Hi, this isn't exactly going to be a fundamentals Friday as you probably expected, but I was just uh running ran across uh this issue that I thought I'd uh check out today. so I thought oh, why not turn on the camera and uh, let you have a look at it as well. So I guess it is kind of fundamental. It you know comes down to uh looking at data sheets and measuring things so you know H it's a fundamental uh problem that we all have to deal with eventually.

Anyway, what I've got is this: Analog Devices ad 8628 It's a really scho as it says zero drift uh single Supply Railto Rail operational amplifier and it's actually a Uh Chopper amplifier as I've done a video on. If you haven't seen a video on Chopper amps I will link that right here down below. so click on that and watch that if you want to. It's one of my uh, uh, much older videos.

Anyway, pretty much one of the best Uh, zero drift uh, low offset voltage operational amplifiers you can get on the market. And yes, it's got that Chopper configuration to get a very low asset voltage. Check out some of the banner specs here. It claims to be the lowest Auto zero uh amplifier noise of presumably any amplifier on the market.

Awesome Super low offset voltage. Only one microv volts that will be a typical Uh voltage as we'll see not a maximum voltage and uh, input offset drift. it's essentially zero. Look at it.

It's 22 microvolts per degree C or 2 nanov volts per degree C So once you once it's got that uh, one micro, let's say this particular chip, the one you've got has 1 microvolt uh offset input, offset voltage, then that value of one microvolts is only going to drift by 2 Nan volts per degre. C Incredibly stable chip. Fantastic. That's one of the reasons I'm using.

Isn't it here now? Uh, it's got railto rail input and output swing. Fantastic! So you can take the input down to Zer volts or you can take it all the way up to the supply voltage and same with the output as well so you can utilize the maximum output range. uh, single Supply operation actually goes down to Uh 22 uh volts or there about. So uh, perfect for my application.

Um High Gain yeah, kind 130 uh DB common mode rejection ratio all that sort of stuff. low input bias current 100 pamps Good enough for my application. Uh, low power supply current, it's under a milliamp. Good enough overload recovery time.

It's not bad for a chopper amp. Um, no external components. it's all internal. Um, all the uh Chopper um uh.

circuitry and stuff all runs internally and the oscillator is all internal and all that sort of stuff. as is typical with chopper amps these days, and it's qualified for automotive applications. Whoop-de-doo But anyway, I built this. Um I used this chip and I discovered a little weird uh problem.

Well, an issue with it that I wasn't really expecting and uh, it. You know it took me a little bit a while to figure out what was actually going on with this thing I actually mentioned on the Amp hour a couple of Uh episodes ago actually. uh, if you want to listen to that, there's a brief description of it on there. And anyway, I thought I would I' I' figured out what the issue is.

but I thought I would double check it by actually building it up uh, from scratch again and actually verifying that it's the case in various Uh scenarios. So that's exactly what I'm going to do today. So I thought I'd turn on the camera and uh, give you a look at it I was slightly off on the supply voltage there. it's actually 2.7 Vol to 5 Vol single Supply operation.

Now, as with um, all off lamps, they don't care whether or not you're using them from a dual Supply or a split Supply subject to input and output. Uh, you know, common mode range and uh, output range and stuff like that. So it can actually operate from Plus Plus - 1.35 Vols That's like, practically, you know. Plusus a single cell.

Um, single alkaline cell to Plusus 2.5 Vol dual Supply So an ideal Um Opamp to be operating from a coin cell or something like that. You know if you got a Um string gauge or something that's operating from Battery Fantastic device. And apparently the chip combines the benefits previous only found in expensive or Auto zeroing or Chopper stabilized amplifiers. And yes, there can be Uh differences there which I won't uh go into.

As you can see, the noise is pretty low too. uh 0.5 microvolts Peak to Peak uh from zero HZ up to Uh 10 Herz So awesome device! It's pretty hard to uh I think you'll be pretty hard pressed to uh get a better device than this. There are lower ones in terms of offset voltage, but not um, sort of like a combined in terms of offset voltage plus noise plus drift. So you know, one of the best Op amps on the market.

Now before we have a look at the Uh problem, I'll just look at some of the specs to uh, give you some background on the issue here now. Uh, these are the electrical characteristics at a supply voltage of 5 volts I.E plusus 2.5 Vols which was pretty close to what I was working at I was actually working off um, three alkaline fresh alkaline cells so it was like a 4 and 1/2 volt. Supply Now, uh, the input offset voltage is the thing that we're concerned with today. This is here are the typical figures for it.

Okay, typical figure at 25 C by the way with the common mode voltage by the way smack in the middle there which was how I was basically operating with a split Supply Now the offset voltage here, the input offset voltage. typically one microvolt and that's the banner spec. But of course you don't take the you know you can't always just take the banner spec. You've got to look at the maximum value as well.

in this case, 5 microv volts maximum. And and if you want to take the full temperature range into account which I don't have to, then uh, it can be as bad as 10 microvolts. That's still pretty low, but you know it's order magnitude worse than the headline spec they tell you on the front page. That's always a tri a um trap for young players with these data sheets.

If you just read these front pages here, these headline specs up here. they don't give you all the detail, they're going to lie, it's basically marketing. Uh, wank up here. Well, they're not.

Well, they're not going to lie, but they're not going to tell you the whole truth and nothing but the truth. This is where you have to look in the electrical specifications here. So this is what we're really concerned with. input bias.

current doesn't really, uh, concern me, um, really and or the rejection ratio and all that sort of stuff doesn't really matter. And of course the uh drift is so low you know, two nanov volts per degree C that gez you don't even have to worry about it, but not that it um, would come into play anyway, even if it was high. Because the lab here is a essentially a fixed temperature, it's only going to vary with the airon on by. you know, plus - half a degree, uh, a a degree or something like that.

Now this looks like an identical page, but it's not. It's actually the next page over the electrical characteristics. Now for a supply voltage of 2.7 Vols We had 5 volts before and once again, common mode Uh voltage 1.35 volts smack in the middle and we basically got exactly the same numbers here. one microvolt typical for the input offset voltage, which is, um, the issue that I had.

Now if you take it, if you scroll further through the data sheet, you come to well, the first thing you come to because they're quite proud of it is the Um input offset voltage here on the X-axis Here we go, it sort of zero is sort of there right about there. and oh, this is at 2.7 volts. Let's look at the 5vt one. there's our zero right right in the center there and our input offset voltage can be plus minus that with with a common mode uh, voltage of 2.5 Vol So smack in the middle and you get that characteristic bell curve kind of shape in your values and these are that number of amplifiers.

So they actually measured you know, 10 20 like 80 amplifiers. So they measured a couple hundred amplifiers here. and this is the response that they actually uh got. And yet it's always typical it's going to be when you get manufacturing process variations like this.

you're typically going to get given enough devices that bell shaped characteristic response, that bell curve as they call it. So uh, let's you know here, most of them are centered around zero and the most of them, the bulk are going to be within plusus 0.5 microvolt. So it's you know. Um, round about that one microvolt.

Typical figure. Not bad. See, here's the majority of them. Oh, there's there's one one microvolt from Plusus one I Guess they only give one microvolt.

Typical. They don't say that it's plus minus anyway, even at 2.7 Vols down here. So when you vary the supply voltage which is what we're going to look at later, then you'll notice that yeah, it's been shifted up a bit. There's our zero point, but still, it's very low.

None of them are getting anywhere near that maximum figure of five microvolts, right? So we expect if we bought some of these chips. you know, odds are not guaranteed, of course, but you can be pretty darn certain it's going to fall within there. You know it's going to be like, ah, at least under two microvolts, for example. You know there's a couple of outliers out here, for example, but gee, you know you expect it to be pretty close to that typical value of 1 microvolts and I Found when I built this thing up, that wasn't the case I was getting up to 5 microvolts offset.

H Let's go to the breadboard. but before we do that, let's take a look at the circuit which we're looking at here. it's the Ad 8628 Opamp. It's a single opamp s so8 uh, package and I've got it set for a non-inverting gain of 100 here.

Or it's actually 101 using 1K and 100K Who cares? Near enough to 100, right? So um, then we're going to have an input offset voltage here. That one microvolt is typical. And of course, that input offset voltage there is going to get multiplied by the gain of your opamp. That's how it works.

It's not just for this opamp, it's for all opamps. It's going to do that. So let's say that this particular device that we have under test is has one microvolt there. We expect 100 microvolts on the output here or .1 molts.

So we'll build it up and we see and we'll see if we get it of course. I've got just got a couple of bypass caps on here I've got a split Supply In this case, it can go anywhere from uh, well, from 2.7 total, um up to 5. Vols Now what I've done is gone. and actually uh, assem this chip on uh, one of my um microcurrent boards here because it has the same Uh configuration just allows me to easily do that.

I removed all other Uh circuitry on there, the rail splitter and all that sort of stuff so it's powered from my external bench. Supply uh, positive, Supply negative Supply and my uh, and there's the ground and uh, that's my output now. I've got my output uh voltage on here and I've got my Uh Supply rail here so it's 5 Vols total across here. so split is plus - 2 1/2 Vol So you know the Uh maximum range that this thing can operate at.

and if you remember, look at our spec here: Typical figure of one microvolt offset voltage at Um and a maximum of five, so we'd expect that typical value. Remember based on that bell curve. So I've got a times uh, well, times 101 gain in here, near enough to 100 and look, what we're getting on the output, we're getting getting 300 odd microvolts or they're about3 molts. If you divide that by 100, that's three microv volts, offset voltage and sure, okay, it's within spec.

You know you think? Okay, not a problem, it's within spec. Okay, fair enough. nothing wrong with the chip. Okay, and that's what I thought the first time.

Oh look, I've just got a bad one, you know? but no I replaced it and I replaced it again I replaced it again and no. All of them had around about this same 300 odd microv volts offset voltage and I was scratching my head for ages trying to figure out what it actually was. And um, yes, I've deliberately shown um this ground Point uh down here as a star point so that everything is referenced to that. the input voltage there is referenced directly to there.

the output voltage is reference to there. and that's really going to matter if you've got any significant current flowing through those tracers. but we don't with our high impedance. uh, you know, 10 me input multimeter and stuff like that.

So the star grounding doesn't really matter in this particular scenario. But I've just drawn it there for uh, completeness. and uh, yes, that input voltage pin three. That's that little blue wire looping across there that's shorted out directly to there.

So um, you know our input. Uh, so we're amplifying our V our input offset voltage by 100 basically. and that's what we're getting regardless of what chip I put in here. and trust me, I've tested quite a few of them.

So now I've got it down to just this scenario here with the split. Supply So what's going on is the typical figure? a load of? Well, I Tried all sorts of things, hacking and slashing my circuit. you know? Uh oh. I've got something.

There's something weird going on the layout. There's something wrong with my power supply. yada yada I was going pouring through the data sheets and everything until I Just thought h I Wonder what happens if I take the supply voltage down and here we go. Let me wind down the wick right and you're remember back here.

Here's the 5V one. You remember there it is: 5 volts typical Figure 1 microvolt, 5 microv volts, 2 points the minimum 2.7 volts. Typical figures remain the same. so you expect you don't expect that offset voltage to change at all.

right? Well, let's look what happens when we wind this power supply down. Look at that. It's dropping. It's dropping.

It's dropping. Look, magic folks. Look what happens when we get down to 2.7 volts. It's bordering on.

You know it's zero. It's like, you know it's very low. In fact, it actually went negative, right? and you'll see it. We wind it up a bit.

it goes positive again, so we can actually almost null that out. You know there it is at. You know, 3 volts and that's going to vary with chip. Why is this thing changing with the supply voltage? According to the spec sheet, it shouldn't.

The typical figure should be the same and it's not just this one chip. I've tried this on many different I many different devices and it's exactly the same now. the thing I actually wanted uh to try today because I've already uh tried this in my uh, real circuit and um, yeah, I' I've verified that this is the case. So what? I actually wanted to try today before I Turn the camera on is does this do it based on a single Supply cuz I've got a split Supply at the moment and effectively this uh Point Here this ground point, the reference point is sitting smack in the middle of that Supply voltage range, right? So what I want to do is power this from a a single Supply So effectively I'm just going to now short out this ground point to this negative point and only Power it from a single Supply So I got to uh I I do this live and I want to blow up my chip if I goof it up? So let me reconfigure it all right.

I've reconfigured it now. So essentially I've just shorted out this point that uh, ground point to the negative Rail and I'm now, uh, running it from a single Supply and uh, let's and here it is 2.7 uh volts, uh, minimum Supply and we're getting about 73 microv volts out or about divided by 100.7 microvolts offset voltage around that typical figure we expect. Now let's wind it up and I haven't tried this. Let's see if it increases it, it is.

It's going up. It's going up, not by a huge amount though. Look at that so we can go up to a maximum Supply voltage of 5 Vols So there you go. It's the fact that don't don't want to go over that.

That is its maximum. Supply Its absolute maximum is six, but I won't take it there. There you go. it hasn't gone up much at all.

so it's the fact that we're actually running this split. Supply uh is the thing that causes the issue here when you actually uh Power it from just a single Supply you don't get this offset issue at all. Strange I don't know why that's the case. And just to show you that it's not that chip I've actually solded another one onto uh, one of my microcurrent boards.

so this doesn't have the maximum chip in it anymore. It's got the uh Analog Devices chip in it. and there we go. There we go.

Let's take it up to 5 volts. There we go. This one's actually real, even higher. This one's uh, 400, uh, 400 microvolts or four microvolts offset voltage.

and once again, it varies with that Supply rail. Check it out down to 2.7 There it is. where it goes down to incredibly low. Unbelievable.

And just to show that, uh, it's different with the maximum chip, I've got uh, one of my original microcurrent boards with the max 4239 in there and uh, it's um, let's take it up and look even it up at 5 Vols There you go. No, it's incredibly low cuz this one's typically 0.1 microv volts um offset voltage. so it's actually lower than The Analog Devices one. but it's drif and its noise is higher and its bandwidth is lower and all that sort of stuff.

But yeah, so there you go. Not a problem on the Maxum chip. Now just for another comparison. I've got a uh, another uh, low drift uh Auto zero Chopper amp Here it's the microchip mCP 6V 01.

It's not quite as good, but it's uh of either of the devices. but it's not bad at all. it's typical or it's max value is plus - 2 microvolts uh, offset voltage. and once again, they've got a percentage of occurrences versus uh, microvolts offset voltage there.

and as you can see, you know they pretty much, um, the sample Falls within that plusus one microvolt. So we should expect less than 100 microvolts on the output. and there it is. That's exactly what we're getting.

Um, if I switch off my supply, of course it just goes BL And there we go. I switch my Supply on. so we're It's jumping around a bit there, but it's close to zero. We're just getting some noise there.

let's wi. actually this one goes down to 1.8 Vols Actually, let's see if this one varies with Supply There we go. Oh well, there. Oh there we go- 350.

So this one does vary. This one is rated from look it's here it is. Plus 1.8 Vols to 5.5 Vols There you go. So uh, that one on the low side there.

but once it gets there you go once. it gets sort of 2.1 volts there, but sort of at at that 2. 7 volts which we were using before for the other ones as a reference. basically zero so well within spec.

Of course we're only talking, you know., 1.2 microv volts or something like that offset voltage. We wind the wick up and no, it's not really increasing so it's doing the same as the max some chip. Pretty much it's offset voltage. pretty.

Oh no. Hang on. no, no there we go. Pretty consistent over the voltage range I Mean you know to get rid of that noise, we could probably like turn on some average in or something like that.

this micr chip ones performing the same as the Maxim one so it looks like The Analog Devices one is something peculiar with that. So I've put my Analog Devices one uh, back in here. and of course, this is, um, not using the split uh bench Supply anymore. It's utilizing the split Supply on my microcurrent here, but it doesn't seem to make any difference whether or not I use the split Supply the Opamp virtual ground split Supply on here or my bench Supply As you'd expect, Uh, really, you wouldn't expect any difference, but there it is.

I Got the average mode on there and down at 2.7 Vols It's you know, you know it's very low. Very respectable. but let's wind the wick up and up and we can turn that off. Actually I'll turn that off.

Boom. There we go. We're getting our 400 microv volts There we go that that average would have taken some time to, uh, creep up there. But look at that.

There you go. So there's something with that split Supply Unbelievable. But even going back to our original one here in Um single Supply rail operation. um, it still does vary.

So from 2.7 we're getting uh, uh, sorry, 0.54 microv volts offset voltage and it does actually does actually go up I Mean you know, it practically doubles there. whoops, practically doubles, um over the operational Supply voltage range. And just for fun, what I've done is I've actually desoldered these .1: Microfarad bypass caps I'm back to a split Supply Uh, plus minus with the Um split ground here and I've replaced that with a single cap between the positive and negative Supply directly across the chip there. And let's have a look at what we get we getting about uh, 200 microv volts or 2 microvolts uh, offset voltage there at 5 volts.

but if we wind that Wick down, it drops significantly faster than what it did before. Look at that. now. now at 4 Vols it's sort of.

you know, at the level we expect and it's gone a bit negative. Look at that isn't that interesting. By the way, that was a 330 nanofarad directly across there before. Now what I've done is just solded the negative one back in there and I'm just using that.

I've got no bypass in the uh positive rail there and uh 2 7 and up we go. look at that, look at that and now it's significantly worse at that 5V rail there Aha and now I've sold it in the other cap once again 100 n So I back to our original configuration: 100n 100n bypass with a split Supply and uh, this is what we're getting this. that's a 2.7 and we can wear look we're back to to our original configuration. basically.

So there it is. There it is. It's got something to do with the bypass caps and presumably, uh, the input I don't know something to do? Well, you know it's probably got something to do with the internal uh operation of the auto zero uh Chopper amp inside this thing. Um, you know I don't know the exact configuration in there, so you know? who knows, right? only.

Um Analog Devices Could uh, tell us that? but uh, uh, there is a difference when we put in different value caps for the bypass in different configurations. so maybe it's got something to do with the input bias currents. Um, to the Opam. But why that matters when they're both effectively tied down to the same star.

Point like this: um I gee, I don't know. So I'd love to give you an exact solution for this one, folks. but uh, unfortunately. I don't That's as far as I've uh gotten.

and really, um, I'm rather perplexed. Maybe it is something, uh, incredibly simple. and it's well, it may turn out to be something embarrassingly simple, but I don't know I just haven't been able to get to the bottom of this uh, stupid thing. and um, it's got something to do with this particular Analog Devices one the others don't actually uh, show this issue at all.

Both the maxim and the microchip part don't show it. it's only this. Analog Devices and it's still an awesome uh chip. By the way, I'm still going to use it.

This is just a really, you know I'm I'm going to put a regulator in there. The power supply is not going to uh, vary anyway, so you know it's not going to be an issue. but I just when I first discovered this I just thought, what the hell's going on there? it's rather interesting. So I thought I'd at least investigate it.

So uh, it's not a huge deal, but it's just interesting. So if anyone from Analog Devices is actually uh, watching and can shed some uh light on this, please do. If you got any, uh, better ideas yourself, leave it in the comments or on the Uh Eev blog forum and maybe it's got something to do with the patented Oh yes, there's the problem. They patented the damn thing.

um, pingpong operation they call it. it's Auto zero and uh, chopping um at the same time in some unique patented configur I Don't know. maybe it's got something to do with that, but it's really, really bizarre. And why those capacitive values and configurations uh, make a difference.

The bypass uh caps I Don't know. in theory, they shouldn't uh, really. maybe it's a you know, something to do with the Um, just the input configuration of this Chip And The Chopping nature and etc. etc.

The topology used in this particular Analog Devices one CU I don't seem to get it with the maximum or the microchip part I Don't think I've ever seen anything like this before. and well, yeah, it could turn out to be embarrassingly simple I Don't know and I'm probably just making a dick out of myself. but anyway, um, that's just some playing around with the Uh input. uh, measuring the input offset voltage.

And by the way, if you're wondering, no, you can't just put your multimeter on there and even if it's a 6 and half digit one and measure your offset voltage on there, it's actually internal generated internal inside the circuitry in there. so you know you have a big blobber solder between those and you know you're still going to get this input offset voltage. It's inherent in the Uh design of all Op amps, even these ultr low um offset uh ones in in this case, you know, like the best on the market is you know, 0.1 microv volts or something like that which is the maximum one I think anyway um typical figure but worst case and yeah so we learn a bit about worst case and bell curves and all sorts of. um Jazz so I hope hope you uh found that interesting? that's just me uh looking at uh a few little issues in input offset voltage so hope you enjoyed that.

And if you like uh well this really wasn't a fundamentals Friday sorry it was just a me around but yeah anyway hopefully you found it interesting and if you do like it, give it a thumbs up. Catch you next time.