Hi yes it's yet another video on a Precision current Source I've done a previous video on this a two part one where I designed a Precision 1 amp current Source well as it so happens I also need a Precision 1 milliamp current source and also a Precision 1 microamp current source and the LTC 65 we used in the previous videos which you if you haven't seen them you should they'll be linked in down below here and that chip. ah you know we had a few issues with the stability of that thing. gave me the Heie jebi So I wanted to try another circuit which is been around for a long time and it's been touted for that time as you one of the world's best uh Precision current sources and it uses the Ref 102 voltage reference from Uh Bur Brown the old Bur Brown which are now owned by TI of course and there's a whole application note which once again I'll link in down below. Check it out, it's um how to to use this Ref 102 uh 10vt voltage reference.

pretty good voltage reference which has been around for donkey years. uh and you can uh hook an Op amp onto it and a Precision resistor Bingo you've got yourself a low current uh IE sub uh 10 milliamp uh Precision current Source great So where we only want one milliamp here and also uh one microamp so not a problem I thought I'd give this a bow with the Ref 102 and the recommended uh Opa 27 op amp here and you've got your load uh standard configuration we which we looked at in a uh previous video. So I've buil it up here on a breadboard and we'll see how it performs. You'll notice no bypass caps at all here.

no load capacitors at all, not really required. They do recommend an input bypass cap but not you know really require. The ultimate application here is going to be powered from a couple of Uh 9vt batteries and of course the aim of this circuit is instead of you know the ground pin here being grounded and you get your Precision 10 volts out of your voltage reference here. Uh what it does is it lifts that ground pin there with the Op amp here and it it still.

This Op Amp does whatever Well, this Op Amp does whatever is required to the ground pin here to keep the voltage reference across here 10 volts. and of course it's wired as a voltage buffer. so the voltage here equals the voltage here minus the offset voltage which we'll uh talk about uh shortly. Then you get a that Precision 10vt voltage reference across your resistor here which can be a nice Precision resistor which we're going to use here today and using Ms law, you can calculate your load current going down to ground here so that's all it is.

Neat little trick raising the ground pin of your voltage reference. let's see if it works. Hey we had issues last time with a application Note data sheet. Let's see if this one one does the business I Think it will because they've done a whole separate application Noe on It has been around for a long time.

fairly confident this thing's going to work at the low currents. and once again, thanks to Uh Uel at the Uh PR V Precision group who uh supplied me samples of these. Very nice in this case. uh Vhp 100 uh reference resistors here.

Um 10K at .01% they're actually this series Vph1 100 actually capable of better than that. I'll link in the data sheet down below. Awesome little resistors. and here they are in a very nice uh, hermetically welded sealed Uh can.

Like this: they're a bulk metal foil uh resistor near almost zero Uh. temperature coefficient. You don't really have to worry about it. these are 0.01% but they are available in Um 05% values to order.

so these are really nice. Uh resistors? Basically no with the Zed foil uh technology. Basically no uh, inductance and no capacitance rise time of like 1 n. Brilliant devices for um, you know, high performance pulse applications things like that.

So, but fantastic for use as a Uh Precision reference resistor like we're going to use here. Oops. Did I mention that these were Zed foil technology? No, they're not. That was the resistors in the previous video.

these ones are the bulk metal foil technology. They have various Technologies available to give different Uh precision and uh, Co and performance and all sorts of stuff. So these are slightly different to the ones we saw previously. And the good thing is you can order them in any value you want.

any resistance value, just specify it doesn't cost anymore. Awesome! And here's our little circuit, the Ref 102 over here and the OPA 27, Which of course you need an OP because this is a 10 volt voltage reference. We power it from like something greater than uh, 12 volts or there. so you need a high voltage Op amp to go along with that.

And the um, uh, open PA 27 does that? Only a 20 uh, microvolt or 10 microvolt offset voltage or something. It's incredibly small, so there you go. Don't need any uh bypass caps on this. I Think we'll get away without it.

and uh, basically we've got our input over here. This is our voltage input. here's our ground, and here's our output here. So I've got the load connected uh, directly across the output to ground.

and in this case, the load is my agilant uh, current up here. So we'll just, you know. So basically the load is just the uh, current shunt resistor inside here. You know, for all practical purposes, it's basically a short circuit.

Power the thing down here. Um, set a current limit. you know, 20 milliamps, something like that, just so nothing blows up. and uh, 13 volts? I Don't know.

13 sounds like a lucky number to me. so we'll use 13. Vols We just have to be above. uh.

10 volts. I'm not sure what the exact value is I think it's a volt or two above that we need to operate. 13's a treat. Now it's pop quiz time.

Just like in the previous video. I'm actually pretty confident this circuit as I've built and hooked up is not going to work. It's not going to give our precise, you know, 10 volts across 10K resistance. It's not going to give ourselves a precise 1 milliamp through our load here to the multimeter.

and you know, if you want to try and figure it out on your own, stop the video now. I Have actually mentioned the reason for it in a previous video. Just think about the circuit and how having a very low value load on here could affect its performance. So there you go.

If you want to go figure it out, please do Otherwise, let's power it on. and uh, see what we get here. We go channel two on and Bingo, look at that. There you go.

It is not. It's not. It's not. It's fairly close, you know.

1.03 Milli But hey, we expect much, much better than that. So what's going on? H Now, if we actually measure the voltage directly across the resistor there, we expect 10 volts out of our I've got the leads backwards. All the electrons will fall out. No, that's all right, But look there it is.

a 10.35 7 basically corresponds with the meter up there, so it's not precisely 10v. So why aren't we getting our precise uh, you know, 10 volts out of our voltage reference H Oh, by the way. Uh, for this video I Couldn't get the super duper accurate Uh C version of The Ref 102 voltage reference. It comes in different grades as most voltage references do if you check out the data sheet.

if you're going to order a voltage reference, be very careful which one you actually order. In this case. I've got the A version, which is only .1% nominal accuracy, so you know it's not as good as it can be. The C version is 0.025% %, but in this case, our error is a whopping 3.5% So it's certainly not the Ref 102 that I'm using here.

It's certainly not going to be our Precision uh Vie resistor here. Not a chance. it's not going to be our Op Amp down there. Well, is it H And if you did, stop the video on trying to figure it out and I hope you did.

Um, the reason is very simple: this: Opa 277. Uh. Opamp. Because of the very low load on here.

effectively a short circuit shorting it down to ground, here, it's not able to. There's going to be a minimum output voltage. It can actually drive on there. So we're going to actually get an error on there due to the Op Amp because it's trying to operate down near its negative rail down here.

And even if as mentioned in the previous video, you use a uh Precision Railto rail up amp, they aren't really rail to rail when you really get down into the you know .1 1% or better uh margin that we're talking at. You know the really low voltages down here. rail to rail might mean you know a 10 molts output voltage or something. Well, you know that might be okay for a normal circuit, not for a Precision current source is going to blow our error uh, budget right out the window.

So what we need to do is put a as we did last time. put a diode in series with this load to lift this voltage. Uh, on the non-inverting input on pin 3 here up by about6 Vols or so. If we have a look at the data sheet for the OPA 277 here, you'll notice our voltage output here.

This is not a railto rail chip by the way. But even if it was as I said, it still wouldn't be good enough. So our minimum output voltage here is going to be our negative rail which is ground plus half a volt so you know it's hopeless. We can't obviously.

so we need to boost the output voltage above that 0.5 volt. so one you know basic silicon diode drop of about 6 Vols should do the business. But let's actually measure it and see what value we get. So I'll measure from uh, the ground pin here to the output voltage because effectively, we're shorting the output here.

So it's you know, driving this thing down to a minimum voltage that it can. So let's have a look. Okay, between pin 4 and six, there we go .46: Vols Close enough to that uh, data sheet minimum value of5 Vols So a one silicon Dio drop should do the business. and once we put that in, um, assuming that, uh, you know it doesn't affect the stability of it I don't think it will.

In this particular case, we should get our precise 1 volt uh 1 milliamp out of there and our precise 10 volts across our Precision reference resistor. So here we go. let's take our load here and I got a little one in uh, 4148 or something like that. So let's put that in series is with our ground and Taada.

Look at that I think we're with inside our error margin and uh, if we measure of course our voltage across our Precision 10K resistor, what do we get? Taada, There's our Uh Precision 10 Vols Awesome! So how close are we there? Well, let's get our trusty calculator here. Uh 1 -9999 21 I Think we're going to be really close here folks. Um, times 100 to give us a percentage and we're looking at Tada 8% Beautiful. Well within our error budget.

So speaking of our error budget, let's take a look at where some of our errors accumulate and these aren't even all of them. Uh, really, you know, so you I don't think I've actually covered them all, which is unbelievable anyway. Um, when you're doing these sorts of precision circuits, all this sort of stuff matters. It really does.

And you can do worst case scenarios and all sorts of things and you know, ah, you know, worst case error budget and in this case it should have been actually pretty high cuz we if we take a look at our Ref 102 voltage reference it the dominant figure here. I've put them all in uh, parts per million by the way. PPM But you can easily convert between Uh PPM and uh, percentage. and if you want to do that, uh 1 PPM equals .01% Basically, that's pretty much what we're looking out here.

but anyway. uh, the people. the dominant figure as you can see here like some of them are going to like, be drifts with temperature and time and stuff like that. But you know, if you just start with your ballpark, you know basic, you know top level Banner speec of the Ref 102 for the basic initial accuracy we're we're looking at you know, 250 parts per million or uh, 0.025% for that CGR chip.

But we don't even have the C grade chip. We've got the a crappy A grade chip that only cost a dollar or something like that. and uh, that is 1,000 PPM So that's going to dominate here. 1,000 PPM So what we actually measured there just before was a value of um 80 PPM uh, you know, different from our Noral expected value assuming of course that our Ulent meter is absolutely, uh, precise and bang on.

But uh yeah, so that's 8% is what we actually measured. You know, so it's well within. So the Ref 102 we're actually using is 1,000 PPM just there on its data sheet accuracy. so we're already bsing it in right there.

Then you've got other stuff like the drift uh, with that temperature is uh, maximum worst case of 2.5 PPM per uh de C So you got to take that into account over uh temperature range. and then you've got uh, load regulation or the the change in the nominal output uh value, uh over the current. so it's a 10. another 10 PPM per milliamp error right there and these sort of Errors can accumulate and interact in various different ways which we won't go into.

but uh, Ref 10? Uh So then then we've got our line regulation I.E Our input voltage here. How does our our 10vt Precision 10vt reference here change with our line or input voltage? Well, one PPM per volt. So if we change that from uh uh, 13 volts that we powered it to 14 volts, we'd expect a 1 PPM change. Hey, let's try that Here we go.

Here's our value: I'm on 13 volts I'm going to up it to 14 Vols There we go. not changing anything measurable there? No, no, it's pretty good. Anyway, that uh is your worst case uh value there. So you qu your know if you're really doing this for serious business, then well, you got to take that into account.

it's all part of your era budget. Then we've got our aging of our reference as well. Well, there's another 20 PPM per ,000 hours of operation right there. Then we have a look at our our opamp really isn't uh contributing much here at all.

the OPA 27 uh uh really the only thing we got to worry about is the Uh offset voltage and the offset voltage of this I think worst case is only 10 microv volts which is equivalent to 1 PPM over our 10vt voltage reference. If you know we were only using a 1vt voltage reference, eh, it' be 10 PPM It'll be a larger percentage of that. But of course if you want to illuminate that, you could use a uh Chopper uh amplifier in there of course instead of uh one of these um, low offset Precision uh references. you can even get much lower than that if you really want to uh you know Guild the Lily and get rid of it.

And then we've got our Uh V resistor here. Well, its basic accuracy is a 100 PPM or a 0.01 uh, 0.01% there. And then you've got the stability of that over temperature and time of and stuff of 2 PPM And then you've got other stuff like overload as well if there's any self-heating But in this case, you know we're not driving any significant current through this and there. no any significant uh Power dissipation in the thing.

So really, we don't care about that. But as you can see, you know all these things like 1,000 PPM and all the other stuff that we could potentially add in here. we're balls in it. In, we're way better.

We measured about 80 PPM Brilliant. But hey, maybe we're just lucky. Eh, you can't expect this to happen all the time. Okay, so we got ourselves a fantastic Precision 1 milliamp current Source But I mentioned I Also want a 1 microamp one? Well, can we just up this resistor here to 10 Meg 10 volts across 10 Meg Going to give us our 1 micro amp? Well, in theory Yes, but in practice No.

And if you want to stop the video again, here's another: Pop Quiz So you might have precisely 10 volts across your precise 10 Meg resistor here, giving you your precise 1 microamp through that resistor. But that doesn't mean one microamp is going to flow down here into your load like this. Why? Because inputs of two Op amps have input bias currents, so there's going to be some of it tiny little smidgen. You know, a couple of bees Dicks is going to fly into the non-inverting input of that Opamp.

Let's go to the data sheet and find out how much. Here's our data sheet again and this is what we need to know the input bias current here for our Uh Opa 277. And by the way, just be careful. this is a column for the 277.

You can also get Jewel and uh. quad versions of this and the different packages is typically the Uh. especially. the Quad versions can have bigger input bias currents than the single or the Dual versions.

That's why they have different columns here, so just be careful with that. So let's go back down here. Where is it? Input bias current? Here we go: IB Oh, look at this plusus 0.5 watt nanoamps. So let's take that as a nominal.

You know? worst case: Uh, one, uh, one Nano amp there and a nanoamp? Doesn't sound like much, but when you're around with Precision current sources like this, well get your calculator out. Figure what uh .5 nanoamps is as an error on one on a 1 microamp. uh, load current There Answer: not 0.05% So worst case, it could be like .1% So you know we like. My goal for this thing is to be much better than 05% there.

So really I Have absolutely no confidence in this in using the OPA 27 opamp. As good as it was for 1 milliamp. No confidence really. Not much at uh, getting this thing to work at one microamp.

but let's try it. But unfortunately I don't have a Precision uh, 10m resistor yet. Uh, still on order. so I've just got a crappy one.

but I have actually measured its value and it's 9912 Meg or thereabout. so you know if it was precise, we' expect about uh, uh, 991 nanoamps out of this thing or 991 microamps. There we go, getting confused with all these decimal points and what do we get? Aha, look at that 1.27 or so microamps. But that's rather curious because we actually expect it so that's above our nominal figure.

But if you have a look at the um uh circuit down here, we actually expect it to be less because it's going to suck away some of the current into here. So we actually expect our current into low, which we're measuring to be under speec, not over speec. So we'll go back to our voltage reference here and measure it. and Tada we're looking at uh, 10.14 volts there.

Uhhuh. So that's exactly the same issue we were getting last time without the diode. So now our our circuit parameters have changed and it looks like where our Uh voltage reference isn't working as well. So we're going to put two diodes in series now.

and what do we get? Well, it's You know, it's a bit lower than what it was before, but if we actually measure the reference voltage now I think we'll find Bingo But of course that's not precisely 10 volts. We'll reading uh, 99.998 or something before. So I've added in a third diode in series now and Bingo hey! I think we've solved it now. The reason for that, of course, is because your voltage drop on a diode is not um, constant.

It's going to vary with the characteristic curve of the diode. Of course, look up any data sheet for the diode and it depends on the current flowing through the diode. That's why we need at least three in series. Once we're down at one microamp, there's you know, bugger, all voltage drop across each of those diodes.

Now, in fact, we can measure that standard 1 N4 4148 diet. Of course, you know any any beginner would expect 0.6v voltage drop across that. What do you get at one micro amp? Let's measure it. 26.

There you go. You think it was some kick ass shocky, but it's not. It's just a Jo Blogs 1 N41 48 Silicon Diode. But hey, it's right down on the lower part of that characteristic curve for the diode.

so the voltage drop is quite small. but we were reading slightly under on our voltage reference here 99.998 if you remember. Uh, precisely put those into the calculator and Tada 1.8 micr. That was pretty darn close to what we were measuring before, as you'd expect, because this is just owns law.

It's got to work as long as you take into account your Um Op amp, as long as you're ensuring that that opamp um has enough uh output, uh, margin there on on the lowside output drive to drive your voltage reference. Oh, and for those curious about the noise, no on, uh, the 1 milliamp one. No, there's bugger, all noise. That's just the noise of the test setup.

Pretty much the good thing about this voltage reference. It does actually have a noise reduction uh pin as well, so you can put a bypass cap on there if you really want to get the noise down. And if we have a look at the 1 microamp Uh current Source output. check this out.

look 50 m per division. That is awful. We got some real nasty crap going on in there, but is it the actual reference itself being unstable? or is it external noise? Well, it's almost I think. Almost certainly external noise because we're talking about a one microamp current Source here.

And of course we've got these huge uh, you know, antenna leads. just you know, hanging off here going into our Uh current, uh, shunt over there. So uh yeah, I expect that and we can probably verify that. Let's actually have a look here.

Let's uh, cut, let's try and couple in some noise into this and hey, look at that. We're coupling in Taada 50 HZ So it looks like we're just picking up all sorts of crap with that big antenna uh wire coming off the uh, load there to our shunt and basically look at this. uh, you don't normally see big spikes like that and aha, let's freeze that and let's have a look at the frequency there. We could see it before, but we're talking 1, 2, 3, 4, five divisions between those big spikes there at 2 milliseconds per division Bingo 10 milliseconds.

What's that? 100 Hertz Fullwave rectified Mains H Coincidence I Think not. No. So what we're doing here is, we're getting crap picked up from external in the room and watch this right? I Don't have my lead lights on. my big lead studio lights on.

Let me switch them on Taada. look at that there some higher frequency I Think if a memory serves me correctly, it was like 64 khz switching or something like that. but more crap just picked up there. So really, what we're talking about here is just, uh, external.

uh, noise picked up from the environment here with these big antenna leads. So if we actually disconnect this antenna here, this horrible looking antenna and we connect our lead over there, so our load is basically shorted out but it's kept on the breadboard now. Bingo Look at that. No more noise I Mean our one micro amp is still there.

If we go down to 5 MTS per division. There we go. we're getting exactly the same noise where we were before and like that's just basically the inherent system noise that's not the voltage reference itself and we're getting some 50 HZ on there as well. So there's absolutely nothing wrong with that circuit and with that one microamp current generator at all.

it is just the noise to do with the system around it. and well, that's perfectly normal and we won't go into Jez. I could do a whole hours video on how you know best to, uh, lower the noise and you know, eliminate it from your system and while you're testing and all sorts of stuff like that. So yeah, I won't go into it.

So there's effectively no problem with this uh current Source down at 1 micro amp either except for the opamp input bias error. So I'm probably going to have to choose another opamp from the OPA 277 because I think we're just going to get a bit too much uh input, uh, bias current there, which introduces an error in the load there. but of could of course you could, uh, you know, tweak the the output voltage of this thing to compensate for that input bias current if you really wanted to. but you know I'll just probably pick another opamp for there.

but the OPA 277 is one specifically uh chosen for this task. Apparently it is a, you know, a pretty Ultra stable sort of opamp so suits this sort of application for raising the ground up quite well. So there you go I Hope you enjoyed that one. Yeah, this will probably my last video on these uh Precision current sources unless uh, something, uh, you know, interesting that comes up.

But anyway, I hope you enjoyed that short little miniseries on these Precision current sources and there's a lot more as I've explained and shown a lot more that can go into this with actually, uh, measuring it and characterizing its performance. and in the case of the one, uh, mic ramp one actually getting you know, a noise-free test environment and stuff like that anyway, uh, might be for another day. If you like the video, please give it a big thumbs up and if you want to discuss it, jump on over to the Eev blog Forum the link is down below. Catch you next time.

Um.