Why build an entire computer on breadboards?

Key Takeaways

is it a good idea to build an entire computer from scratch on bread boards like this well it might seem like a strange question for someone who's basically made an entire youtube channel out of doing precisely that and I sell kits with all the parts so you can do it yourself so obviously I'm a pretty big fan of building computers on bread boards and the big reason I really like it is that it forces you to think through how everything works you know what I felt like something was missing with these projects where you get a premade circuit board and you just solder a bunch of components to it and don't get me wrong it's great if you want to learn how to solder but once you know that I I just don't see how attaching a bunch of components to a circuit board teaches you much about electronics in fact the typical experience is that you solder the whole thing together plug it in and it it just works the first time there's nothing to troubleshoot there's no need to understand how it works and so you don't this on the other hand you know it really forces you to think about each individual connection to each pin and as you build each section testing it you know each as you go there's a very good chance things aren't going to work perfectly the first time and that's a good thing you know forces you to think about why and it helps you build a deeper intuition for how it all works and so that's what I really like about doing big projects like this on bread boards but aside from being a lot more work which which i think is a good thing there are definitely some caveats to be aware of when building complex projects like this on bread boards first not all bread boards are the same there's a big difference in quality between the cheap $2.00 bread boards you can get and a higher quality you know eight or nine dollar bread board and it's true they look pretty similar but let's take a look inside each row on their bread board is connected with a metal strip that grabs the wires that are inserted and if I dig under the backing here I can remove that metal piece so we can take a closer look at it and here it is is what it looks like if we take a closer look you can see the wires are inserted like this here and make contact with with the metal here and in a good quality breadboard these this metal is nice and flexible and you can see even if you insert the wire it kind of a weird angle like this it it flexes like that and then pops back nicely you can also see it's nicely shaped so that when you insert the wire from the top which is which is what you'd be doing you know it kind of finds its way in there nicely so these make very good contact with the wire bring back so they last a long time and everything else now let's take a look at the cheaper breadboards a construction on the cheaper breadboard is pretty much the same so it's actually pretty hard to tell the quality just by looking at a breadboard which is unfortunate because it means it's easy to pay for a high-quality breadboard and end up receiving one that's lower quality and I may not even be the fault of the person selling it to you since they may just not realize that there's such a big difference but anyway here's the same piece from the cheap breadboard so let's take a look at the difference and so you can see it's the same basic shape but right away you can see there's a huge difference here you know the metal is just not not as springy and so it doesn't snap back together and so it may not make great contact with the wire as you can see here and it actually gets worse you know over time if you you know insert something big you can actually bend this out of shape and I mean that's maybe a little bit extreme for what you might stick into a breadboard but once that's bent out of shape you know nothing you know you're not going to make great contact there and so the quality of this breadboard and the quality of the connections you're going to get is is much worse so there's a big quality difference and you can even see the you know just the shape of the of the the top there where the wire goes in is very different and and much less consistent on the lower quality one so the quality of the breadboard matters you can run into a lot of problems building a more complex project like a computer on these cheaper bread boards because he just can't be assured that all the wires are making good contact and you can check out my website for more information on what bread boards I recommend and of course if you get any of my kids they're all gonna come with these high quality bread boards but okay even with the best bread boards there's still a lot of limitations to building a complex design like a computer on bread boards versus a custom printed circuit board or even these you know soldered prototyping boards and that's because the physical properties of the conductors in a circuit whether that's the traces on a printed circuit board or the wires and and bread boards in in something like this you know those conductors and everything have physical properties that affect the circuit for example it's easy to look at two wires like this you know let's say these two wires here and think well okay you know this wire connects this point here to this point here this wire connects this point here to this point up here and maybe they don't really have anything to do with each other and that's not entirely true you know anytime you have two pieces of metal close together like you have these two wires that are close together maybe you actually have a capacitor and that's why the schematic symbol for a capacitor is two plates next to each other but not touching and the way a capacitor works is you have a difference in electrical potential on either side and a charge builds up between these two plates and another way of saying that you have a difference in electrical potential between two points is saying that you have a voltage across those points that's you know that's what a voltage is it's always you know sort of a voltage measured between two points and it's just the difference in electrical potential between those points but with a capacitor in here if you try to change that voltage the capacitor will will actually try to prevent that you know so if you try to increase the voltage by by you know adding more charge to one side the capacitor is actually going to absorb that charge and charge up a bit before the the voltage between these points actually changes and then if you try to decrease the voltage here by by pulling some of that charge away the capacitor will discharge as much as it can to keep the voltage from dropping and you know sometimes you want that so for example on the power rails you actually don't want the voltage to fluctuate right we have five volts coming in and you know we want five volts everywhere on our power rails to be actually 5 volts we don't want it dropping and things like that so if if a chip is switching circuit on and off and it has to draw more current to do that we don't want the voltage on that power rail to drop and so it's actually a good practice to add some capacitors just across the power rail like this so actually add a couple of point 1 micro farad capacitors here across the power rails to help stabilize the 5 volts that are on those power rails and really the the best practice here is to have one of these capacitors for every chip that's on on in your circuit so for example we could have a capacitor here directly from 5 volts to ground across this power rail for for this chip here like this so that means that if this chip has has any change in the amount of current it needs to draw from them from its power rails it's always going to see a consistent 5 volts or as close as close as we can get to that and so the closer you can put the capacitor to the power inputs for any particular chip the the more the better stabilized the power is going to be for for that chip and that's I guess one of the other maybe drawbacks of breadboards is that it's it's you know kind of hard or at least inconvenient to get these capacitors directly across the power rails of a chip like this you've got to kind of put it across and some of these chips they're the powers are just gonna be an inconvenient place but I think you know for what we're doing it's probably good enough just to have a couple capacitors here these will still help stabilize the power rails without you know really getting in the way too much so that's fine you know we can add capacitors here if we want to stabilize the voltage and keep the voltage from changing like we do on our power rails but elsewhere you know we have signals that need to change voltage rapidly you know any of these signals need needs to change voltage rapidly because they're you know carrying signals and and and and that has to change and so the stray capacitance that inherently exists in the breadboard or even a printed circuit board can cause problems and you know since one factor that determines how much charge a capacitor can hold is the is actually the physical area of the conductive plates in that capacitor you're more likely to have a lot more capacitance in a breadboard circuit just because you've got a lot more metal in here you know in the breadboard and in all these wires then you might have in other types of circuits now another related phenomenon is inductance and if you've learned a little bit of physics you might know that anytime you have a current flowing through a wire like this there's a magnetic field that is generated around that wire of course if you have a lot of wire especially wound up like this you can use that magnetic field to do work and that's how a motor works and that actually works in Reverse as well so right now I'm using current to generate a magnetic field that's pushing against this fix magnet here but any motor is also a generator and that's because a changing magnetic field around the wire will induce a voltage in the wire so when I spin this the magnetic field from the permanent magnet changes relative to the wire and induces a voltage in the wire so what does this mean for a breadboard computer well you know if you've got a wire with a current going through it you're gonna have a magnetic field around that wire and if the current flowing through the wire changes then the magnetic field is going to change so if we go from a smaller current to a larger current we're gonna go from a smaller magnetic field to a larger magnetic field but remember if you have a changing magnetic field around the wire then the voltage that'll induce a voltage in the wire and it just so happens that the voltage that's induced will oppose the change in current and say you've got capacitance which opposes a change in voltage and inductance which opposes a change in current and both effects are relatively small unless that voltage or current is changing very rapidly well how rapidly you know if we take the 6502 computer that we're building I've said that I plan to run this at one megahertz so are we gonna have any issues with one megahertz signals you know so that's potentially alternating between zero and five volts a million times per second you know that can be a problem well we can run a little experiment I've got a breadboard here with a bunch of connections and I can feed a signal in on one side and you can see I'm measuring that signal as well and we look on the oscilloscope you can see there's this 100 kilohertz sine wave that we're measuring going in and so that goes in on this side and then it goes through a whole bunch of connections and then we can measure it coming out over here now I also have a resistor here to kind of isolate the input and output measurements and aside from that induced current that we're going to be looking at that I mentioned I don't expect any current flowing through that resistor so Ohm's law says no current flowing through it means no change in voltage so we should measure the same voltage going in here and coming out over here so here's the signal going in it's a 1 kilohertz sine wave and we're measuring that over on the left here but we'd also measure the output on the right and so if I overlay that you can see it's pretty much the same thing and you know that's not too much of a surprise since we basically just have a wire going through the breadboard but if we increase the frequency and can dial this up so that was 100 kilohertz so if we keep going up when got to one night one megahertz zoom in here and it still looks like pretty much the same signal going in and out but if we keep going higher and higher frequencies zoom in here you actually start to see something happening here you see the output which is the yellow actually now the yellow is the input so you see so you see the green which is the output is actually shifting a little bit from the yellow so there's there's a phase shift happening there which is kind of interesting and that is perhaps because that inductance and capacitance that I talked about is going to resist a change in voltage it's going to resist a change in current and so it may delay and actually cause that phase shift so we're starting to see that and if we go higher and higher frequencies you'll see not only does the phase shift increase but you also see the green which is the output is actually decreasing so it's starting to attenuate and so if we go higher and higher frequencies and that's about as high as we can go up to 20 megahertz as this is as high as I can go with this but you can start to see that the green signal which is our output is being attenuated it's being phase shifted and it's being attenuated and actually what I can do is I can automatically sweep through all of the frequencies here so to do a let's see frequency response analysis and run analysis and what this will do is I'll actually sweep through all the frequencies from a hundred Hertz all the way up to 20 megahertz which is as fast as this will go and what you're seeing is it's plotting the phase shift which is I believe the red line so you can see that's staying pretty close to zero and then it's also plotting the gain or in our case attenuation which is the blue line and that's also staying pretty close to zero and here we are at 100 kilohertz and still pretty close to zero as we approach 1 megahertz you see the phase shift is starting to change and that's you know what we saw when we were looking at it ourselves and then as we get you know looks kind of 5 megahertz or Beyond we're starting to see the blue line is also dipping by you know a few decibels and so that's showing that at higher and higher frequencies we're starting to lose some of our signal integrity something else is kind of interesting if we go back and just look at the waveforms here remember that the attenuation and phase shift that we're seeing is result of both capacitance between conductors here as well as inductance which is actually a magnetic field that's generated around these conductors so if we actually change sort of the physical shape here or properties or you know sort of relationships between the different wires you can see just as I poke at this the the phase shift and the the attenuation that we're seeing is is changing a bit as well and you know this is at 20 megahertz but I think that's pretty interesting to see that so you can see at higher frequencies we are starting to see some signal integrity issues but you know we want to run our computer at 1 megahertz here and 1 megahertz looks like you know no problem right we're you know still zero attenuation our phase shift here hasn't shifted by very much at all 1 megahertz and even the attenuation that we do see is exaggerated quite a bit because of that resistor I added to make this more of a worst-case demonstration so even in this you know pretty messy scenario that we've got here things look pretty clean up to 1 megahertz so we ought to be fine right well it's not so simple that a clock for the computer is 1 megahertz but it's not a sine wave like we've been looking at it's a square wave you know just toggles from 0 to 5 volts but in reality any signal including a square wave is actually made up of sine waves so here's a 1 megahertz sine wave and I'm generating with this first formula here which has a frequency of 1 million and so you can see the period here is 1 times 10 to the minus sixth or one microsecond so that's 1 megahertz but of course it's a sine wave and we're looking for a square wave well we can get closer to a square wave by adding another sine wave to it so here is a 3 megahertz sine wave and I've made it one-third of the amplitude and if I add these two together so f of X and G of X if I add those together we get this which is a little bit more square shaped but if I add to that H of X which is a 5 megahertz square wave at 150 amplitude so if we add H of X you can see it gets a little more square shaped and we can keep going I can add this next sine wave which is 7 megahertz and you can see it gets even more square shaped you know the slopes are a little bit steeper here the tops are a little bit flatter and in fact to get a true square wave we could add up all of the odd-numbered multiples of the fun mental frequency which in this case is one megahertz and if we add all of those up all the way to infinity and that would give us a true square wave so let's take a look at this other set up here that I've got it actually got a formula here that represents the sum of odd-numbered multiples of a 1 megahertz sine wave and I've got this 2k minus 1 factor in here as well as dividing by 2 K minus 1 so if K goes you know 1 2 3 4 5 then 2k minus 1 is going to go 1 3 5 7 9 so that gets all of the odd multiples and then we're summing up all of those for K going from 1 all the way up to N which right now n is also 1 so right now we're only getting the first term which is why we just see that 1 megahertz sine wave then I've got some other stuff in here just to kind of scale this up and we're adding 2 and a half here just so that this this wave overall goes from you know around zero at the bottom to sort of around 5 at the top just to kind of give us something like that 0 to 5 volt square wave that we're looking for but now this makes it easy for us to add more terms because I can just change that N and so this is where we were before with n equal to 4 with 4 terms but we can keep going and the more and more terms that I add the more and more it looks like a square wave but also the more terms that we add the more the higher the frequency so this looks like a nice sharp square wave but the frequency you know their maximum frequency that I'm using to get that is over 100 megahertz so you know in reality because high frequencies like 115 megahertz are gonna be very susceptible to you know small amounts of capacitance from small amounts of inductance like we've got in a breadboard circuit we're definitely not going to see you know perfect square waves with you know perfectly steep slopes like this or anything like that in real life so the question is you know is that a big deal well my last video we looked at this timing diagram for the 6502 cpu and we figured out that we weren't going to have any problems come complying with all of the constraints that are set out in here but there's one natural timing requirement that I kind of glossed over that actually turns out to be one of the hardest constraints to meet and that's you know this time here this T which is the fall time and TR which is the rise time of the clock so what is that requirement fall time and rise time well we flip over here we can see full time rise time is a maximum of five nanoseconds five nanoseconds is not a lot of time and just to get a picture of that I can add two lines here that are five nanoseconds apart so I'll add these two and these are centered around you know five times ten to the minus seventh which is half a microsecond so it's centered around this point you know sort of half way through which is where we have this transition and then it's bracketed you know plus or minus two and a half nanoseconds basically ten to the minus nine so you know minus two and a half to two plus two and a half is going to give us a gap here between these two lines of five nanoseconds and so as you can see you know five nanoseconds is really short so let me zoom in here and see if we can you know see what's going on a little bit better so I'll change the x-axis to go from four times ten to the minus seven to six times ten to the minus seven so there we can see things a little bit better and you can see we're going from you know five volts down to zero volts within that five nanosecond window but remember for this signal to exist that we're graphing here this red signal there must be a component of it that's you know a hundred and fifteen megahertz so if we don't think our breadboard circuits gonna handle 115 megahertz signal all that well then we shouldn't expect to see a square wave with slopes you know quite this steep so if we go down to you know let's say we're maximum frequency is closer to you know maybe thirty megahertz you can see here you know even in theory we definitely can't transition from five volts down to zero volts within this five nanosecond window but you know one piece of good news is we don't actually have to get all the way from five volts down to zero volts in that timeframe you know we just have to get from high to low and it turns out that high is really anything above about three and a half volts and low is anything below about one and a half volts so really you know something like like this would be in spec right because we're going from from the absolute minimum of what's considered high to the absolute maximum of what's considered low we're doing that within that five nanosecond window so this would technically be in and so maybe that maximum frequency is really something more like you know it's hard to see but maybe it's maybe it's like 40 43 45 megahertz and of course this is just a model that shows all of the frequencies up to forty five megahertz and then none of the frequencies above that in real life you know we'd still have all of the frequencies it's just that as we get higher they become more and more attenuated not to mention phase-shifted as we saw but it you know the point is that the model you know isn't perfect here but hopefully it gives you just you know some sort of idea of what's going on the other reality is that this 555 timer clock module that I've been using is not exactly great at meeting that 5 nano second requirement either and it's been working fine if we take a look at it you can see you know this is 5 nanoseconds per division so you know going from you know let's say it's 1 and a half volts here up to the 3 and a half volts which is actually about as high as it gets is a little bit more than 5 nanoseconds so you know it's not technically in spec but it's been working fine so for you know an educational or hobby project like this you know particularly if we aren't running close to the 14 megahertz limit of the processor or we aren't at any temperature extremes or anything like that yeah there's plenty of Tolerance and and close enough is probably good enough but let's hook up this 1 megahertz oscillator and measure it and see how close we really are now I've run out of room so I'll need to add another breadboard here you know connect the power rails and add the oscillator here and the oscillator is pretty straightforward you know we just connect power and ground to it so power goes here to the sort top left pin ground to the bottom right pin over here and that powers it up and then this top right pin we're just going to get a 1 megahertz square wave coming out of that so let me hook up the power here and we can take a look at that obviously the computer is not going to do anything because we have no clock connected to it now let's measure the output of our oscillator so with the oscilloscope probe here rather than using the ground lead the best way to measure this because the ground lead I mean this is a wire it's got inductance in it it's gonna have all the same problems that all these other wires have so to get the best measurement you want to get rid of this you also want to get rid of the clip and you just want to use the probe and then actually this little ring here is is also ground so you can use this little ground thing stick that on there like that and then you've got a ground and then the probe so I'll put the probe here on the output that we want to measure and then the ground thing is just going to connect or sort of you know be stuck to the case here and that's that's gonna be ground because the case is grounded so that's the best way to get a really accurate measurement of what's coming directly out of the oscillator here and if we take a look at that we can see we've got a 1 megahertz square wave here and if we zoom in a little bit you can see there is a little bit of overshoot and a little bit of ringing there kind of like we expected but overall it's a pretty sharp edge so let's zoom in here too so this is now 5 nanoseconds per division so between these vertical lines here these vertical gridlines is 5 nanoseconds so you can see the actual rise time is much less than 5 nanoseconds for the for the whole thing we kind of shift things over here so let's say our rise is starting right there we're already up to 5 volts within about you know 2 and a half maybe 3 nanoseconds so measuring what's coming directly out of the oscillator like this we've got a really great signal but of course we're kind of avoiding as much of the breadboard as possible but with this measurement technique in reality by the time our signal gets to our processor it might be a degraded a little bit but let's let's take a look at that let's do that so what we'll do is I'll connect the clock in to the rest of our clock signal from the rest of our computer so that's there and we'll connect that down to our clock there and so you can see we're taking the output of the clock and it's going in or directly into both the interface adapter as well as up to the the CPU here and so the question is going to be what is our clock look like as it's going into the CPU here so we can do the same thing I can measure get this little ground probe into our ground and then get our regular probe into that part of the breadboard and that'll measure the the clock signal up here and if you notice it's it's really not that bad again you know this is five nanosecond per division here so we're getting you know all the way from zero volts up to five volts in you know maybe it's got taken a little bit more time you know maybe this the slopes not quite as steep perhaps I'll edit this video to overlay the two of these so you can so you can really compare but at least to my eye looking at this just my memory of what the other one looked like it looks like the slope is a little bit less steep but not terribly so and this is actually really encouraging because you remember that when we were looking you know theoretically at the just doing the math we needed you know 100 megahertz plus signals to get from zero volts to 5 volts within 5 nano seconds so this is this is a very steep so we must actually be able to maintain those those really high frequency or at least enough of those really high frequency signals to be able to get something like this so this is very encouraging though of course I've done everything I can to get the best possible case here you know obviously I'm using high-quality breadboards these wires are as short as they could possibly be and they route as directly as I could make them route and I think that is important for the clock circuit here so if you know some of these other signals maybe you know you can have longer wires or whatever but the clock signal is one where you know because the datasheet has that spec of 5 nanosecond rise time 5 nanosecond fall time in order to meet that spec it seems like it's worth you know going to a little bit of extra effort to make those connections nice and neat and as short as they need to be so everything looks good and we're running a clock at 1 megahertz so let's reset and it looks like there is still one problem which is the our program to display hello world doesn't actually seem to be working anymore so if we reset you know we could try to disconnect reconnect power but it's just not working though it is sort of doing something so let me actually disconnect and reconnect power disconnect this reconnect it and you can see the LCD comes up and it's not initialized but if we reset the computer so the hello world program runs you can see the LCD does appear to initialize you know it gets set to to line mode the carcere shows up but it doesn't do anything else so it looks like there still might be some sort of problem running the computer at this higher speed of one megahertz so what's going on well there's something I remember seeing in the datasheet for the LCD controller you know a while ago when we first hooked it up and that is that there there are several places in the datasheet where it says you have to be sure that it's not in the busy state before sending an instruction right and when instructions being executed no instruction other than a busy flag read instruction can be executed the busy flag is set to 1 when instructions being executed check it to make sure it's zero before sending in another instruction you know be sure it's not in the busy state before sending instruction and at the time we first hooked the LCD module up I kind of glossed over this because we're running the clock so slowly that there was basically no chance that the LCD would still be executing one instruction by the time we sent it the next instruction in fact I don't even know if I mentioned at the time but I do remember seeing this in the datasheet and making a mental note that I could you know probably ignore it while we were running the clock really slowly but that eventually I need to come back and actually add the code to read the busy flag and wait before sending each instruction or character to the LCD module but now that we've got a good clock signal it looks like everything else is working because the LCD is being initialized in the first instructions sent to the LCD is the one that initializes it and sets it to two line display mode you puts it in this state it's just that all the rest of the instructions are probably coming too quickly before it's ready for any of them so these instructions you know turning the display on cursor on clearing the display these party aren't actually working even though the cursor is on and the display is cleared it's just that those are the defaults when the LCD first powers on anyway and then of course these commands were it actually prints each character aren't working either you know all of these instructions are being sent so quickly that the LCD isn't actually done executing the first instruction before the rest of them are sent and that's because we don't have any code to check for that busy flag and wait to send the next instruction and the LCD is ready and at 1 megahertz it's just going too fast for the LCD so since we've got a good clock signal I I think all we need to do is just add the check for the busy flag and we can do that entirely in software and that's what I'm going to do in the next video and remember if you want to follow along with these videos you can get all the parts I'm using over at my website Eater net slash 6502 and for now at least we're still able to ship kits during the ongoing quran virus at pandemic and I'll keep my website up to date if that changes but you know if you're stuck inside looking for something to do got you covered you