Hi. There's a couple of annoying problems in electronics. One of them is slow rising input signals on digital circuitry and other stuff, and the other one is noise on comparative inputs and such like. So we're going to take a look at these 2 problems today and how we can fix them.

There's basically one solution for both. So let's take a look at slow rise in input signals and the problems they can cause with digital circuitry. Now, look at the Dave CAD drawing here: I've got a seven, 4hc, one, six, one 4-bit are synchronous counter jellybean logic I've got it wired up here and I've got the four LEDs on the output and I've just got like a 2 Hertz input square wave and it's counting up in binary just fine. And by the way, don't worry about all these extra pins here.

This is just a complicated chip wood that has a data load input I've just set them all load. It has various enables and stuff like that. So I've just tied all those high Basically, clock in binary out these settings through our input clock. Here you can see it's our 5 volts peak-to-peak so from 0 to 5 volts, it's just got an offset.

Their duty cycle is 50% and the rise time of the positive edge is 30 nanoseconds. And if we go over to our scope here and have a look at our input signal, sure enough, there's our 50% duty cycle. It's just a regular clock all the way in and we can see that our rise time is about 45 nanoseconds. It's near enough It's going to have, you know, some delays you the cables and other stuff, but it's basically of nice sharp rising edge.

Edie, it's working just fine as you'd expect. So what I've done is I slowed it down a little bit to 500 nanoseconds for the rise time. This is a positive edge triggered our clock. You can see the count is still just fine.

We go up to 1 microsecond here near enough and it still seems to be working just fine. We're at 2 microseconds and it all still well. Yep, it all still looks good. There was no zero there.

Do you notice that there was no zero? It did not go to zero. All the LEDs didn't turn off. You see that it's starting to play up. That's at 2 microseconds.

and if I go higher for microseconds. Oops, our first lead there is two stuck on what's going on. This thing has completely failed and you might think look, this is a perfectly fine edge. There's no noise on it.

Aha trap for young players. Let's go to the data sheet now. I've got the data sheet here for the exact National Semiconductor chip that we're using. The Seven, 4hc, one, six, one, the exact same brand.

Different brand. logic families like this will vary slightly due to slightly different process technology and stuff like that, but basically are they're all pretty much identical among manufacturers and it's got all sorts of specs in here for propagation delay and set-up time and all sorts of stuff. And if you go right down to the bottom of the specs here, you'll find that there's a maximum input rise time. Uh-huh, This one's actually maximum input rise and fall time as well.

But we're only concerned with the rise time here for the positive clock edge. I Haven't changed the full time here at all and you'll notice that it does actually change with voltage. So at 2 volts, it's a thousand nanoseconds. One microsecond is the maximum rise time.

A bit at the 4.5 or say 5 volts we're using that, then it the maximum rise time you can have is 500 nanoseconds. And where we saw that at you know, a 1 or 2 microseconds, it started to play up and sure enough, it's outside of the spec. It's not guaranteed if I take it right up to 50 microseconds here. Look, it's just.

it's just flashing the two most significant digits. It's just gone crazy. But 50 microseconds? that's still a pretty fast, more fast ish input edge. If you actually look at your square wave here, you might think that's practically instantaneous, right? It's it's the super fast input edge, not nanoseconds.

But hey, it's you know, tens of microseconds so it should work. but it doesn't. And what's going on here is that the input is that transitioning through the threshold voltage for your digital logic and you might be familiar with this. IV and High and V in low for the digital take.

The data sheet has specs on this, and instead of a really fast transition through that, we've got a real slow transition through and it's spending too much time in that undefined region and hence the gate might start to oscillate or do something weird, and you might think, well, what's the big deal here? Just make sure you have fast input edges. Well, this can be a problem on, say, for example, a reset pin of a microcontroller. for example. you might traditionally hook that up to an RC circuit on the input so that it shorts the pin.

When the power supply powers on and resets the chip for a certain amount of time, the capacitor charges up and then you've got a slow rise in input signal. If you feed that innocent digital logic that can go into a metastable state and ruin your day. or if you're designing a crude RC oscillator with some spare digital gates or something like that that can ruin your day as well. So a slow transition through this undefined threshold region can cause our oscillation or some other weird effect inside, unknown effect inside the gate.

It's basically not guaranteed. that's why there's a maximum spec for that rise time. Now, this probably shouldn't be confused with metastability, which is something I'm sure I've mentioned in a previous video somewhere I haven't done a specific video on it? Maybe I will in the future, but metastability has a similar end result in that the gate can oscillate these weird stuff like that. But in the case of metastability in clocked our systems, that has to do with our set up and hold times and not the actual slew rate of the input signal being in the undefined region.

But some people might call this sort of oscillation caused by this input signal a metastable state. But yet, just don't confuse it with metastability, which is a different R phenomenon. So let's take a look at our next issue here, which is noise on comparators. Now we've got a classic jellybean comparator here.

the Lm317 svatah CH on the non-inverting input here, which is our just a resistive divider from our 5 volt rail. So 2 1k resistors given us 2.5 volts a reference voltage here and the inverting input is our input signal. Now our input signal can be anything it can be, say a battery. This could be a classic low battery indicator.

For example, we have a reference voltage of battery drops below the reference voltage and your LED turns on or it signals are you know, the micro or something else. And here's our output: LED And if we turn that pot there we go, we can turn our LED off and on. No worries whatsoever, this is working just fine. Or is it? Hmm, what if we try and get it right on the edge like that? It could be a hold your tongue at the right angle.

I Can see that lead flicker, flicker, flicker flicker flicker. What's going on? Let's take a look at the scope. I've got my scopes set up at a slowish time base where there's at 500 milliseconds per division I in normal trigger mode so that it doesn't keep auto sweeping across the screen so it'll only capture on the positive edge when we transition through there. So here we go: I'll turn the pot, the LED should turn off and bingo.

it's captured it. No worries whatsoever, right? It looks fine. Hmm. Let's actually zoom in here.

What? What? What? Wha? What's going on there? Look, We've got some horrible oscillation there. Let's try that again, shall we? It'll our auto reset up. There we go and captured the other edge there. But there we go.

Bingo, We captured it again and you can see that we've got some oscillation in there. Wow Look at that. That's horrible. You can see that we're also going to get that when the LED switches from off to on as well.

On the negative edge, we're going to get the same sort of oscillation going on there and well, this might be just fine and dandy if you. Yeah, it just got a low battery LED circuit or something like that. Okay, you know, not a real issue, but hey, oscillations are undesirable. They can draw extra power I know, the LEDs draw in power but hey, you got an ultra-low power circuit and that wasn't going to a LED it was going off to some.

you know, triggering your micro or something like that. an interrupt service routine that could really when you die look I'll turn it really fast and it's sort of like narrow there. but if I turn it really slow, watch this boom look how long that is. the slower I did it and just like before with the rise time, the slower I did it the more oscillation that we actually got.

So what's here we go I'll go really slow again and wow look at that. whereas if I go really fast boom look there's not much of it. but there is still a solution and that could cause you know end of problems. Depending on the circuit implementation you've got real trap that one.

So there's the two problems: input, slew rate and also art noise on comparator circuits. What's happening here and how do we fix it? Well, let's go to the whiteboard so we'll jump straight into the solution. the Schmitt trigger. You might have got it from the title of the video and it's absolutely classic built in building block.

It's used everywhere and you may not even have realized it's used in places now. I'll very briefly and very simplistically. It's important to realize what's happening in a CMOS digital circuit. You might have seen this if You saw these.

seen the internal circuitry inside a typical Y CMOS gate. In this case, it's a CMOS inverter. Then you basically just got two MOSFETs Like this, the gates are tied together. you've got a P-channel an N-channel MOSFET, and the output here.

That's it. Now, we won't get into a detailed characteristic curves of MOSFETs and how they work, but suffice it to say that when your input signal changes because the two gates are tied together like this one transistors supposed to be on when the other one's off and vice-versa But that's not always the case if you have a slow change in input signal a that you know a slow slow rate like we saw like that. There's going to be a point where both of these transistors are on or partially on at the same time and that can cause excess current and other issues in digital logic. especially when they've just got a non Schmitt trigger input like this.

a standard CMOS input. They don't like having slow input slew rates. They like having just to be switched incredibly fast so that even one transistor is on all the other ones off. So let's get into Schmitt triggers.

now. you might have this symbol before inside an inverter gate, or on a microcontroller datasheet or something like that on one of the digital inputs. Let the reset art pin of a micro for example might have in the internal block diagram. might actually have this little symbol here and it could actually also be the other way around.

It could also be like that as well, but it doesn't. We won't get into details, it's the same thing. that means that this inverter or this particular part of the logic has a Schmitt trigger on the input. and shortly, we'll see why this symbol is actually a good representation visual representation of what actually happens in a Schmitt trigger.

So start the solution by taking a look at our classic comparator circuit here now. Oh, I'm just like using an Op amp here, but can be an Op Amp or a comparator. Op amps can't be used as comparators as I've done in mention in my Op Amp video which will be linked in down below if you haven't seen that. So we've got our input.

we've got our output positive and negative rail because that's going to make our just our theory and a little bit of math simpler and our input is tied to ground like this. So this is actually a zero crossing detector building block circuit. So if you want to detect if your circuit wants to detect a zero crossing on an input signal, this is how you do it. You might notice that I flipped this around because I kind of goofed it I wasn't thinking Anyway, we'll see why in a second.

Now let's take the example of our zero crossing detector works exactly the same for this circuit that we saw on the breadboard before, where the input where we had a single rail here, ground, and five volts and then our input was hearth rail at 2.5 volts works exactly the same way. So you should know how a basic comparator works. If our input, our inverting input is greater than in this case. if our input waveform goes high like this, above zero volts.

If it's greater than our reference voltage on the non-inverting input ie. ground, then our output is going to swing negative because the negative input is more positive than the positive input. So it's going. The output is just going to switch negative like this.

or if we had a grounded reference like we did in the previous example, to just go to ground. And likewise, once that input signal goes below the reference voltage ie. below zero volts here. boom, it switches the other way and that's it.

Simple, but we didn't see that in our example. We saw that we had some oscillation on the transition here, so it might be actually obvious what's going on here. In the real world, you get noise on signals. Nothing is absolutely perfect.

and in the case of a comparator here, that input threshold this input only needs to be a fraction 1/2 A/b stick a smidgen above or below the reference voltage here, be it on a micro volt, and then over whatever. you don't actually know better. Think of it basically as practically zero. It just needs to be above by any ridiculously small amount and it will switch.

And of course, your input signal is going to have noise on it. Always will. It's just a practical reality of real-world circuits now. I've greatly exaggerated the noise on our input sine-wave here like this, and you'll notice that it actually transitions through.

The noise actually transitions through that ground there several times. So in this case, if we take that down here, okay, this one's going negative. So it's going to go negative like this. but then you notice that because it went above zero.

So it's going to go negative. But then it transitioned because of noise briefly down below ground like this. So it's going to switch boom like that because of that matching one. They're basically going to transition several times like that before it actually stays down like this.

and likewise, over here. it's going to transition a few times there. Judah Noise. And that is what we saw on our practical circuit because there's no hysteresis or no Schmitt trigger action on our comparator here.

So our solution is to add what's called hysteresis otherwise known as a Schmitt trigger. A Schmitt trigger uses his to recess to give you the Schmitt trigger action. You can think of them as basically the same thing. If a circuit has hysteresis, it's basically acting as a Schmitt trigger.

And the way we do this is we actually want to widen the margin on the input where this are switches between positive and negative to prevent noise, Any certain level of noise from actually giving us multiple oscillations, multiple transitions like that. And the way we do this is to use positive feedback as opposed to negative feedback which you're familiar with with amplifiers. Amplifiers use negative feedback to amplify things what we're going to use. Positive feedback.

Usually positive feedback is a bad thing in electronics. You usually don't want it, but in this case it's very, very helpful because we can add Bingo! You might notice this looks somewhat like your negative feedback amplifier, but it goes back to the non-inverting input. So it's positive feedback. So what this positive feedback gives us and I'll go through working out these values in a minute is it gives us not just that, zero volt threshold voltage and zero volts, is it? Even if it's smidgen above or smidgen below, you get that oscillation.

What we want to do is add two different threshold levels. the upper threshold level and the lower threshold level. About that zero volt point or about 2.5 What we had in the circle four doesn't matter, then we can. If we have these two thresholds like this, then the noise like this is not going to affect it.

Let me show you, now, in this case, the zero volt threshold doesn't exist so we can ignore that. We've just got an upper and a lower one. which in this case, I've just arbitrarily said +10 millivolts and minus 10 millivolts about ground. Because let's say our noise in our system might be you know, 5 millivolts or 1 millivolt or something like that.

We want to avoid that level of noise causing our false triggering. So in this case, well, let's just assume it's high. Like normally we would have had the oscillation here if we just had that one zero volt level. but because we've got this upper threshold, it won't switch until it gets too that particular threshold right there and then it will go negative.

Just like before. the upper threshold works exactly the same way as our zero volts. but the key is because of the positive feedback action and we'll see how it works in a second. It won't then transition back high until it actually until the signal goes back below the lower threshold.

So in this case, you see how it transitioned back down under the upper threshold. there. it's not going to do anything. it only cares about the lower threshold before it switches back.

So this noise about here is not going to cause any problems whatsoever. It's only when it then it first goes through is lower threshold will it actually boom, switch back like that. And then once again, even if it transitions multiple times because of noise through this lower threshold, it's not going to flip back until it reaches the upper threshold again. So bingo with by adding some hysteresis, which is the difference between the upper threshold voltage and lower threshold voltage by Schmitt trigger action.

We've added a margin in there that eliminates our noise problems and we get a nice clean switching waveform beauty. So let's take a quick look at how this would actually work. Using some typical values here, let's say we had our plus/minus our rails here plus 10 volts minus 10 volts and our two resistors here. where the 9k will call that R 1 and 1k R 2.

You might realize why I've chosen 9k because it comes out as a nice round number. Now the formula is and it's very simple math. The upper threshold voltage here is: you might recognize this: r2 over r1 plus r2 times V out maximum. That's a voltage.

That's the voltage divider formula. You already should already know that. So it's basically the voltage divider. So you put nine r 1k over 10k total.

It gives you naught point. one times V out, which is the maximum voltage this particular comparator goes to. It may not always go to the rail, but let's assume it's ideal. goes to the rail to ten volts, then UT in this particular case would be plus one volt.

easy. And likewise for the lower threshold voltage down here exactly the same way LT equals minus 1 volt. So in this case, instead of 10 millivolt example we use before, the example would be plus 1 volts and minus 1 volts. So you could have half a volt of noise on there and it wouldn't cause a problem.

So what's the mechanism that makes this work? Well, it's fairly simple. Remember that we had ground here before. Well, ground is now here. and the voltage at the reference voltage here is the point on this voltage divider.

a reference to this ground. Here this could be some other voltage, could be two and a half volts or whatever. Doesn't matter, and the formula just gets a bit Messier But it works the same way, right? So let's take the case where our output is high. Okay, so output here is plus 10 volts.

So instead of our reference voltage being zero volts like it was before we are ground there, it's now that naught. plus 1 volts Because of this voltage divider action, this reference voltage is now plus 1 volt. So it won't action. It won't do anything.

It won't switch unless it gets to plus 1 volt here. So once that signal goes once, just once, over even a smidgen over. Just like our our issue before with zero volts, even particularly a micro volt and nano volt over whatever it is, the first time it does that, it'll switch down like this and the output will then switch to minus 10 volts down here. What have we got on our voltage divide? Instead of having plus 1 volt here? Because we've got minus 10 volts here.

we're going to have minus 1 volt here and our reference voltage is change to see how we had a fixed reference voltage before. But now we've got a reference voltage which toggles between two different values written depending on our output state. It's clever. So our reference comparator voltage here is immediately.

let's just say instantaneously switched from plus 1 volt to minus 1 volt. Like this and be that noise doesn't cause an issue anymore because it now needs to go through the lower threshold. Point down here the minus 1 volt before it can switch back. Beautiful.

And that is hysteresis. And that's the Schmitt trio. So why is it called a Schmitt trigger? Well, as common in electronics, it's named after the inventor in this case autosh Witt in 1934. And why is it called a trigger? Because well, kind of warm fuzzy because once it goes through this fresh hold, it triggers it into the next condition.

Something like that, it's a Schmitt trigger. So I Promise to tell you where the Schmitt trigger symbol came from? Well, you have to look at the transfer function and I won't go into details, but it's basically the output referred to the in putting a graphical representation like this. And in this case the output voltage are switches between plus 10 volts and minus 10 volts and then the input voltage switches between plus and minus 1 volt. Like this, you can see that if you invert it like that to scale on the input.

Bingo. That is what it looks like. So you can see that our Mitt symbol is actually well. In this particular case, they put a slope on it like that and that is our Schmitt trigger.

It can also go the opposite direction and that's supposed to imply a directionality, but it doesn't. just draw it either way. fine. and sometimes it's drawn like sloped like that.

either that way or that way slightly. Or sometimes it's drawn with vertical lines like we get in. This ideal representation here doesn't really matter. You get the idea.

So let's go back to our breadboard and actually fix this to see if our oscillation disappears when we add some positive feedback. Now in this case, I won't go through the calculations because using the Lm317 an open collector output, we're driving a lead like this and it's like to add in a resistor going back like this. You know it's either ground like this, but there's going to be some drop across there and it's not going to be quite ground and it's going to be a little bit tricky. but it doesn't matter.

Just whack a resistor in there. you know, 10k, 20k, something like that. See if it makes a difference. So here's our original wire circuit with no hysteresis.

no positive feedback whatsoever. In this case, we've got 1k 1k resistor here now. I'm going to add a 15k resistor from the output here back to our non-inverting input. so we've added some positive feedback there and the calculation gets a little bit tricky with the voltage divider and you know everything else.

But and the fact that we've got two Led an open collector output doesn't matter. 15k, it's just like order of magnitude bigger. I Just happen to have a 15k line on the bench so it'll add some hysteresis to this and we should see a difference. So this is before.

Okay, I'll do this all in a single shot to show that there's no funny business. Okay, there's our positive and negative thresholds. I've got that auto sorry normal triggering on the scope so we can see that I'm not going to change any settings and I've got my 15k resistor here. Let me plug it in and come on here we go And Bingo fixed.

There we go look at that even if we go right in. we've only got the one transition. winner winner chicken dinner. We just added some hysteresis to that and our noise problem is gone.

Schmitt Triggers Beautiful. And as for fixing our dodgy clock input here? well, I won't bore you with the details, but suffice it to say you just either use a counter that has a Schmitt trigger on the clock input, which most of them do I chose the last M4 HC 161 because that in particular didn't have a Schmitt trigger input or you can just add, you know, a seven for once for Schmitt trigger inverter in there. No worries, they'll fix the problem. Try it yourself.

Now We saw how you can do an analog Schmitt trigger with the comparator and the feedback resistors. The positive feedback. That's all hunky-dory But how do they do it inside? say a seven for HC one for Schmitt inverter or inside any other chip that has a CMOS digital chip that has a Schmitt trigger action on the input. Well, they can't have resistors in there and do it or analog II that you know that just doesn't work in the process.

They have to do it using discrete MOSFET transistors And here's a shot of how they actually do it from our Fairchild and it is rather clever. Look at the input here. it actually uses four stacked MOSFETs on the input to antenna channel down the bottom, two P channel up the top and this is different from the arrangement I showed on the whiteboard before with just the single transistor top and bottom and then they've got additional two MOSFETs there P 3 and N 3 which act as our source followers, actually going from the output, feeding back the output, effectively feeding back the output signal back to the middle of either the upper or the lower totem segment up there on the input and that can be used as a Schmitt trigger action. It's fantastic.

So let's take a brief and simplistic look at how it works. If you've got 0 volts in on the input, then the upper two MOSFETs P1 and P2 are going to be on. P3 is going to be off and N1 and N2 down the bottom of the pole totem pole. There I got to be off and N3 is going to be on.

But if we start to raise that input voltage by a little bit, let's just say one volt for argument's sake, then it sort of reaches a threshold where N1 which was off before, actually switches on and that forms a voltage divider between VCC and ground with N3. So N1 and N3 are now forming that voltage divider similar to what the resistors were doing in that analog E-type configuration. and a similar sort of things happen in here as that input voltage Rises. And likewise, if you went down from 5 volts down to then the same mirror thing would happen with P1 and P3 at the top, they would act as our voltage dividers between VCC and ground so that voltage divider of M1 and N 3 then biases are the source of N2.

They're said to a particular threshold level which then when a transition and then if you raise the input voltage even more, it transitions through that and switches that on and boom It did, flips a trigger point over and and the mirror starts to happen. With the art P 1 and P 2 and P 3 at the top. That's a very clever system I Love it! And then of course I passed that on the output. You'll see that out is in the middle there.

then the other two P 4 and n 4. There are just another inverter with a and then an output driver using P 6 and n 6. But that's how you basically configure a Schmitt trigger inside any sort of modern CMOS digital logic. And I mentioned way back towards the start of this video that when you have both of the transistors on in a typical configuration, a ramp that inputs slowly.

Both transistors can partially turn on. You can draw excess current and also if it oscillates, you can draw excess current as well. Well, Unfortunately, the Schmitt trigger our configuration doesn't give you a free lunch here. It doesn't let you escape that this if you have a slow changing input, it still does actually increase the current.

But of course it doesn't oscillate and souls all that sort of problem. But you can see here that this is the supply current in milliamps versus the input voltage in volts for a Schmitt trigger gate here. and you can see that at various threshold at the two threshold voltages there, it's sort of peaks at, you know, at much higher current. So yeah, if you ramp it through, it can still cause little current spikes.

So yeah, you don't completely get away with that unfortunately. So there you go: I Hope you enjoyed that. Look at the Schmitt trigger and hysteresis. Anyway, if you liked it, please give it a big thumbs up.

And if you want to discuss it, links down below or comments. Catch you next time you.