Copy Initials Fact with airSlate SignNow

Copy initials fact

hello and welcome back okay so i've actually just finished the earlier control pcb so i'm waiting for that to be manufactured pretty sure this is working very stably this now finishes at value 24 just to satisfy the crazy people who think counting should start at one and not zero okay so i've been giving some thought to better ways of getting some code from my pc onto the processor because at the moment i'm just literally using my eeprom programmer to program code onto a rom each time i make the slightest change i really need to get that iteration time down a bit if i'm going to write some big and interesting programs so let's talk about some of the ideas i've had and what i'm thinking about doing let's start off with looking at the current memory map i've got the 32k rom that we write the programs onto the upper half of memory is 32 kilobytes of ram that we use as stack and working space now i've actually had a few different ideas over the course of constructing this build but the most recent plan i was working to when i put together this memory configuration was to have the the ram and rom split as it was but build a module that i could fit into the rom zip socket so the idea is i would have a rom emulator that could allow me to upload program code directly to a little child board and then when i was done i could write that code onto a rom chip to swap in and out easily now i still think the rom emulator is a good idea but there is one thing that's unsatisfactory about it the emulator would require like a modern microcontroller and some chips that are a bit more advanced i'm trying to make the main build out of and whilst it would only be a debug development tool i'd still kind of rather avoid it now ever since i built the uart i've been thinking that could be used for getting code into the build in fact right now i could write a program stick it onto the rom and have it received bytes over a protocol from the pc build a program in the ram and shift execution to it but that program would only take a very small fraction of the 32 kilobytes of rom and so what i really need is a way of having more ram in the system there's a few ways i could do that one of which would be to introduce banking with banking you have some kind of indirect mapping between one or more regions of memory and a larger pool of memory outside of the standard address range so my initial thoughts was to have four 16k slots and then have a series of 16k pages where i could define a mapping from the larger address space into the smaller address space now obviously the idea here would be to make the default map at least one of the rom portions into the first slot of memory and then it would be able to do whatever shuffling was necessary and change the mapping to whatever it wanted and if we really needed it we could make all of memory into ram this kind of memory remapping is what a lot of the 8-bit computers that appeared later on used to have more than 64 kilobytes of addressable memory older pcs also used a system called ems which is expanded memory system which had a set of pages that could be mapped into a region in high memory which allowed those pcs to move beyond the one megabyte limit of the original 16-bit x86 processors after thinking about this for a while i decided against it but it's definitely worth mentioning because it was an interesting possibility so what i actually want to do is something simpler i like to go with a flat simple 64 kilobytes of ram now in terms of my executing programs this takes away all the complexity of handling the banking and the difficulty of trying to split stuff between the ram and rom but i still have to solve the problem of how i'm going to get my programs into the build so i'm still going to have the 32 kilobyte rom chip i'm going to build circuitry to copy that rom chip into the first half of ram on first power up and this will afford me a great deal of flexibility i'm thinking i can have a program on a rom chip that can monitor the uart to receive code from the pc or i could have a program that loads some kind of a menu of programs up from some long term storage but i could still continue to use the eeprom programmer in the same way i currently am and i actually gained the advantage that instead of separately allocating variables and constant data i could store my variables inside that base 32 kilobytes and pre-initialize them to useful values which would make the code slightly simpler well this technique is used elsewhere and it's referred to as shadow ram and so what i'd like to do today is work out how to retrofit the extra 32 kilobytes of ram into my build and then in a future video i'll look at the circuit for actually performing the copy now let's take a look at what it would take to add the extra 32k ram let's rename these to something a bit more sensible now the wiring of the second ram chip will be exactly the same as this one apart from the different chip enable line we need to add a capacitor for it but it really is going to be that simple rooting that's going to be tough but problem for another day now i've got to work out the copy circuit and design a whole new pcb that's going to include that circuit but for now what i want to do is take the simplest method i can come up with to get 64k of ram into the build so i can just test a few things today now if we look at the rom chip we're reminded that all of these lines are actually in common between all of the memory chips apart from the two halves of ram have got a different chip enable and rom has the second chip enable but over here what would be the right enable line is just hard wired to vcc that's active low now we've got ram chips that are dip 28s like this so if i could fit one of those in here but wire this line to the right enable that should give us 64k of memory on the current board and i'd like to try and do that okay now i'm just going to pull this memory module out and see what we can do okay now if i'm going to hack together a test form of the circuit i want to build i need to gain access to this line here which is the right enable i also want the memory acknowledge line which is here and the reset line which is here i could probably get with a long wire from elsewhere but if i'm soldering to here i may as well do that one now i will need to be very careful not to provide too much torque on that i've sold it to one of the other pins as well partially because i wouldn't mind testing the timing off that one because this is just part of the delay chain that looks like a bunch of good connections now this is a round chip wiring is pretty much the same as the rom chips but this pin is the right enable which here on the rom socket here is wired to five volts so what we need to do is work out a way of getting a round chip in here but not connecting that pin up so i'm thinking put it in a socket the socket on a little bit of board and get some pins on the bottom there and then i can remove the one pin we don't want connected and put a connector in there i've got two pcbs here because i think we can do something very similar for getting the wrong chip connected back up and that's going to be the pin that would be right enabled modern through-hole plated holes are so much easier to solder than these strip boards i'll be honest solving this problem was either going to be very complicated breadboard wiring that would have taken quite a long time or a really tricky piece of soldering i actually contemplated going straight to pcb on my faults here okay i've actually done that the opposite way when i planned so i will put this wire in and i'll need to cut this wire i've lifted a track there only really i'm interested in the first two pins here but in the right angles the fives are the smallest i've got so a notch here is in the wrong place if i put this ram chip in here if i connect that which is the right enable line on here it doesn't go on to anything else okay i'm guessing for that pin that sunk didn't have a good solder joint that one at the end doesn't look great now hopefully you'll understand this so this sits in like that that's the right enable pin connect that to the right enable pin there and in theory this should now be a 64k ram module okay crazy oversight i almost forgot i need to split the tracks by a cruel twist of fate that's every pin working apart from the one we're most interested in it's difficult to see but i can just about make it out but that's a dry joint around that pin finally now we've got the interesting challenge of trying to work out how to test this let's go see if we can make it work in the build okay so i spent some time writing some test code to try out this new ram chip now actually test it out we've got some quite complicated operations to do so let me explain what i've done this program includes two binary files in it which are two other programs so fibonacci dot bin is the fibonacci sequence calculating code that you've seen me run before i need some better naming on this but test2.bin is a fresh piece of code that i've just written now i've been putting a lot of work into improving the tool chain i'm using here so the assembler's got some new functionality for example this include bin option but also i've got the capability now of putting expressions into constants so this is a quick memory copy routine it transfers four bytes in each iteration of the loop and this basically just uses these two labels to work out how many iterations of the loop is necessary so instead of calculating it and then storing the value i can just type in expressions and the assembler can work it all out from the whatever inputs are necessary it's quite a powerful feature and it's going to be very useful as i write more code so i've got a few print statements on the lcd but the core of this work is this copy which takes the contents of test2.bin and it uploads to address 9000 which is an address in the ram region and then the second mem copy loop is copying the fibonacci rom to address zero a triple zero and then i call address nine triple zero which is this bit of rom i have uploaded first now i've been improving the ascender output so i've now got a better quality of dump from the assembler so i can help debug it but you can see here the chunks of binary that it's stuck into the file so there's the regular fibonacci rom and here's my new loader rom so when this starts executing this is now in address zero nine zero zero zero so this is executing from ram so this is pretty cool in itself because prior to this we've never executed code from a ram chip apart from possibly attempting to do so by accident when things went wrong so this working is a good proof that we can do that i don't see any reason why we won't but it's kind of essential for the long term plans so now what this does is this takes the code at zero a zero zero zero and it performs a similar mem copy to put that which is the fibonacci code back down into address zero so this is copying the contents of the fibonacci rom into the new ram chip that we're wiring onto our hacky little piece of vero board and then it calls it to execute it so if this all works as planned what we should have is the fibonacci sequence code running entirely from ram without a rom chip in the memory module let's give that a try okay if you look closely you'll notice this has changed here slightly when i first tried to plug it back in i accidentally knocked the header i was using and actually tore the pads off so i had a little bit of a panic there but i have managed to reconnect that pin from the chip and wire this header back into the same location but this time i've glued it down so i can use it without breaking it but it's a good thing i'm replacing this board right let's give it a try okay so now it's copying the loader so we can see s-i and d-i here have radically different addresses now it's copying the fibonacci rom so these still have different addresses in this one the top bit isn't set and here the top bit is set because this is a rom address and this is a ram address you can just see the counter cycling down here till it finishes now it's prompting me to put the ram module in the second line along is the right enable line that we did this to get access to so now it's going to copy from the location it loaded the fibonacci sequence program into ram down into the new ram address 0 so in the same location it would be if it was running from rom so now we see source address in the high memory bank and the destination address in the low memory bank but that's ram as well now so when this cycle's down to zero it's going to start executing it that looks promising yes excellent this is completed correctly and given the right result okay from a circuitry perspective this probably isn't something you'll see as a big step forward this is really just a building block behind this video the work i did on the tool chain adding the functionality of the to the assembler necessary to support this new work was actually uh quite a lot of effort and i'm actually really pleased with how far forward that's moved now i think i've really got something i can write some more significant bits of code with next step here is to use the fact that we're now in entirely ram 64k of ram to build the circuit that will copy the contents of a rom chip into the ram at first power on and then we can write code that takes advantage of all of that ram and not have any restriction there's actually a big added advantage to that the ram chips we can get are much faster than the rom chips so this kind of system is actually used on a lot of early computers it's called shadow ram because the ram chips are so much faster than the rom chips it allows the processors to execute faster with less waiting around for the memory and the rough work i've done at the moment on the cpu in determining what's actually constraining the maximum clock rate includes the rom chip in the critical path so this will potentially open the door to increase the clock rate a little bit which i know a lot of you are quite keen to see i hope you found this interesting and i've been keenly watching the tracking data on the next batch of pcbs so i've actually got quite a lot of choice on what section of the circuit to work on next but i will hopefully be back quite soon alright goodbye