposite direction? 2.7 Yeah it does. So there we go, that's a 2.7 It's actually gone the other way. Tada So there you go I Hope you enjoyed that. Followup fundamental: Friday to the previous video and if you like the concept, give it a big thumbs up.

and if you want to discuss it, you know where to do it. the EV blog Forum or you can leave comments on the blog website or on YouTube Catch you next time.