Usagi's discord crew are seriously some of the smartest retro-tech people in the world when it comes to these 70s and 80s minis and server machines. Absolute brilliance and it's so good that they are around helping people like you preserve the less well documented aspects of our digital toddlerhood. Great stuff Adrian!
That is why an open source is so important. To have a place to publish when the project is over. But it needs to be in the culture or otherwise all the legal people complications
@@oidpolar6302 You are right. But it is not the terminology or ideology that is important, but the people. I do cracking and piracy since the Commodore PET times (which was of course voluntary work for the kids back in the time, who never could have payed the astronomical price-ideas of the corporations and growing game publishers (they and capitalism spreads like the plague)) and voluntarily work in the open source community since I was a kid. We live in great times now. But my fear is that this will change. After the next mayor privacy scandals the mood may change and also the legislation ... which may make the illegal and unethical handling of data (of strangers!) from big-corp and start-ups very unattractive and threatened with prison. The same inevitably will happen when those embarrassing hypes, like the K.I. nonsense implode or (what already happened) the "crypto-bro's" make new friends and opportunities in prison showers. Then free services like open repositories and free storage, free websites, sponsored projects and the like will have a hard time. Anyways, at any point in the future we have to ask and answer those questions. A free society (including free software or voluntary/free work on software) cannot be based on unethical, immoral concepts. Those corporations today, who provide this "free" storage will not hesitate to delete the data. Even the recovery of such data will be hard ... BECAUSE MONEY IS INVOLVED! So be aware that if we do not take precautions for the future (which will cost our OWN MONEY!), all that wonderful work, data, acquired and with great effort renewed knowledge ... may be lost! This happened in the past and happens right now to a rate most people do not even wanna know or believe. Not only with printed or (older) historical data, work and art, but right now ... with all kinds of different media. F.E. Good luck with QIC-80 tapes!:) Take that with a grain of salt and please not as fear-mongering or politics-critique (there is and was nothing better than capitalism and democracy. who says otherwise can p___ off to this wonderful North-Korea:P). Just to be aware that Google, Start-Ups, Tech-Bros and especially Elon Musk are NO ALTRUISTS! Have a good one, and don't let "them" catch you ... Oidpolar, exidy and community!:)
I used to service Plexus machines in the 80s. I never came across a P/20 though. P/30, P/35, P/55 and P/75, but not a P/20. They were quite good machines. Very reliable.
I'd still really like to see a diagnosis of the power supply reset circuit. One of your other commenters and I on the last video had a debate about whether your fix was potentially bypassing a potentially more serious problem in the power supply That is actively happening, or if it wasn't that big a deal and your reset circuit wasn't an issue. To that end, I'd love to see the power supply revisited and rescoped out both to check the voltages again as the other user suggested should be done, and to see if we can get an understanding of The reset circuit and how it monitors the power. It looked like a complicated circuit and it would be interesting to see what it's responding to or if it's just broken.
I think the reason /stand commands didn't run in Unix is that they are not compiled for Unix but compiled directly for the CPU. They probably try to talk directly to the peripherals rather than speaking to the kernel. Sort of like an ATTiny doesn't have an operating system, it just runs the raw code. So even if the program could execute, it might try to access a protected memory space in Unix since it thinks it has full control of the memory map, Unix might intercept that and force the program to dump and exit.
/stand is for munix (mini-unix) and sash (standalone shell). There is a complete set of man2 files with .2s suffix. Among them, srcheof.2s is most fancy since it's designed just for installer tape head positioning. :-)
I was suprised about how different is from other contemporary Unixes (AIX, SCO, IRIX...). They usually use the "bios" only to setup whatever devices you have and load the kernel from there. After that, they usually had a tape or disk partition for statically linked binaries mounted in memory on /sbin to be used like this /stand. This is quite cool, but I guess one needs to maintain those tools separatelly for each hardware which is not as practical.
I had zero interest in this plexus machine because I thought "what can you possibly do with it with no documentation" until I heard what was on it. That was a game changer! Well done to you and the discord team. That is a once in a lifetime score.
It has been quite the roller-coaster ride of a series for the P/20. So awesome that the community real came through to help out making this such a successful adventure!
Such a lovely series, it was very exciting detectives work! Thank you for not giving up and sharing. Also thanks a lot to the community and all the help.
Unix System V svr2 commercial source license was for $43,000, with three months of support, and a $16,000 price per additional CPU…. That’s some expensive system you just resurrected !
Considering the small customer base of hardware like this and the level of support required, this machine might have cost several 100.000 USD. All to be thrown into a landfill after a few years of use.
If it works like the SGI 4D/20 (another SysV based computer), you only needed a source license (which was, indeed, expensive) if you were planning to recompile the operating system itself. If you didn't get the source license but did get a compiler, you still got all the necessary library and header files for compiling other programs, just not the actual kernel code. (I'm pretty sure you got at least some of the libraries even if you DIDN'T get a compiler option, because they were the same DLL files the OS itself needed to run.)
This has been one of my favourite series on the channel. Congratulations on getting it running! The community really came through with this one. Love it!
The com port loses connection because in balanced power mode unused devices are shut down when the machine goes to sleep mode. You can change individual power saving settings for devices though to stop it from putting it to sleep due to inactivity. Putty also has an option for how often it should send keep alive and on lower baud rates that often needs to be increased so the emu can keep signal sync. I know it's convenient, but sleep mode isn't the greatest option for hardware. Low c states still have current bleed issues even on modern CPUs... yeah be a cold day in hell before that ever gets fixed. Congrats on getting the old girl running :)
the transistor circuit that un select cal.ce- when in reset / power down is probably against content corruption of the registers; The bus is in a unknown state and stray voltages/signals may be present on the bus that could cause a random write to one of the registers is the chip is selected.
I find it so interesting that these old unix device names still are part of being used today. If you use a LTO tape drive, the drive is like /dev/st0 and the "no rewind" function is /dev/nst0, notice the extra "n". And that is found in the device names here as well, it's simply fascinating :)
This series was awesome. Bringing in other communities and even having NCommander give it a go, really made this feel more like a on going event. Great work all around.
My car broke down. Luckily my car particular had the engineer and mechanic who designed it stored in a secret compartment in the glove box. With both of them free of the carbonite we were able to get it going again.
@32:00 I reviewed the scope meter, it does pretty well, the multimeter is actually very accurate, I tested it on my calibrator and reference standards, like any of these devices the scope is compromised a bit, but is fine for a lot of work.
I had no idea what you were talking about 95% of the time but I really enjoyed this series nonetheless. I get enjoyment out of seeing people talk about things they are deeply knowledgable about.
That Reset inverter circuit may exist to provide spurious write lock-out while the system is powering up or down. Tying /CS low, permanently might cause Config memory corruption.
My experience with battery damaged chips in arcade machines is that some may be dead already and those that aren't are gonna fail soon. I fixed a Qix (original Taito) and it took several months for the last chips with corrosion to die and on average one per week died. After like three or four months, the machine kept working for much longer intervals between chip failure. It's pretty much working for years now. By heating up the clock chip the battery liquid that made it into the chip has heated up and probably cracked the die. Same thing would have happened after hours of normal operation.
@adriansdigitalbasement - I notice that a garbage date from the clock chip resulted in a year of 2036, which means the system can actually do current dates. I wonder if it's possible to hack the date/time config code to allow entering y2k+ dates? Maybe something for usagi's crew to tinker with?
It was a common convention, at the time of this computer, to have the SCSI controller at ether 3 or 7, and the tape drive at 7. Also, keep in mind the that SCSI bus is exactly that, a buss that you hang devices off of it, and the controller is just another device like the disk (with termination at both ends).
Minute 41:05. Another failure mode not listed is R52 short (maybe an external path due to corrosion). R51 and R52 are not a voltage divider in this configuration. Base voltage will always be 0,6V-0,7V with or without R52. The purpose of R52 is to reduce the switch off time of the transistor (recombination of minority carriers in BE junction).
This is honestly the most compelling project I think I've ever watched you work on. I love it! 😀 I think it's because it started as such an enigma and there was so much to learn.
When I was working at a contractor in 1983 we had a development machine a PDP 11/70 that ran System V UNIX. So System V 5.2 release 8. We had an RL02 drive but the copy of UNIX arrived on tape. Good job everyone!
Really enjoyed the series on this beast! Great to know another piece of history has been saved. Thank goodness it landed in the right pair of hands to save it from obscurity and the trash.
Am looking forward to seeing this beast actually do something useful, thus fully giving it the final respect & dignity due its station. Much respect for your accomplishment! Kudos!
Incredible work from you and the community. What an absolute retro tech treasure trove. Massive props to the original developers of this machine for putting together such a comprehensive and well-engineered debugging and configuration interface as well, way ahead of it's time.
Not into retro computing personally, but i just absolutely love seeing the troubleshooting and repair process. Calming and fascinating. Especially in the case of obsecure machines that have been forgotten. Loved the longer format. Always excited when you do cuts for troubleshooting as i know you are coming back with a discovery. Been eating your channel up lately.
Congratulations on getting this beastie working. One thing: you may wish to insulate the inside of the cover with some vinyl tape where the clock module is, because what's facing out is the POSITIVE terminal of the battery. If that goes to ground it could be inconvenient at best, and could cause a problem at worst.
Wow, Adrian that really cheered me up, I've been having a tough time and seeing you (and the team) overcome this incredible technical challenge has really made me feel great. Thanks.
25:50 od is *the* classic octal dump tool under Unix all the time ... beside the fact it es been constantly extended (to even show hex, ASCII and different formats nowadays)
In 1985 the Ethernet standard wasn't completely developed and approved; it was still in flux. I am not sure an Ethernet card from that era would be able to communicate with a modern Ethernet based network. SLIP was around in 1984, and we know the Plexus has serial ports. I would look into that as an option.
@@eshwayriExcelan Multibus Ethernet adapter was available for this system... Just because the stupid standard didn't exist doesn't mean ethernet wasn't being utilized by many systems.
Really enjoyed this series - reminded me of my time working as a DEC and Solaris sysadmin in the 90s and early 00s. I've installed and updated unix from various tape formats.. DAT, DLT etc. Installed and configured SCSI cards+devices, added memory... Early on we did all that and only got DEC/Sun support in to fix issues where processor boards needed replacing... Those were heady days indeed!
I think you could transfer files to it by using tar or cpio from a serial port and feeding it from a Pi. Might take a bit of experimentation to sort out the handshaking and baud rate on the getty process. You even try cat ing the input from a serial port to disk, but this sort of stuff was pretty standard back in the day for software porting.
The big disks are SMD, and could have been attached to a SMD controller on the multibus. Should be easy to just look for mentions of a SMD driver. Edit: ID 3 is missing because on this machine ID 3 is the host. The host does not always take address 7.
I'm pretty sure ID 7 is the standard host port for controllers with 16 (0-15) IDs, while this system only has 8 (0-7) and port 3 is the standard host port.
It would be nice if the unix on this system has a slip or (very fancy for the time) ppp implementation installed. With that, and a linux machine, you could have it 'talk' tcp/ip over a serial link, and connect it to your internal network (using the linux machine as router). That would make moving files to its disk a lot easier, and would get it 'connected to the internet' even without ethernet card.
this series is proper! that handy python script is unbelievably helpful. you've brought a stone cold dead machine to life, on our screens. thank you for sticking with this and documenting the process, watching you work through this stuff is an education in itself. and a great little machine to work on.
I think the video was done just fine Adrian.. no need to apologize to us.. what a great result too, how fun is this?! I can't wait to see what you have planned for this monster.. My mom used to use a Unisys machine that ran a Unix system on it for payroll, she worked for Burlington Coat Factory (the main office in Burlington, NJ) she was the payroll manager I suppose back in the 80s she wasn't really called manager as women in offices, eh, it wasn't like that back then, but anyway.. yeah and then even when she moved to the township job she upgraded to in the late 80s, she was using the SAME machine and same Unisys system.. it was like a strange DOS machine, red monochrome monitor (yes, RED) and well I suppose the software was some type of spreadsheet. I was young back then but was VERY into computers so I got to play with it any time I showed up at her office!! Neat stuff!
Tis has been a really awesome journey with the Plexus, looking forward to you getting the remainder of the hardware functional.
6 місяців тому+4
Love the P/20 episodes! Not everyday you see an early UNIX system in action. Looking forward to seeing the FDD fixed and some programs compiled for it. May I suggest "One UNIX Workstation, 8 ASCII Gamers LAN" video 😆?
I set up a multi node UUCP network in the early 90's and it can transfer lot's of things. I haven't played with it in a couple of decades but I did link a Linux host to an old sco open server 5 box to do a data recovery. I'll have a play with a modern Linux build and if successful I join your discord and do a cheat sheet on setting it up. Nice to see an old System 5 box up and running. My oldest working Unix box is an SGI Indy.
I think you deserve a break, now that you've gotten it working somewhat stably. You've certainly earned it! What a roller coaster of emotions your series on the P/20 has evoked. This video was totally worth the nearly two hours spent to watch. (Thunderstorms interfering with my WiFi Extender, so I kept losing signal.)
What a great challenge to follow along, I really enjoyed watching this series Adrian. It would be great to see this machine with an Ethernet interface.
Great series of videos and glad you got this unique machine running... I love old hardware like this - much more interesting than just an old PC.... thanks!
Well that was an interesting rabbit hole to go down and explore.. what’s the next exciting expedition going to be .. I guess we’ll have to wait. Cheers mate.
This P/20 mini series kept me coming back just to see if it could finally run error free. Seeing the troubleshooting process and help from the discord users was pretty awesome.
congrads on this one, it was next level difficult. It was also great to see it show up on ncommanders page. I would have loved to see him do a deep software upgrade dive on it. Can't imagine something like a kernel compile on this machine.
Congratulations on getting this antique up and running! I think these PLEXUS units were used for other uses besides multi-user systems. The case looked strangely familiar to me and I recall seeing a system that looked like this at a Printworld show in the mid-80s when my family was looking at typesetting systems for their prepress graphics company. They settled for a Varityper Epics 20/20 system, another unique system that would be a goldmine to find someday, by the way. The company I saw sold RIP software and imagesetter combination for printing films and RC-paper output that looked very similar to this. The computer system is connected directly to the high-resolution imagesetter via one of the SCSI ports because a serial connection is too slow and imagesetters need a continuous feed of data otherwise buffer under runs will occur.
Yay! Nice to see the pressing problems sorted and a resultant wall command post of "It freakin' works!" 🍻 - Shout out to the folks assembled via Usagi's discord to help unravel the mysteries. Like so many other people here, I kept having a chortle every time "BATT" signal was mentioned, along with a quick mental image of a caped crusader 😂 Good stuff Adrian - look forward to seeing the mighty Plexus in a future episode, after you've had a bit of a well earned break from it.
uucp dates back to the days when computers would call each other by modem and send a burst of queued files. You can imagine that scheduling calls for late nights and weekends would save money on phone bills.
Yes awesome! I have used a PC version named UUPC and I spent a whole week reading the manuals to understand how to configure it. Awesome sleepless nights!
Usenet was the first 'social networking' protocol originally running over UUCP, and later over TCP/IP. UUCP based mail was the first almost internet mail, and if you connected to a machine that HAD internet access mas a gateway it WAS internet mail.
As much as I love Commodore machines (big part of my youth), I have thoroughly enjoyed this series, as it has been the perfect balance of the crossover period between minis and micros, and was a nice balance between your normal content, and David's, as it had a mix of both styles. Plus, it is just a damn sexy machine!!!
The last time I was that glued to the screen, I had just super-glued a damaged chunk of plastic back on and it leaked out under my fingers where I was holding it in place without me realizing it....
I remember when I got started as an IT rookie we had what I think was a HP9000 d350 that required a tape to be in place in order to boot. It might have been a slightly older model but it's been 25 years .. I just remember it being a HP9000 and it had to have a special DAT tape in place for it to boot, even though it also had an LTO1 drive. It was replaced by a hp9000 that did not need a tape to boot so it was a model from the early 90s(90-95) somewhere. It was already running hp-ux though.
I’m at 10:16 , and just throw a guess; The return vs expected values look like a “bit error” in either i/o bytes? There’s a slight “logic pattern” between the two values, like frozen or missing (floating) one or more bits to the byte or nibble…? I know, the answer comes later in the video, but it’s like watching “true crime” for logic boards. Have to guess, right…? 😅 [Edit @ 1:03:26 ; I was wrong - but not entirely…? (Comforting myself…) Great Series and work! 👍]
That boot loader environment is really interesting - I'm used to seeing built-in commands in a *nix boot loader (like "ls" or the like, things to help you find where your old kernel is after you bork the new one, etc.), but the fact that you can run everything in /stand is really interesting. Where does /stand live I wonder? Is the /stand you see in the boot loader the same as the /stand that lives on the host's actual root partition? Historically I believe that was a sort of rescue partition that did two neat things: 1) all the binaries are statically linked so if say "/usr/lib" isn't available (as it wouldn't be when booting off rescue media or booting single user w/o other partitions mounted) you can still run all those commands and 2) it was one binary with multiple hard links to it that acted differently based on how it was called (think "busybox"). I'm wondering why /stand binaries dump core on the running system - I'm thinking they only can run in that weird boot loader environment. Would love to know more about this!
Great series. I know you mentioned the editing on this, but I thought it was great and kept the whole vid moving (we know you can solder some components together ;-)
This was the best since the SWTPC. Games are not my main goal in retro-computing so these systems interest me more. I think that you should do a future video or series where you show us around the software in the system, and demonstrate the multiuser capabilities.
Back in the old days of usenet, we'd use text files that were self extracting sh scripts using uuencode. You can use a command to package a set of source code to Kermit or something like that into this format, and then extract it using bog standard /bin/sh to create a .tar.Z file (assuming uncompress is on the system), then decompress it, untar it, and then compile it. Then you'll be able to use Kermit or whatever to transfer files.
+1 for uuencode/uudecode. You can use it to faithfully transfer binary files over the serial port with nothing at the other end but standard Unix utilities. Good for compressed tar files!
Great episode. As an aside I noticed the Beagle Bros t-shirt you were wearing and that brought back fond memories from my early 1980's Apple II programming days using some of their utilities.
![Lp. Сердце Вселенной #60 РОЖДЕНИЕ ЛОЛОЛОШКИ [Финал] • Майнкрафт](http://i.ytimg.com/vi/YoR0pAV9FVQ/mqdefault.jpg)
Usagi's discord crew are seriously some of the smartest retro-tech people in the world when it comes to these 70s and 80s minis and server machines. Absolute brilliance and it's so good that they are around helping people like you preserve the less well documented aspects of our digital toddlerhood. Great stuff Adrian!
That is why an open source is so important. To have a place to publish when the project is over. But it needs to be in the culture or otherwise all the legal people complications
@@oidpolar6302 You are right. But it is not the terminology or ideology that is important, but the people. I do cracking and piracy since the Commodore PET times (which was of course voluntary work for the kids back in the time, who never could have payed the astronomical price-ideas of the corporations and growing game publishers (they and capitalism spreads like the plague)) and voluntarily work in the open source community since I was a kid. We live in great times now. But my fear is that this will change. After the next mayor privacy scandals the mood may change and also the legislation ... which may make the illegal and unethical handling of data (of strangers!) from big-corp and start-ups very unattractive and threatened with prison. The same inevitably will happen when those embarrassing hypes, like the K.I. nonsense implode or (what already happened) the "crypto-bro's" make new friends and opportunities in prison showers. Then free services like open repositories and free storage, free websites, sponsored projects and the like will have a hard time. Anyways, at any point in the future we have to ask and answer those questions. A free society (including free software or voluntary/free work on software) cannot be based on unethical, immoral concepts.
Those corporations today, who provide this "free" storage will not hesitate to delete the data. Even the recovery of such data will be hard ... BECAUSE MONEY IS INVOLVED! So be aware that if we do not take precautions for the future (which will cost our OWN MONEY!), all that wonderful work, data, acquired and with great effort renewed knowledge ... may be lost! This happened in the past and happens right now to a rate most people do not even wanna know or believe. Not only with printed or (older) historical data, work and art, but right now ... with all kinds of different media. F.E. Good luck with QIC-80 tapes!:)
Take that with a grain of salt and please not as fear-mongering or politics-critique (there is and was nothing better than capitalism and democracy. who says otherwise can p___ off to this wonderful North-Korea:P). Just to be aware that Google, Start-Ups, Tech-Bros and especially Elon Musk are NO ALTRUISTS!
Have a good one, and don't let "them" catch you ... Oidpolar, exidy and community!:)
I used to service Plexus machines in the 80s. I never came across a P/20 though. P/30, P/35, P/55 and P/75, but not a P/20. They were quite good machines. Very reliable.
I'd still really like to see a diagnosis of the power supply reset circuit. One of your other commenters and I on the last video had a debate about whether your fix was potentially bypassing a potentially more serious problem in the power supply That is actively happening, or if it wasn't that big a deal and your reset circuit wasn't an issue. To that end, I'd love to see the power supply revisited and rescoped out both to check the voltages again as the other user suggested should be done, and to see if we can get an understanding of The reset circuit and how it monitors the power. It looked like a complicated circuit and it would be interesting to see what it's responding to or if it's just broken.
"Fully operational" reminds me of Emperor Palpatine saying "witness the power of this fully armed and operational battle station."
Therefore, the Plexus is the Death Star. Lol.
I guess it reminds me of At The Drive-In. This station is non operational
If he ran it on sun power it would be a Solar Plexus.
Glad to see it wasn't just me.
Beware the liability of the fan port :)
I think the reason /stand commands didn't run in Unix is that they are not compiled for Unix but compiled directly for the CPU. They probably try to talk directly to the peripherals rather than speaking to the kernel. Sort of like an ATTiny doesn't have an operating system, it just runs the raw code. So even if the program could execute, it might try to access a protected memory space in Unix since it thinks it has full control of the memory map, Unix might intercept that and force the program to dump and exit.
Yep that is almost certainly the case -- luckily there are alternative unix-friendly version on the disk too.
/stand is for munix (mini-unix) and sash (standalone shell).
There is a complete set of man2 files with .2s suffix. Among them, srcheof.2s is most fancy since it's designed just for installer tape head positioning. :-)
@@adriansdigitalbasement Indeed so. Ed 7 Unix had these and others I would think. We were somewhat amused to discover there was a stand alone cat. 🐈
It looks like ccal is available as /etc/ccal, interesting that it doesn't appear under a /bin directory.
I was suprised about how different is from other contemporary Unixes (AIX, SCO, IRIX...). They usually use the "bios" only to setup whatever devices you have and load the kernel from there. After that, they usually had a tape or disk partition for statically linked binaries mounted in memory on /sbin to be used like this /stand.
This is quite cool, but I guess one needs to maintain those tools separatelly for each hardware which is not as practical.
I had zero interest in this plexus machine because I thought "what can you possibly do with it with no documentation" until I heard what was on it.
That was a game changer! Well done to you and the discord team. That is a once in a lifetime score.
Would be funny it was done by a disgruntled employee. I'll show them, here are the docs in the wild......
Much like DOS 6 source code.
This is really a fun series. Almost like an archival treasure hunt. Every new episode has surprises.
It has been quite the roller-coaster ride of a series for the P/20. So awesome that the community real came through to help out making this such a successful adventure!
Such a lovely series, it was very exciting detectives work! Thank you for not giving up and sharing. Also thanks a lot to the community and all the help.
Unix System V svr2 commercial source license was for $43,000, with three months of support, and a $16,000 price per additional CPU…. That’s some expensive system you just resurrected !
Considering the small customer base of hardware like this and the level of support required, this machine might have cost several 100.000 USD. All to be thrown into a landfill after a few years of use.
just the operating system ? nice scam they had going back then the hardware i can understand it seems advanced for the time
If it works like the SGI 4D/20 (another SysV based computer), you only needed a source license (which was, indeed, expensive) if you were planning to recompile the operating system itself. If you didn't get the source license but did get a compiler, you still got all the necessary library and header files for compiling other programs, just not the actual kernel code. (I'm pretty sure you got at least some of the libraries even if you DIDN'T get a compiler option, because they were the same DLL files the OS itself needed to run.)
System 5.2 Release 8
@@belstar1128Unix was also really advanced at the time
Paused the C compiler test just to verify you had it print Hellorld as expected. Nice work reviving this super interesting machine!
Hellorld on all the things. The entries for TI-83 and TI-86 in the Hellorld wiki are mine. :)
Truly these Plexus videos are some of your best work ever, Adrian! Thank you for making them!
P20 is a handsome machine and just reconfirms my belief that all machines need more blinking lights on them. Love this series.
This has been one of my favourite series on the channel. Congratulations on getting it running! The community really came through with this one. Love it!
The com port loses connection because in balanced power mode unused devices are shut down when the machine goes to sleep mode. You can change individual power saving settings for devices though to stop it from putting it to sleep due to inactivity. Putty also has an option for how often it should send keep alive and on lower baud rates that often needs to be increased so the emu can keep signal sync. I know it's convenient, but sleep mode isn't the greatest option for hardware. Low c states still have current bleed issues even on modern CPUs... yeah be a cold day in hell before that ever gets fixed. Congrats on getting the old girl running :)
the transistor circuit that un select cal.ce- when in reset / power down is probably against content corruption of the registers; The bus is in a unknown state and stray voltages/signals may be present on the bus that could cause a random write to one of the registers is the chip is selected.
I find it so interesting that these old unix device names still are part of being used today. If you use a LTO tape drive, the drive is like /dev/st0 and the "no rewind" function is /dev/nst0, notice the extra "n". And that is found in the device names here as well, it's simply fascinating :)
I was surprised the cpio format was from that era as well, I didn't think we still used such old formats
@@kernelramdisk3348 Red Hat's rpm package format still uses cpio format! So it is "current" technology, LOL!
This series has been amazing. Hidden technical specifications on the hard drive for the win!
You’re the friggn Columbo of old OS’s, “one more thing” 😂
This series was awesome. Bringing in other communities and even having NCommander give it a go, really made this feel more like a on going event. Great work all around.
This would be a nice machine to bring to a retro computer party with many terminals connected and then do 80s era chatting for example.
My car broke down. Luckily my car particular had the engineer and mechanic who designed it stored in a secret compartment in the glove box. With both of them free of the carbonite we were able to get it going again.
Your upbeat attitude through this, despite all the setbacks, is inspiring.
@32:00 I reviewed the scope meter, it does pretty well, the multimeter is actually very accurate, I tested it on my calibrator and reference standards, like any of these devices the scope is compromised a bit, but is fine for a lot of work.
I had no idea what you were talking about 95% of the time but I really enjoyed this series nonetheless. I get enjoyment out of seeing people talk about things they are deeply knowledgable about.
That Reset inverter circuit may exist to provide spurious write lock-out while the system is powering up or down. Tying /CS low, permanently might cause Config memory corruption.
So great to see you digging into something that isn't a mass-produced 80s 8-bit microcomputer.
My experience with battery damaged chips in arcade machines is that some may be dead already and those that aren't are gonna fail soon. I fixed a Qix (original Taito) and it took several months for the last chips with corrosion to die and on average one per week died. After like three or four months, the machine kept working for much longer intervals between chip failure. It's pretty much working for years now.
By heating up the clock chip the battery liquid that made it into the chip has heated up and probably cracked the die. Same thing would have happened after hours of normal operation.
@adriansdigitalbasement - I notice that a garbage date from the clock chip resulted in a year of 2036, which means the system can actually do current dates. I wonder if it's possible to hack the date/time config code to allow entering y2k+ dates? Maybe something for usagi's crew to tinker with?
Congratulations! Such an epic repair series! Very inspiring.
Adrian: “Well, that pretty much tells us one thing”
Also Adrian: *immediately lists at least three things*
It was a common convention, at the time of this computer, to have the SCSI controller at ether 3 or 7, and the tape drive at 7.
Also, keep in mind the that SCSI bus is exactly that, a buss that you hang devices off of it, and the controller is just another device like the disk (with termination at both ends).
Minute 41:05. Another failure mode not listed is R52 short (maybe an external path due to corrosion). R51 and R52 are not a voltage divider in this configuration. Base voltage will always be 0,6V-0,7V with or without R52. The purpose of R52 is to reduce the switch off time of the transistor (recombination of minority carriers in BE junction).
Every time I heard "BATT Signal", my brain filled in "Na na na na na na na na na na na na na na na na... " 🙂
me too! to the bat cave!
Damn! I miss those good ol' @Barnacles unboxings with the bat-knife(tm)
Holy fan-out, BATTman!
No, to the bat basement!
Same here and couldn't stop laughing.
This is honestly the most compelling project I think I've ever watched you work on. I love it! 😀
I think it's because it started as such an enigma and there was so much to learn.
When I was working at a contractor in 1983 we had a development machine a PDP 11/70 that ran System V UNIX. So System V 5.2 release 8. We had an RL02 drive but the copy of UNIX arrived on tape. Good job everyone!
System V ran on the PDP 11!?
@@roundduckkira The very first release did but I don't think the later releases did. However I am not an expert.
Love your content. I understand less than 10%, but your enthusiasm and ability to troubleshoot is mesmerising.
Really enjoyed the series on this beast! Great to know another piece of history has been saved. Thank goodness it landed in the right pair of hands to save it from obscurity and the trash.
A triumph! Well done to Adrian and all involved from Discord.
Am looking forward to seeing this beast actually do something useful, thus fully giving it the final respect & dignity due its station. Much respect for your accomplishment! Kudos!
Adrian you have me riveted from start to finish. Your the Sherlock Holmes of the computer world.😊
Incredible work from you and the community. What an absolute retro tech treasure trove. Massive props to the original developers of this machine for putting together such a comprehensive and well-engineered debugging and configuration interface as well, way ahead of it's time.
Not into retro computing personally, but i just absolutely love seeing the troubleshooting and repair process. Calming and fascinating. Especially in the case of obsecure machines that have been forgotten.
Loved the longer format. Always excited when you do cuts for troubleshooting as i know you are coming back with a discovery.
Been eating your channel up lately.
Congratulations on getting this beastie working. One thing: you may wish to insulate the inside of the cover with some vinyl tape where the clock module is, because what's facing out is the POSITIVE terminal of the battery. If that goes to ground it could be inconvenient at best, and could cause a problem at worst.
Wow, Adrian that really cheered me up, I've been having a tough time and seeing you (and the team) overcome this incredible technical challenge has really made me feel great. Thanks.
It's fun watching you having so much fun with this project.
Freaking fun!
The combination of luck and technical expertise in this series of videos was just incredible. Well done and thanks!
One more 'Well done' to add to the collection. Your tenacity is impressive.
25:50 od is *the* classic octal dump tool under Unix all the time ... beside the fact it es been constantly extended (to even show hex, ASCII and different formats nowadays)
Ethernet card would be so cool.
was one ever made?
In 1985 the Ethernet standard wasn't completely developed and approved; it was still in flux. I am not sure an Ethernet card from that era would be able to communicate with a modern Ethernet based network. SLIP was around in 1984, and we know the Plexus has serial ports. I would look into that as an option.
Webserver... Put it on the internet
@@eshwayriExcelan Multibus Ethernet adapter was available for this system... Just because the stupid standard didn't exist doesn't mean ethernet wasn't being utilized by many systems.
Attach a serial to Ethernet adapter and can at least get on the console
Really enjoyed this series - reminded me of my time working as a DEC and Solaris sysadmin in the 90s and early 00s. I've installed and updated unix from various tape formats.. DAT, DLT etc. Installed and configured SCSI cards+devices, added memory... Early on we did all that and only got DEC/Sun support in to fix issues where processor boards needed replacing... Those were heady days indeed!
Your troubleshooting technique is spot on!!…that’s half the battle. You have inspired me to tackle my dead Amiga 4000T…thanks
I think you could transfer files to it by using tar or cpio from a serial port and feeding it from a Pi. Might take a bit of experimentation to sort out the handshaking and baud rate on the getty process. You even try cat ing the input from a serial port to disk, but this sort of stuff was pretty standard back in the day for software porting.
The big disks are SMD, and could have been attached to a SMD controller on the multibus. Should be easy to just look for mentions of a SMD driver. Edit: ID 3 is missing because on this machine ID 3 is the host. The host does not always take address 7.
I'm pretty sure ID 7 is the standard host port for controllers with 16 (0-15) IDs, while this system only has 8 (0-7) and port 3 is the standard host port.
@@CollinBaillie "Standards are great, everyone should have one". I have seen systems use ID 6 (DEC VAX) and ID 0 (Nortel PBX) too.
It would be nice if the unix on this system has a slip or (very fancy for the time) ppp implementation installed. With that, and a linux machine, you could have it 'talk' tcp/ip over a serial link, and connect it to your internal network (using the linux machine as router). That would make moving files to its disk a lot easier, and would get it 'connected to the internet' even without ethernet card.
this series is proper! that handy python script is unbelievably helpful. you've brought a stone cold dead machine to life, on our screens. thank you for sticking with this and documenting the process, watching you work through this stuff is an education in itself. and a great little machine to work on.
What a series of videos with an amazing successful outcome. Kudos to all who assisted. Well done, Adrian
I think the video was done just fine Adrian.. no need to apologize to us.. what a great result too, how fun is this?! I can't wait to see what you have planned for this monster.. My mom used to use a Unisys machine that ran a Unix system on it for payroll, she worked for Burlington Coat Factory (the main office in Burlington, NJ) she was the payroll manager I suppose back in the 80s she wasn't really called manager as women in offices, eh, it wasn't like that back then, but anyway.. yeah and then even when she moved to the township job she upgraded to in the late 80s, she was using the SAME machine and same Unisys system.. it was like a strange DOS machine, red monochrome monitor (yes, RED) and well I suppose the software was some type of spreadsheet. I was young back then but was VERY into computers so I got to play with it any time I showed up at her office!! Neat stuff!
Tis has been a really awesome journey with the Plexus, looking forward to you getting the remainder of the hardware functional.
Love the P/20 episodes!
Not everyday you see an early UNIX system in action.
Looking forward to seeing the FDD fixed and some programs compiled for it.
May I suggest "One UNIX Workstation, 8 ASCII Gamers LAN" video 😆?
of all the systems ive seen on your channel this one is the COOLEST
You are an inspiration. I’ve got a server that randomly reboots and I have been putting it off and putting it off. Pain, pain, PAIN.
Wow I was totally riveted to this mini series, what determination and persistence. Well done!
such a cool looking soulful machine you got there nice nice nice i hope in 5k years it's cherished
I set up a multi node UUCP network in the early 90's and it can transfer lot's of things. I haven't played with it in a couple of decades but I did link a Linux host to an old sco open server 5 box to do a data recovery. I'll have a play with a modern Linux build and if successful I join your discord and do a cheat sheet on setting it up. Nice to see an old System 5 box up and running. My oldest working Unix box is an SGI Indy.
I think you deserve a break, now that you've gotten it working somewhat stably. You've certainly earned it! What a roller coaster of emotions your series on the P/20 has evoked.
This video was totally worth the nearly two hours spent to watch. (Thunderstorms interfering with my WiFi Extender, so I kept losing signal.)
What a great challenge to follow along, I really enjoyed watching this series Adrian. It would be great to see this machine with an Ethernet interface.
Great series of videos and glad you got this unique machine running... I love old hardware like this - much more interesting than just an old PC.... thanks!
I am loving the journey of this machine. Can't wait to see the next video on it
Thanks for this great series about the P20. And also my full respect to the helping discord guys.
Well that was an interesting rabbit hole to go down and explore.. what’s the next exciting expedition going to be .. I guess we’ll have to wait. Cheers mate.
This P/20 mini series kept me coming back just to see if it could finally run error free. Seeing the troubleshooting process and help from the discord users was pretty awesome.
Amazing work Adrian! I've enjoyed reading the comments from the UNIX greybeards.
congrads on this one, it was next level difficult. It was also great to see it show up on ncommanders page. I would have loved to see him do a deep software upgrade dive on it. Can't imagine something like a kernel compile on this machine.
Congratulations on getting this antique up and running!
I think these PLEXUS units were used for other uses besides multi-user systems. The case looked strangely familiar to me and I recall seeing a system that looked like this at a Printworld show in the mid-80s when my family was looking at typesetting systems for their prepress graphics company. They settled for a Varityper Epics 20/20 system, another unique system that would be a goldmine to find someday, by the way.
The company I saw sold RIP software and imagesetter combination for printing films and RC-paper output that looked very similar to this. The computer system is connected directly to the high-resolution imagesetter via one of the SCSI ports because a serial connection is too slow and imagesetters need a continuous feed of data otherwise buffer under runs will occur.
Yay! Nice to see the pressing problems sorted and a resultant wall command post of "It freakin' works!" 🍻 - Shout out to the folks assembled via Usagi's discord to help unravel the mysteries.
Like so many other people here, I kept having a chortle every time "BATT" signal was mentioned, along with a quick mental image of a caped crusader 😂
Good stuff Adrian - look forward to seeing the mighty Plexus in a future episode, after you've had a bit of a well earned break from it.
uucp dates back to the days when computers would call each other by modem and send a burst of queued files. You can imagine that scheduling calls for late nights and weekends would save money on phone bills.
Yes awesome! I have used a PC version named UUPC and I spent a whole week reading the manuals to understand how to configure it. Awesome sleepless nights!
Usenet was the first 'social networking' protocol originally running over UUCP, and later over TCP/IP. UUCP based mail was the first almost internet mail, and if you connected to a machine that HAD internet access mas a gateway it WAS internet mail.
@@leosmith848 Can you say "bang path"? I knew you could.
Batched many an email over UUCP.
Imagine having a scheduled dial out and some hacker war dialing with ToneLoc just dialed in. What a pain!
As much as I love Commodore machines (big part of my youth), I have thoroughly enjoyed this series, as it has been the perfect balance of the crossover period between minis and micros, and was a nice balance between your normal content, and David's, as it had a mix of both styles. Plus, it is just a damn sexy machine!!!
I would be interested in a little review of that Oscilloscope. Great vid as always!
The last time I was that glued to the screen, I was watching The Silence of the Lambs.
The last time I was that glued to the screen, I had just super-glued a damaged chunk of plastic back on and it leaked out under my fingers where I was holding it in place without me realizing it....
@@USFrozen Super-glueing my fingers together. Been there, done that. Been there again. done it again. And again. And again, And...
I remember when I got started as an IT rookie we had what I think was a HP9000 d350 that required a tape to be in place in order to boot. It might have been a slightly older model but it's been 25 years .. I just remember it being a HP9000 and it had to have a special DAT tape in place for it to boot, even though it also had an LTO1 drive.
It was replaced by a hp9000 that did not need a tape to boot so it was a model from the early 90s(90-95) somewhere.
It was already running hp-ux though.
Outstanding result! This has probably been my favorite series ever.
Awesome repair!! Can't wait to see what more you can do with it!!
Regarding SCSI ID #3 being the computer's interface, I seem to recall that was also the case on Sun Microsystems computers as well.
For something so obscure, this had me sitting on the edge of my seat. I hope the next video will explore the software a bit more.
Great job! Congrats to you and the Usagi Discord community! So much fun to follow along.
So awesome to see this working again! I couldn’t believe the luck finding all that documentation on the drive. What an interesting series!
What an amazing journey thanks for sharing that with us!
Love this series as both UNIX history and retro hardware general are fascinating to me, very excited to see what you do with the system in the future.
1:02:00 Sys5.2r8 - my guess: Unix System V Rev. 2 Subrevision 8 / SVR2.8, which came out ~1984
I’m at 10:16 , and just throw a guess; The return vs expected values look like a “bit error” in either i/o bytes? There’s a slight “logic pattern” between the two values, like frozen or missing (floating) one or more bits to the byte or nibble…?
I know, the answer comes later in the video, but it’s like watching “true crime” for logic boards. Have to guess, right…? 😅
[Edit @ 1:03:26 ; I was wrong - but not entirely…? (Comforting myself…) Great Series and work! 👍]
That boot loader environment is really interesting - I'm used to seeing built-in commands in a *nix boot loader (like "ls" or the like, things to help you find where your old kernel is after you bork the new one, etc.), but the fact that you can run everything in /stand is really interesting. Where does /stand live I wonder? Is the /stand you see in the boot loader the same as the /stand that lives on the host's actual root partition? Historically I believe that was a sort of rescue partition that did two neat things: 1) all the binaries are statically linked so if say "/usr/lib" isn't available (as it wouldn't be when booting off rescue media or booting single user w/o other partitions mounted) you can still run all those commands and 2) it was one binary with multiple hard links to it that acted differently based on how it was called (think "busybox"). I'm wondering why /stand binaries dump core on the running system - I'm thinking they only can run in that weird boot loader environment. Would love to know more about this!
Great series. I know you mentioned the editing on this, but I thought it was great and kept the whole vid moving (we know you can solder some components together ;-)
A great series, thank you. I look forward to Plexus updates in the future
This was the best since the SWTPC. Games are not my main goal in retro-computing so these systems interest me more. I think that you should do a future video or series where you show us around the software in the system, and demonstrate the multiuser capabilities.
Love the series. Love that you got it working. Great job! Thanks for sharing.
Back in the old days of usenet, we'd use text files that were self extracting sh scripts using uuencode. You can use a command to package a set of source code to Kermit or something like that into this format, and then extract it using bog standard /bin/sh to create a .tar.Z file (assuming uncompress is on the system), then decompress it, untar it, and then compile it. Then you'll be able to use Kermit or whatever to transfer files.
+1 for uuencode/uudecode. You can use it to faithfully transfer binary files over the serial port with nothing at the other end but standard Unix utilities. Good for compressed tar files!
Loved watching this series. The machines I’ve never heard of can be fascinating.
Great episode. As an aside I noticed the Beagle Bros t-shirt you were wearing and that brought back fond memories from my early 1980's Apple II programming days using some of their utilities.
this is quite a journey! can't wait for the next part !