Transcript for:
Restoring the Plexus P20: Modern Revival of a Vintage Unix Workstation

well hello everyone and welcome back to adrianne's digital basement we're back on the plexus P20 this machine is a very rare dual processor multi-user Unix workstation from around 1985 and I've been on quite a journey trying to get this machine working this is part five in the series if you haven't seen the previous Parts I really recommend you go start from the beginning if you haven't seen any of them or at least catch up on the ones you missed because as usual this part five is going to be continuation where I left off in the last part to give a quick recap in part one I took this machine apart and we saw the Glorious huge motherboard in here and we took out the hard drive and I imaged it so I got a good image of that old hard drive which if you think about it is what like 40 years old at this point and yet it still works in part two I had to service the power supplies because there were Reas in there and plus there were some other faults so I got those serviced and I actually powered up the computer for the first time and After figuring out the pinout of the serial ports I was able to actually get some action on the boot ROM part three I hooked up the hard drive the original drive back to the system powered it on and the computer worked it freaking booted up into Unix I couldn't believe what I was seeing and I thought that was going to be the end and just be me playing around with the system but it quickly went wrong and the computer failed there were two faults that developed one of them was that the reset circuit that's inside the power supply had gone bad and was just keeping the computer in reset so the system would never actually boot up once I bypassed that it turned out there was a critical failure that the diagnostic ROM was detecting and it would prevent the computer from booting whatsoever and then that brought us to part four and in that part I talk a lot about the architecture of the system and the work that's been going on in the background on Discord on both the hard drive recovering data from that and the design of the motherboard and reverse engineering of the ROMs in order to fix that critical error in part three I had to know what was wrong with the Diagnostics what it was stopping on and the great people on the Discord channel that David set up that's David at usagi electric set up on his Discord server they were able to get to the bottom of exactly what was wrong with that error code why it was stopping and it turned out that it was actually a flip-flop chip that was bad had gone bad and it was right in the area of the battery leakage had happened from The Leaky battery so probably was related to that but it's amazing it worked for a little while and then it died that main flip-flop generated a signal that was necessary to allow this processor here which is the one that boots up the system to talk to the main memory and that signal was bad which meant that the processor could never communicate with the ram so once I swapped out that ship that fixed that problem but then we ended up with another issue the diagnostic said there was a failure in the clock circuit which is almost certainly the clock chip right here which is the same clock chip that's used on the IBM PCAT 5170 and it also said there was a multi-bus error multibus being the expansion chassis on the front of the system I fiddled around with the motherboard for a bit to see if I could get rid of that error but even power cycling and checking the connections and stuff like that it didn't go away it happens every single time so in this video I hope to get to the bottom of of both of these errors and get this system finally booted back into Unix once and for all and hopefully it stays working this time it doesn't just die after 20 minutes so without further Ado let's get right to it [Music] all right we have the patented talking hand over the motherboard and that's because this thing is so big it really doesn't fit on the bench so I got to do a lot of the work over here which is why there's a little oscilloscope sitting here and the multimeter all right so I want to address a couple things a few people pointed out that I probably needed to have the multibus expansion installed on the front in order for the system to boot without an error and that this board needed to be installed this board is only required if you're using the second power supply there's a connector right here that can go to the power supply and that feeds the pins that go to the back plane that Supply power to the other cards so you can either connect the power supply here or you can connect the power supply directly to the back plane CU it also has a power supply connector now when you look at the back plane cuz you can see the traces clear as day the power supply connector on the back plane also just goes to these same pins right here they don't feed the motherboard that the motherboard has no idea that power supply is there or not there it just feeds this card or obviously if you have any other multibus cards installed it will feed those with the requisite power I'm assuming normally you would plug the power supply into this if you're using this board just because you're not going to put any stress on the traces that are on that multiplane back plane the power that goes into this connector will just be turned into heat through all these resistors but obviously if you're not going to use multipl you don't even need this and you just unplug the power supply which is what I've done and that simplifies that whole process now back to this card cage on the front here the reason why it's actually on the system right now is because I just wanted to validate that it was not required for the system to boot properly but I know this actually wasn't so even though you see terminating resistors right there I had this thing booted up into Unix without this card installed so in part three this thing was never in the machine and it was definitely running Unix without any kind of issue and then we had the failure of the chip on the board my assumption is that this system came in two configurations one didn't have any of this multibus stuff this was missing this was missing and so was that second power supply so the motherboard obviously has to work properly without these installed because well that is the p50 configuration this is the P20 which means it came with this and this and that extra power supply and if you're going to run everything together then you just need to make sure the resistor board is in there so the power supply doesn't have any regulation issues with this all installed now back to the issues that this thing is having we have those two errors I think the first thing I need to do right now is power this thing back up because it's been a little bit of time since I last worked on it and let's make sure that it still has that same failure all right I have the key replacement the screwdriver it's like I stole the car here or something let's turn the computer on all right is it coming out of reset yes it is there it is so all the usual noises are happening self test revision 3.3 I'm assuming this thing will fail in exactly the same way now back in part three we had the problem where it couldn't talk to the ram this self test would fail very quickly and it would fail with that four F was it 4 f02 that was unable to talk to the ram but now there it is takes longer cuz I think these tests run sequentially and it gets to the multibus and it says it has a failure and then it gets to the clock and then it says a failure I think it's running a couple more tests there and yeah it told us the config value or config info is invalid what that's talking about is just like on a PC when you fire it up when there's no battery you'll get like a Coss check some failure you go into the cosos you configure it save it then everything is good it's telling us that the config info is invalid well of course it's invalid because the clock chip has no battery on it and those Motorola mc46 818 that's the chip that's in there when they're uninitialized because there's no battery on there when you power them up they just have a bunch of random garbage in there and it needs to be configured by the OS or the ROM or whatever and then with a battery obviously that gets saved or usually with those chips they stay running while the computer is powered up so as long as you don't turn it off you're actually going to have a valid configuration it's just once you shut it off it will lose that those uh settings anyhow I have validated the system is broken in exactly the same way it was in part four which is good good I don't like things when they spontaneously fix themselves and this actually brings me to the next awesome thing that the folks on Discord figured out about this system they have been consistently working on reverse engineering the ROM and they have figured out that there are a bunch of commands that you can do to do things while you're in this ROM even in the halted state that we're in right now so let me demonstrate some of the stuff that they have figured out one of the things that were figured out is that if you hit shift and one which is the exclamation point what that does is it exits out of the halt State and you can see we're at the plexus primary boot or Vision 1.2 prompt which is exactly where we were in part three before the system failed now if I hit enter it should load the Unix kernel and start booting the OS and yes that's exactly what it's doing it should print how much RAM it has available and then load the system and right here is where things go wrong the system is actually halted it's Frozen right now so we have a Unix 1.2 and the real available memory and if I recall from part three what happens next is it asks us to set the clock and I have a feeling it's probably trying to talk to the clock right now which I don't think is working because it says clock failed and that's probably why the system is hanging up right now so I'm going to use the old screwdriver here and let's reboot the computer by switching it into reset so there we go and out of reset okay that's going to boot back up that's my at tiny in there that's allowing that to happen now here's another cool thing that they figured out I have this terminal profile for putty here configured to send contrl H when you hit the backspace key and that was because when it was booted into Unix in part 3 that's what I found it needed for me to have the proper backspace but if you change it to this control h or decel 127 and I press the delete key it drops us into a menu called dma now dma as I mentioned in part four the last part that is the name of the processor that's running the Diagnostics right now the other one being job and it seems that there are a whole bunch of commands in here that you can do additional Diagnostics which we don't fully understand yet I did a bunch of button mashing in here I was playing around with this earlier and I found if you hit the T key it runs all the tests but it's going to do it in a way that is going to be verbose so it's actually going to give us info about what's running as opposed to when you first power on the machine it just says nothing until you get an actual there so you can see it tested the ROMs the static Ram the sios it tested the ram which was failing before right that was what we were stuck on in part three it's now testing the job processor the map the mapper the scuzzy here it's testing the multibus let's see what happens multibus passed but now it's testing the clock calendar and yeah check it out config checksum error 85 so that's the error of the config checkum I guess that's I don't know whatever a specific error code and then we get that final warning but all of the tests pass except the clock calendar so strangely the multibus tests passed but let's talk about what we're seeing here with this clock output so on this system it appears that the clock chip is assigned to d0 and then 000000 that's where the clock chip resides in memory the 68,000 is a memory mapped IO system it doesn't have separate IO address space like Intel systems so every type of peripheral on this system has to be mapped into the main memory space and it's a 16 megabyte memory space so you have to carve out little chunks for the very system scuzzy and the clock and you know whatever all the different stuff well this is where the clock chip is and the dma processor was obviously trying to read from the clock chip uh which is incidentally only an 8 bit chip and that's why it's skipping every other address here because of course this is a 16-bit processor well it's a 32-bit processor really but it's uh reading everything in 16bit chunks but since it's an only 8bit chip it's only going to happen on every other memory location so it was trying to read out all the various values inside the clock chip which is going to include like the time and a little bit of NV Ram that's in there and it received a one but it was expecting a two and received a two and it got a four and you know so on and so forth and you can like scroll through all this stuff here and it got back nothing of what it expected obviously there are some values in here that are going to be just garbage and that's obviously because the configuration is lost it's just garbage inside the chip but a bunch of these things will have consistent values because they're status bits and stuff like that and it's not getting anything back that it expects every single address gets back the wrong thing well that pretty much tells us one thing that tells us either the clock chip is bad or the logic that connects the clock chip to the CPU is bad because there's going to be some buffers and stuff like that or potentially the select Logic for the clock chip is bad as well because just like any other peripheral you need to turn it on and tell it hey I'm talking to you respond back please so it's got to be one of those things that is the problem now when it comes to the multibus error that we're seeing up here this multibus failure here notice that it comes before the clock test well I asked the folks on Discord to look at the flow of the diagnostic and I wanted to know does this ROM check the contents of the clock chip of the NV Ram in the clock chip that's the configuration data does it check it for multibus configuration like if you have a card in there that's a multibus card does it go to the clock chip and say you know hey load the config and maybe in there there's some values for whether you have multibus cards and that causes this multibus test to actually fail because it's reading back garbage that actually says that there's a multibus card which of course there isn't because there's no cards in there and that's going to fail and the reason why I think this is exactly what's happening is if we go back all the way to part two and we look at what happened with this machine on one of the times I turned it on it got a multi-bus error but it didn't get the clock failure but it got a multi-bus error and then it said config value invalid you know like the the check something was wrong with the config value well on the next reboot the multibus error went away I think what happens is the system automatically once it detects uh invalid configuration in the NV Ram reconfigures it probably sets it up to like a base value with like the B rate 9200 bod and no multibus cards and whatever stuff like that and then it should write that into the clock chip so on all subsequent reboots it shouldn't give you a multibus error but except it did on that first one because of the corrupt data well in this case it can't do that it detects that the config value is invalid but it can't write back to the chip cuz something's wrong with that part of the circuit right we've already determined that and therefore every time we power this thing on it's going to get a multibus error so I have a feeling that if I fix the clock circuit the multibus error will go away it will might happen one time but on the subsequent reboots it'll never come back of course I have no battery in this machine so powering it off might cause the multibus airor to come back and then you got to do one reboot cycle and then it should go away again or I can install a battery so that once it sets up the config values then it should go away my presumption is inside the operating system there's probably a command in there to configure the multibus cards for instance if you're adding like the eight Port serial card that's a multibus card to bring this up to 16 users there's probably something to configure that it writes into the NV Ram so that on subsequent boots it's able to test that card goes and checks what the contents of that nvram is it says oh you have one of those serial boards cool I'll do a little test on there good passes multibus good or or there's a problem multibus failure so that's my logic and my reasoning for why we're getting two of these failures right now and why I think they'll both go away once I fix the clock calendar chip before I go back over there and start poking around the clock chip to see what activity we see on the pins I'm just going to take a little bit of a tangent here remember I think it was in part two I was trying to run some commands that I saw in the ROM code and it looked like it was trying to load them off of tape and was course not working cuz I don't have the tape well what you see here on screen is that ROM code and we have various things like format mkfs restore file system check when I typed all these commands none of them actually worked well over on Discord a user transrail found some old code here for an older version of one of these plexus system but you can see here it has the plexus bootrom 1.0 and it seems that it's talking about using these commands but the interesting thing is these commands exist both on the hard drive and on the tape and when you just type mkfs it seems to try to load it off tape but for default but the also on the hard drive and I think they're on this hard drive too so this stand directory which I think stands for stand alone which means it doesn't require the rest of Unix to be running I think we can run these from the boot ROM there in the colon prompt so back in putty here the thing has failed so again we just push the exclamation point and that drops us to this colon prompt well I don't want to run the mkfs command that's make file system right but trans rail actually found a list from I think some source code for an old version of Unix all of the commands that we can see in the ROM I mean pretty much everything that's in the ROM is here except for I don't think Deon fig is actually there but the rest of this stuff is and this looks like it's actually using these commands here to move these onto tape right see it says tape colon and it's using DD in in file out fi over the tape but some of these are going to work so one of them is stand slh help and I know I tried to run that in part two and it didn't work but let's give that a try right here stand slel the hard drive is connected and spinning right now by the way and what's going to happen it's going to hang this what's going to happen I see the hard drive LED blinking it's oh that worked that freaking worked look at this display tape content so yeah this seems like this is a tape thing these things are actually on the hard drive they should be look at this file two a copy of the plexus system 5 Unix the dis formatting config program disc as in hard disk or floppy disc I don't know it doesn't really say construct a file system restore the file system check the consistency file system check you can use DD to make copies of stuff there's the file dump program which I guess will list the contents of a file interesting so we have du for disk usage LS and Cat type return for more information about this item okay just a few more files here file 20 the minimum file system required to boot plexus system 5 and file 21 a complete release of plexus system 5 in cpio format all right right so this tape contains the programs need to bring up software on a new system or restore the system so this is coming from the hard drive but what's cool about this is it's possible We Could reconstruct a tape image I don't have a tape to put in the drive but maybe with the blue scuzzy we can emulate the tape drive and then we can recreate it and actually be able to start from scratch and have a larger hard drive partition for instance you know if we use an emulated hard drive I mean there's all sorts of possibilities right type one followed by carriage return to display the menu or type two to display the tape Contents I think this is exactly what we've seen already oh look at this more information this menu display the tape contents recrea a system software Standalone programs I mean there's let's look at five here I want to know if I can format a floppy disc formatting the dis sparing tracks initializing the dis is also used to configure your logical disc please consult your manual okay I think this is for the hard drive let's look at number eight though disc format information for format oh look different hard drives uh what's kind of hilarious is it says 8 in and 14 inch the thing is the hard drive that's in here is a 5 and a quarter inch so I have a feeling this stuff is like left over from the old versions as far as I'm aware the p25 and the P40 those like the big heavy duty plexus systems that had a lot of cards in them and stuff like that they used eight and 14-inch hard drives and it seems like when they ported that stuff over to this they never fixed this so yeah clearly um this is inaccurate information so let's look at number four for Standalone programs here certain Standalone programs must be run before booting Unix they perform a number of tasks necessary to allow Unix to run on new system catastrophic System error results may require these programs to run okay so like you know making new file systems or checking the file system things like that and definitely the way the disk partitions are set up on this thing it doesn't really has a partition table but the format is Dev disk and then 0 S1 so I think it's scuzzy id0 and slice one slice two slice three things like that when we actually booted up the system in part three the user partition was being mounted and the root partition was being mounted I think it was like slice one and slice two or something like that and the next page here talks about The Logical partitions or volumes on the first default disc and then we have something called isbc discs are these Intel multibus drives is instead of dsk and then PD is the imsc drive and then zy logic I think people will probably know more about this than me but I have a feeling this is probably all leftover stuff from the old systems as well that definitely had a whole bunch of cards in them for accessing discs in different ways ah but look at this Dev RFP 0m is the floppy drive and then rpt 0m is the tape and then you can do it with no rewind all right okay so what we know for sure is that I can run these commands directly off the disc but they need to be in that stand directory and I just don't know if like LS and Cat and stuff is they would be on the tape if we had that but I don't think we do so how do we exit out of this I think we do minus one okay so let's try stand LS let's see if it's going to give us a directory listing okay well that's a bit odd we got two dollar signs and oh it works it freaking works so we're able to see the hard drive like the main drive here right so I guess what happens if we run this again is it's actually asking me to add any kind of uh command line parameters so if we want to look inside a directory cuz I do want to look in the stand directory here so there it is those are all the commands that we can run right from this boot menu so without even booty to OS these are all available for us cat whatever ccal is oh that might have something to do with the calendar let's try that stand oh oh no the backspace key is misconfigured it's not control h which it needs to be in here stand slash SE Cal I have a feeling this has to do with the clock calendar that would make sense I don't think it's going to tell us anything I mean the clock calendar is clearly not working as the diagnostic test showed us what the heck is this system diagnostic configuration rev10 diagnostic console oh does this let us set a different serial port to be used as the diagnostic console B rates oh this is it this is definitely how you configure the the clock calendar the defaults that are in there it's like the C setup for the system are we going to change the B rates um no we just leave those alone and this is not going to work anyways memory size 512k no it's 2.0 I don't know if we type 2.0 MB doesn't really matter right because it's going to figure this out autoo boot is set to no obviously we have set to yes when you power up the machine it's probably just going to do the slash Unix and just boot right in automatically model p15 all right maybe this is how you can figure whether it's got the multibus or not ICP okay so IP is the inte or Intelligent Communication processor that is what adds those extra serial ports if the system thinks it's a P20 and it has the ICP installed it's going to probably check those and result in the multibus error this is all confirming exactly what I was thinking about the clock calendar was causing the multibus error all right ACP is another communication process like asynchronous or something like that I I saw that in some document look at that ethernet does anyone have a multibus ethernet card can we get this thing on the freaking internet would that be cool or what clock speed what I'm pretty sure this thing runs at more than 10 MHz I think it runs at 12 A2 maybe this has something to do with the like timing Loops or something like that scuzzy controller zero in system I mean yes right it has a scuzzy controller scuzzy zero Lun zero is it hard drive tape or floppy or none well I know that the omt 5200 that's in there the hard drive is configured as Lun Zer the floppy driver is configured as Lun one so this is definitely HD I I'm typing this stuff it doesn't really matter matter obviously this is a floppy and then none for this I guess you just go through them this scuzzy controller one in the system it's funny how it calls it a controller but it's I guess it's talking about the actual Drive uh yes it is is scy one Lun zero a hard drive tape or floppy well I happen to know that is the tape drive and we just do none on these and there's nothing else is we have another scuzzy controller in here two four oh it's asking us all the IDS I I guess we have multiple controllers I could put more hard drives in this thing scuzi id7 isn't that the actual main motherboard but we'll just say no to that is all data correct yes update configuration yes I mean it's not going to work the clock chip is bad but it's so cool we can configure the system right here in the boot menu before booting the OS I I just didn't know how that would work and and now we do so let's do stand LS and actually what I wonder if I can just do that uh no okay stand LS all right I hear the hard drive doing stuff it's funny how long it takes just to do a simple LS but I guess because the OS is not running maybe the ROM code is very slow and inefficient but I know now I can type stand here and there we go we can see the other commands Okay so cats obvious C ccal is obviously the configuration core maybe a core dump dis usage file backup format file system check we already know the file system is good file system DB I don't know what that means help LS make file system OD which I think was the uh dump and then restore let's just try stand OD and let's see if that does anything I'm assuming it's going to ask for something to dump and I could try dumping the kernel which I think is just slash Unix yep SL Unix okay there it is look at that it's dumping the contents right all right that was a very fun diversion let me use my old screwdriver here and I'm going to shut the system down because it's quite noisy and the reason why I did that is because I think we need to do a little poking around to see how the circuit is designed that actually connects to the clock chip now going back to part four I had mentioned that we were able to recover some data off the hard drive and in that data was a net list and a parts list for this actual computer yeah it's completely mindboggling well in the diagnostic process of trying to find the bad flip-flop that was causing the ram not to work I had to go through and I have my papers over here I did a whole bunch of like manual work going back and forth and cross referencing all these different f files to figure out exactly what chips connect to what signals and how to like Trace them through to figure out where the signals were going wrong well it turns out the Discord channel has come through again and there's now a utility to do it for me I don't have to do it by hand anymore and it's going to come in super handy because we're going to do the same work on this clock chip to try to figure out how it connects to everything so what we have here is a python script called netlist and let's take a look at that real quick and it well it uh I don't know does a bunch of stuff it's actually not very long of a program but I'm not very good at at python whoops I don't know how I was editing there so we'll exit out of that and what it's looking at is this C.T file I think this is like an amalgamation of the data that was recovered off the hard drive in a way that be can be quickly parsed by this python script well I just took a look at the computer and the clock chip is part number u2h and all we have to do is run this command with the parameter of the chip that I'm trying to enumerate all the signals on and we do that and boom all that work I was doing manually is now done in mere seconds not even seconds in a second so that's the clock chip MC 14688 24 pin dip these are all the Nets now unfortunately a lot of these nets and that's just the way the files are they're not really named anything it's just net and then number as opposed to a name like ground or you know M reset but what this does it shows each of the pins that is connected to something and it shows which chip on which pin it goes to now for ground for instance it doesn't really show it and that's because if you look at the net list underground it's like a zillion chips on there so there's no point to look at that so anytime there's a plus 5 volts or VCC or ground it doesn't show it now this is a clock chip it doesn't run on the normal 5vt circuit it runs on a bat circuit bat and that's because that's normally hooked up to the clock battery because this chip needs to stay energized while the system is off also due to a little quirk in the way this thing runs anytime there's a pin that's not connected to anything instead of saying NC for no connection it doesn't list it so the reason why there's no pin one is because that chip must be just not connected to anything on this machine now a feature I had asked to be added to this and Peter who created this is working on it is to a enumerate if the pins are not connected to like print that so we don't you know Miss that but also to use a commonly available databases that are existing out there to tell us if the signal is input output or or both and also if it's a standard chip like a 74 LS 0000 to tell us what actually each pin is so there are databases out there that have that information readily available and for this clock chip for instance well we don't really know what any of this stuff is here not without looking up the data sheet so that one may never actually be look up aable but on a lot of other components they are commonly available in databases and it's just a handy timesaver to not have to go look that stuff up manually either way when we take a look at this a few things we can ascertain right away so M reset I happen to know this is the reset signal that goes into that chip from the reset circuit and where the power supply stopped working for reset that is where I connected the at tiny reset signal in I actually soldered onto the chip right on the clock chip on pin 18 cuz I knew that was the input for the reset circuit and I happen to have cut the trace on the back of the motherboard that goes to the power supply cuz the power supply circuit doesn't work so I just cut that Trace so that signal is reset active low and then the bat signal we already know that that's the battery and this signal right here cal. CE calendar chip enable I'm assuming and whenever it has a little Dash after it it's the older convention but that means that that is an active low signal on a more modern design you might write a slal doce or cal. CE with a line over it all of those things mean the same thing it just means that signal is active when it's low if that signal doesn't go low that means that ship is never going to communicate on the bus that goes the same for ROMs or anything else you're trying to troubleshoot when there's an enabl problem now when we take a look at these other pins here four five 6 7even this this seems a lot like data lines or something like that so let's just take a look at one of these other chips here so all we have to do is just run this again with this other chip oh this is LS 244 so yeah this must be a data line and in fact uh what was the net I don't know the net we were on was like 734 or 732 or whatever and 734 yeah okay so see this so it has ba dma so dma is the processor that's running the Diagnostics ba is buffered address that's what I'm assuming that is and I happen to know this chip uses multiplexed addressing and data lines that means that the eight lines both handle addressing and data so if we do this other chip here oh that's a lot of stuff on here this is going to be data yep BD dma so buffered data dma 7 6543 21 etc etc but there is data line zero so we have seven all the way 6543 2 1 Z yep okay so we know for sure going back without even checking the data sheet that almost certainly these signals right here are going to be all the data and address lines so we know for sure if one of these chips is LS 244 or the ls 245 aren't being selected correctly then the clock chip will never be able to talk on the data bus I think let's not get ahead of ourselves first thing you always need to check if you're trying to figure out why something's not communicating on the bus a chip that is is always a chip enable line so let's take a look at Pin 13 I'll get the AOL scope hooked up the scrope probe goes to the Zoe ZT 703s whatever this thing is a few people have asked me about this like oh what am I doing using this thing I bought one of these on a whim this thing is cheap it was like 70 bucks shipped it's a decent oscilloscope it's actually not bad I don't love it but I don't hate it either and I I like the form factor and it it's an okay multimeter and after some firmware Updates this thing has gotten this is actually not a terrible device now the reason why I did not review this is there are so many other reviews of this and by the time I got it like 50 other channels already done it and I thought what's the point but if people want to see me review this then I can but ultimately it's a decent device if you need a cheap $7 oscilloscope and meter it's not bad USBC updates and charging and the screen is big it's quite visible and the update rates are decent the UI is frustrating at times but not that bad I like it it's not as good in cell scope as the hand Tech portable scope I have but the UI is better on this and the screen is better and my eyes aren't that good anymore anyhow it's hooked up to the chip here and when I power this on the chip select line on here needs to be at zero volts or around there when the chip is being accessed at the minimum so let's turn this on I have the little plastic tool I'm going to use okay it went up to 5 volts so that's okay it's like it only needs to go low while the chip is being accessed if it's not being accessed it's fine it's up at 5 volts now over on the screen over there I can see the Diagnostics running it just says self test 3.3 and we'll keep an eye on this right here and see if it goes low and nope over on the console I can see that we have the self test failure now for the multi buzz and the clock and this never budged at all now even though the update rate seems a little bit slow right now I happen to know while this is triggered it actually does update much more quickly so if we had a pulse where it went low we would have seen that it would show up on this thing that's one of things I don't like about is I don't like how the update rate is so slow on this when it's just sitting here not triggered uh it's just in the Run position but anyways it would have shown something and we got nothing and that means without a doubt there is no way let's turn this off cuz it is so noisy there's no way that that clock chip can work if it's not selected so I need to go back to the computer over there and let's figure out exactly what that thing is connected to and see what selects it and then we'll fix it all right so back with the python script we are looking for this line here it's going to two things a resistor and a transistor let's start with r47 and let's see what that goes to so r47 is on the Cal C okay that goes to the bat signal battery I happen to know cuz I looked at this when I was putting on the little at tiny the bat signal is just the 5vt rail through a diode because of course it needs to either charge the battery with 5 volts or it needs to power up the clock chip with 5 volts or not 5 volts whatever the original battery was probably like 3.6 or something like that so if we take a look here designator bat not found did I type it wrong I don't know bat well okay well anyways it's right there and it's enumerating it goes to various things so we know there cr1 that's the diode that then goes to the 5vt rail in fact I think if we do this with cr1 we'll see it's going to VCC yes there it is VCC and bat so that's uh there's a little bit of a voltage drop through there so the VCC becomes you know whatever6 volts less but that does work because when I was probing around to install the at tiny I almost connected to this bat signal which would have kept the at tiny running all the time which it wasn't necessary and I I decided to go not onto the clock chip I actually took 5 Vols off a chip that was right below it I'll write down the signals we need to check we need to just make sure that bat does have that 4 point whatever volts on it because the clock chip isn't going to work without it but it looks like the chip select circuit obviously is connected to it as well so we just need to make sure that that looks good all right the other component was Q2 so we have a transistor there so let's take a look at that all right there it is it's connected to ground it's connected to the chips sleck line and it's connected to two other resistors so R51 one let's see what that is okay that seems to go to the reset signal that goes to my at tiny all right and that the reset signal goes off to various things that is fine and then what's r52 r52 that just simply goes to ground so one of the things we can tell for sure here is that none of these nets connect to logic chips none of them at all well the only one that was connected to anything was this m reset but we know that the chip needs to be enabled when the computer is not in reset so assume it's connected to that something to do with it like I don't know maybe it's a power saving thing or something like that I'm going to think about this a little bit but yeah it's definitely not what I expected because I would expect it to be coming from a logic chip so anytime the CPU tried to talk to the click chip it would be chip selected otherwise you know it would be idle but I guess that's not the case and actually if we think about it for a second remember that the data and address lines for the clock chip are connected through buffers LS 244s and 245s that means that you could actually keep the clock ch selected all the time and that wouldn't have any negative effect on the rest of the system because obviously there's logic that activates these this LS 244 or the ls 245 to connect the data or the address line to the clock chip so maybe the clock chip is just enabled all the time something like that I don't know let me go have and think on this all right well I was over on Discord chatting with super sbga who helped me before with a bunch of stuff in the previous parts and super SVGA came up with this simulation of the circuit and we took the values of the resistors that I actually saw on the board because they actually were different even though it said 10K if we looked at the uh parts list here uh wherever it was down here yeah 10K and 10K I looked at the board and it actually had 520 ohm resistors here and here it looked like what this circuit is doing and super vest VGA kind of talked me through it a little bit it looks like as I thought the clock chip is either selected all the time while the system is running or it's not selected when the system is off and this is probably some kind of a power saving strategy when the system is running on battery probably the clock chip shouldn't be selected and if you look at this circuit let's just talk through a little bit so here's the 5volt battery signal right and right now the transistor is not turned on if this transistor turns on what happens is this line through this resistor gets connected to ground and actually will pull the Cal C the chips sck line down to 5 volts right now you can see right down below I can't point at it cuz if I move my mouse or no uh no if I move my mouse it goes off of it the voltage of this signal is 5 volts and that is when the reset line is not at 5 volts so while the system is in reset which means the reset line is down around Zer volts or the system is just off altogether which means the reset line is down around 0 volts then the battery signal is just routed directly to the Cal C line through a 520 Ohm resistor when the system comes out of reset which means that the at tiny pulls the reset line up to 5 volts then you can see we have electrons flowing through here and the Cal C line now becomes 68 molts so that is equivalent of a low signal which enables the chip select line on the clock chip clearly the circuit is not working properly when I power on the computer the system is actually in reset for a couple seconds that's what the at tiny does so this calc line should be at 5 volts initially for a couple seconds and then drop down to zero volts well it never drops right and when the at tiny takes system out of Reset 5 volts is available on R51 here and that should bring this down around 5 volts and that's just not happening now when it comes to failed components obviously if this resistor were missing well that's all it's going to be up there at 5 volts no matter what and if this resistor was missing now according to the simulation even with that resistor open still works as desired at 5 volts the chip would be enabled so it's not that resistor and I don't think it's this resistor because they would be at Zer volts all the time so the only possible candidates really are this resistor here this 2K R51 or the transistor being bad or the clock chip I suppose the clock chip signal could be bad internally and it's forcing this transistor up to 5 volts although if that were the case that would be shorting through the transistor to ground and the transistor would have got hot and died already so I think need to go over there and check a few things I need to check that the base on the transistor is 650 Mills or whatever it says there I need to check that the emitter is actually connected to ground because if it wasn't if we take away this ground then it's always going to be stuck at 5 volts so I need to check that it's actually grounded and I guess um I think that's it I don't know what other things could cause this to go wrong so yeah I need to check that resistor well I don't even to check that resistor because this is a voltage divider right here all I need to do is just check the voltage on the base to make sure that that's there and just make sure that the emitter is actually connected and I suppose I'll check to make sure the collector is also connected to here because If This Were Somehow broken like a tracer was broken here and this resistor just went right to Cal C and that the transistor wasn't actually connected then that would mean that it would also be held at 5 volts all the time okay I know what to do I'm going to go do that right now all right I am back and um well I don't really see anything wrong actually so the base is around 6 50 Ms that's what I saw on the multimeter and then the collector was connected to the chip itself I actually have full continuity between there and there and the emitter is definitely connected to ground so that really just leaves the transistor as the only bad part that could be bad here now this whole circuit seems to be something to do with power saving as I said and one easy way to test without even taking this out is I'm just going to ground the collector here or just ground the Cal line Al together that shouldn't cause c a problem it's not going to cause a short cuz this resistor is here 520 ohms it's going to waste a little bit of current but it's no big deal this is just a test if this does work then I'll just fix the circuit I have to take the board out though and desolder it so let's go do that right now so what I've done is I clipped onto the chip select line and I hooked up a 100 ohm resistor and that goes to ground and this thing should be constantly enabled now so with the asilis scope hooked up let's power on the machine and yep indeed it's down at Ground now so this is working perfectly all right well the system booted and that's it that's the boot we just had and yet it's still telling us that we have a clock failure in the same error as well so let's go in here and let's run that clock test now I think by the way I can run that test yeah the H command lets me run it independently run clock a TA I think is what I type no we are still having issues I've been poking around the clock chip I don't see anything wrong all the signals like there's a address strobe and a data strobe signal those are good read write looks good as well in fact when I run the Diagnostics the TA you know where it lists all the memory addresses it's getting wrong I can see it pulsing everything looks like it's working now the question is why do we not have a working clock calendar my feeling is that I must have somehow damaged that clock chip when I was installing the at tiny I guess I don't know what else it could be although in part four when I was using the Arduino to boot this thing I hadn't yet soldered anything onto the clock chip which is where the at tiny is and I just don't remember if I actually got the system booted or not I'm going to have to go review my footage to see if I ever let the thing finish all its Diagnostics and try to boot the system maybe I did maybe I didn't well you know to be honest now that I think about it I guess it's kind of irrelevant the system is not working right now even though it looks like it should so I think what I need to do is I'm going to socket the clock chip and that will allow me to swap a different one in there I'll just take the at tiny off I I have I have extras of those chips I'll swap a different one in I could even use a Dallas chip in this thing and that that should work so I'm going to go do that right now okay we're ready for a test this is the chip that was in there I've removed it and I have a a bunch of chips here spares Dallas is but other 146 818s that are compatible so there's a attache part in the socket there now I have the at tiny removed right now so I'm going to have to manually bring the computer out of reset all right here we go powering up computer's in reset as it should be and we'll bring it out reset now here we go oh one other thing for that whole circuit with the transistor I decided to remove one of the resistors R51 which if we take a look at the list here that says it was a 10K or was this one here but it was actually like a 510 Ohm resistor so I remove that and that actually renders the whole transistor circuit in operative so I just put a little jumper link on the back side of the board to bypass the transistor entirely look look the clock has not failed so we're just getting a multibus failure which is exactly what I would expect to happen let me grab the keyboard here my hypothesis is if I reset the computer we will not get that error anymore but let's go into the dma mode here and let's do ta this is a clock test P freaking past and T9 is the multibus test specifically that's going to pass right right okay so wait I this chip was bad I mean it was working I mean it was close to the battery but it didn't look crusty but I did solder onto the legs so maybe I damaged it by doing that I guess I don't know okay so to test my hypothesis about the multibus failure I'm going to disconnect the resistor which is going to put the computer in reset reconnect it and we're going to reboot and if my hunch is correct we're not going to get the multibus failure anymore so the computer's in reset and now it's out of reset now if we do get a multibus failure it's not that big a deal because we know how to run the clock calendar setup utility now and I'm just thinking here so I think what I'm going to do is I think I'm going to install one of these Dallas chips I just have to do the little Dremel trick to put a new battery on it cuz these are all dead I have uh how many do I have I have three of them here the kind that will work in this machine oh we still got the multibus failure interesting okay well I have the hard drive stuff disconnected now I'm going to power the machine down reconnect the hard drive and the scuzi and let's see if I run that setup utility if that fixes this problem all right drive is reconnected to power scu's reconnected let's power the machine up and I'm going to clip the resistor on to bring it out a reset there we go I can hear the drive spinning up all right multibus failure yes we know wow clock chip died as well I'm still bit fiberg assed about that okay diagnostic console change the B rates I'm just going to leave that as is we have 2.0 megabytes autoboot no let's put it to P20 since that's what this is but we do not have the ICP or the ACP or ethernet clock speed should be 12.5 as far as I'm aware scuzzy controller zero is in the system yes it's a hard drive then we have a floppy then we have nothing scy controller one is the tape drive so that's TP I mean it doesn't really matter does it these other lens aren't used and these other scuzzy controllers are not used where is three it's funny how it skips over three all data correct I'm going to say yes let's update the clock calendar okay let me reboot it again by taking the resistor off all right system is booting back up if all goes well we won't get the multibus error anymore I'm keeping my fingers crossed that that's the case it's just corrupt data oh in incidentally when I was poking around on the bad clock chip and I had the oscilloscope connected without a doubt I could see the clock chip being accessed early on in the boot process when you first power on the machine this chip is read a bunch of times and that's even before it gets to the clock test so that means it is reading configuration stuff oh clock failed again no no no is that error different a007 oh you got to be kidding me What's happen happening here okay before I jump to conclusions I'm not going to worry because the way I'm doing the reset with the clip leads is not not ideal I think ultimately glitching the reset line is probably not a good idea you know using the clip lead that's going to cause all sorts of weird issues and okay yeah we lost the config info okay whatever let's boot the system completely so I hit exclamation point and I'm just going to push enter and hopefully the system boots uh hard drive light is Blinky Blinky and let's see if it boots before it was freezing right after it would show the free memory I think because the clock chip wasn't working so fingers crossed this actually works oh yeah that was it it's working now so the bad clock chip keeps the system from booting oh look at that so it remembers the last time the system booted that's the time I typed in and remember the first time we booted the system in part three it was 19 1989 so that confirms that was the last time this system was working is the date correct I'm just going to say yes just to skip over this do you want to check the file system um I think we're going to say yes because the system did crash and I'll let this run through the check that was it was crashing in part three when it was getting those errors right I feel a big sense of relief here though that yeah the clock chip went bad and that that caused this thing not to work so I think what I'm going to do is take this Dallas chip here I'm going to put the uh 2032 mod on it use a Dremel to kind of eate the case a little bit and then I will stick this in there and this should work in place of that and then we should have a clock calendar that actually is able to save the settings but the system is booting up here which is excellent there we go we're ready for login root and there happens to be no root password on here we did learn that and there we go we're freaking logged in now I'm curi ious if there's a reboot command so I should be able to run that ccal Command right here as well yeah um oh cor dumped okay you cannot run that command while you're in the operating system I guess these stand commands do not work uh while the OS is running they do not and what I'm going to do now is modify the Dallas chip and I will reinstall the at tiny so we have a proper working reset circuit and hopefully hopefully after that the system is fully operational okay the motherboard has been modified I actually put the at tiny On The Backs side and I have my Dallas chip all modified here with a CR 2032 the one thing I forgot to mention is the original Motorola chip that I took off the board actually has an auto detection between Intel and Motorola mode but all the later versions of that chip don't have that so for the Dallas chip to work you actually have to put a 10K or a 4.7k between VCC and the sense pin which is pin one it's Motorola inel mode change pin so to use this Dallas chip though I do need to hook that pin up to something and I hooked it up with a resistor on the back side of the board as well this is exactly the same thing I had to do on my BBC Master 128 when I installed the replacement clock chip because I think it also used the same exact Motorola chip in there originally and I put the necrow Weare replacement chip in there instead of one of these Dallas chips and that one also required pin one be pulled up to 5 volts otherwise it wouldn't work quite right so I'm going to stick the Dallas chip in here with the battery let's see what happens and I'm ready to steal this thing I mean uh turn it on with a screwdriver here we go and I won't need to do the reset thing anymore should come out of reset there it is it came out of reset on its own now incidentally when I had the Hitachi chip in there that was the replacement chip that I put in instead of the Motorola chip I think it also needed that resistor on the backside probably explains why it wasn't quite working with the multibus we were still getting that error I think it was because of that although I guess we're going to find out shortly if that was actually the case tape dve making all sorts of Noise Okay multibus failed as long as we don't get the clock failure no clock failure that's good all right config invalid totally understandable so what we're going to do is we're going to run that stand CC config program not CC config ccal this will run off the scuzzy drive all right diagnostic consult B rates nope memory size 2.0 megab Auto boot no I'm just going to leave this as p15 for now just so it doesn't even do a multi uh multibus test anymore uh definitely scuzzy controller hard drive floppy and the rest is nothing and I'm actually going to say no for the tape drive because I actually don't know if the tape drive is id1 or or not so we'll just skip past all this stuff here all data correct yes update configuration yes I'm going to do the old reset here with the screwdriver let's see what happens okay there we go system is rebooting so theoretically we should not get the multibus error hopefully and before when I didn't have the resistor installed when we did this I got that proc error and a clock error and I think that was because the resistor was missing let's see what happens I kind of remember on the BBC Master when I didn't have the resistor in place there was a bunch of garbage that came back from the clock chip when you didn't have the resistor in place so let's see what happens I think we're golden everyone I think we're freaking golden because uh it's doing a parody check right now oh no proc failed interesting procs mean I wonder so we didn't get the clock failure we didn't get the multibus failure well we can run that utility again stand CC Cal oops that's not it stand ccal but look it definitely retains settings 2 megabytes Auto boot no so let's P20 you know what before I press enter here I'm going to hook the osilloscope up to the clock pin on the 6810 I've never done that so let's see what it actually is running at all right yeah it's running at exactly 10 MHz at least uh well at least the processor I I probed there so we'll just leave that as 10 MHz and uh we'll just go through these I'll just hit enter since it seemed to have retained everything all data correct yes update information yes we will I'll reset the computer again there we go so interesting is the multibus error it went away but then we got the procs error what does that mean what is procs so while we wait it's kind of cool that the Dallas chip seems to be working definitely saved the settings and the battery on there should keep it going for you know years and years and years it's a very efficient ship the motor will need that giant nikad battery to retain the settings for any reasonable length of time but the one that the Dallas has there's that proc error again I guess I should power the machine off and back on and see if that goes away I'll just let the machine cool down for a moment I guess this is a real test as well is the battery working I used a little mini Dremel here to cut away at the side of it and then my little engraver tool to cut through the pin I like to personally cut the battery off that's internal that's epoxied in there I like to cut the little metal part that goes from the chip to that battery that way the battery I put on there the CR2 32 is not trying to like back charge the little whatever coin sell that's inside of there okay come on no procs error no freaking procs error no proc fail let's go in here and let's run the tests again I think I just hit T this will run the entire gamut of Diagnostics here the funny thing about that proc airor is when the system booted up before like when the clock was corrupted we didn't get it and now we are so it feels like it's something to do with the config settings that I set up there because the error appeared after I set up the clock calendar with that utility okay wait look at this procs sc7 not expected failed sc7 ah is that scuzzy scuzzy controller 7 and I hit no let's fix that I'm going to reboot the system and I'm going to put yes on seven and see if that does anything all right hit yes and it's asking what it is I don't know what it is I don't know I'm just going to put HD I I don't know what else to put there and we'll say yes update config sure and I've escaped right to the dma menu here and I think if we do H we can just run that particular test TF no same thing oh different error code though look at that F22 and before we were getting F35 now I notice if I hit Q aha look at this system configuration sc0 sc7 is the tape drive sc7 I mean the tape drive is a separate scuzzi device it's not part of the interposer which is definitely id0 so sc7 it's got to be the tape drive okay I'm going to put yes and we're going to do tape TP and there's nothing else on that one is this correct yes it is update config yes reboot the machine over there and let's see if this does the trick it's quite the learning experience it was interesting to learn that in the dma menu Q gives us the system configuration and normally when the Diagnostics pass we will see this exact message I think it just kind of tells you like what it's found and let's see if that that does the trick you know what the fact is how it skips scuzi ID3 I think that kind of tells us that the computer itself is scuzzy ID3 that's unusual I don't think I've run across that in my in my entire career oh yes that was it that was freaking it this thing is working now hit enter and we're going to boot into Unix quite the learning experience but you know what the clock is good I can turn this off and tomorrow it will freaking remember all settings because that battery is on there that is awesome we ended up in single user mode so I think we could do that and this should put us yeah it should oh look at the do the time oh it's because it's corrupted and it's reading it back from the actual clock let's check the file systems one more time this thing is working question is is it going to stay working or is it going to fail in other spectacular way I don't know all right there we are it's logged in it freaking is working again oh that is just so great what I should do is try to do a shutdown here shutdown program system being shut down right now so that was a broadcast to all other users on the system I'm going to turn this off tonight I'm going to go to bed and then we will see in the morning if this thing boots up properly without any errors I really hope I really hope this thing works I'll see you in the morning it's been a couple days since I took out the bad clock chip the Dallas has been in there with the battery the whole time and I think it's now time to test this thing did the Dallas clock chip retain all my settings and will we actually get a boot with absolutely zero errors guess we're going to have to find out I have a lot of confidence that this is going to work so I'm going to reinstall the multibus card cage Plus board and put the cover back on this thing cuz I want this thing to have good ventilation and if it does work I want to let it run for a while kind of stress test it let the air circulate inside here and just make sure this thing is actually stable I only got about 30 minutes or 20 minutes out of it the first time this booted up and worked so hopefully we can break that record and maybe get a few hours out of this thing and the plexus is all back together the one thing I ran into is that Dallas clock chip sticks out a little far and it actually caught the lip of the case as I was sliding on and it knocked it out of the socket it actually clears the side of the case without an issue and it even clears the lip you just have to make sure when you slide this on you don't sort of tilt It One Direction which just makes it a little too close I'll have to put a note on the back that says when we moving case be cautious of the battery on this side just to remind myself not to knock that out next time all right let's power up the Beast here we go all right we don't see the plexus Banner because I had to restart putty when I first turn this on I always forget that when my lab machine goes to sleep for whatever reason putty loses connection with the comport hopefully it is actually working and one thing you can tell is oh wait look yes it's freaking working all right so Auto boot is turned off all I have to do here is hit enter by the way system 5.2 what does that mean h i don't know there we go yep run level two automatically is the time correct June 4th yeah that's it 131 that is literally what time it is right now ignore the year being wrong that's a problem with the clock set utility time is correct should let's check the file systems again I mean we don't really need to but may as well this thing is actually freaking working wow there we go I just broadcast it freaking Works to all the other users on the machine which is no one right now this thing is actually fixed at least it's fixed for now part of me can't really believe that this computer is actually working so I think what I'm going to do now is just let this thing sit here boot it up into Unix for a while and make sure that it doesn't fail I wanted to at least run for let's just say 3 hours I'm just going to pick that amount of time arbitrarily and hopefully it stays working and then we can consider this thing fixed at least fixed for now and it's been a few hours since I powered this up and the plexus P20 has been working perfectly well not quite perfectly I think I figured out there's something going on with a floppy drive on here I don't think the spindle is spinning and anytime you try to access the floppy Drive what happens is the interposer gets confused and the floppy LED stays on and then the hard drive is no longer accessible so it actually crashes the system well it doesn't crash it just halts the whole system cuz even if you try to go to a second serial port and log into another console it can't do it because the hard drive is inaccessible it's probably one of the negatives of using the interposer that handles the floppy drive and the hard drive at the same time I think what's going on is because the drive is not spinning there's no index pulse going to the interposer that must confuse it so when it comes to future plans for this machine now that it's working there's still Lots I want to do first off we have to finish sanitizing the hard drive image that we've created of this thing that way I can upload that to GitHub so people can take a look at it I've had a lot of requests for getting access to the hard drive image so hopefully by the time this video is out I will have already been able to do that and it will be uploaded and I'll link below in the description to that image I also obviously need to fix the floppy drive because that's not working so take that out see if I can get that working I'll do it externally on a PC also I want to get the blue scuzzy working working with this machine I talked to Eric who's one of the lead maintainers for the blue scuzzi project and he gave me some tips and pointers on configuration options to use because this thing obviously is expecting the onty 5200 in there for the hard drive to work so in a future video I'm going to try that out cuz I need to let Eric know if there's any fixes that need to be made to the source for the blue scuzzi to add this weird rare computer to the compatibility list I'd also like to try to figure out how to get files on and off this computer the floppy Drive obviously if that were working would be one possible way or potentially copying Kermit source code or Z modem source code and then transferring files over the serial Port but right now this seems to have a uucp mechanism for transferring mail between various systems over the serial ports but I think I'd like to get something a little bit more direct like the floppy drive or some other way to get stuff on and off of this if the blue scuzzi works I can obviously take the SD card out and put that in a mon computer and then hopefully we can figure out how to mount those images on something modern to copy files on and off this as well anyhow as you can see there's still lots more to do with this machine I might need to take a little bit of a break though and work on some other stuff for a while it's been sucking up all my time this beast but I do like it and I'm really excited that the thing is working and I'm definitely going to do some more stress testing of this thing and let it run for even longer uh than just a few hours just to make sure that nothing else is ready to break on this thing but so far so good 100% stable and the last thing I want to add is I do do apologize for this video I know is a little bit all over the place especially while I was working on the clock circuit I edited some of the video already before I've done this outro and I noticed uh I don't know it's not up to my usual standard so hopefully you liked it if you did thumbs up if you didn't you know what to do a huge thanks again to everyone over on Discord who has been helping me with this thing I really don't think I would have been able to get this working or it would have taken me way longer if it wasn't for the help that everyone over there has been giving me and also a huge thanks to the folks who were helping me with the disc image and recovering the data off this because obviously that hidden data that was on that lost partition was the other part of this machine that was absolutely essential for getting it working so that's going to be it huge thanks to my patrons names are going up side the screen they get early access to videos they make it possible to do this fulltime so yeah big thanks to them thumbs up subscribe usual YouTube junk and that's going to be that so stay healthy stay safe and I'll see you next time bye-bye [Music]