Huh. I saw the Polish-sounding name on the tag at 20:08 and looked this guy up. He had an interesting life! :o Fought in the WW2 during the German invasion and the Warsaw Uprising. Escaped from Warsaw after the Uprising through the sewage canals, then emigrated to USA to work in RCA and NASA. He studied in 3 different universities in Poland, Germany and USA (MIT) and somehow got the Apollo memory module on the way. :o
I commend you for even attempting to fix the memory module at all. When I realised it was potted I thought any repair would be impossible, I can't believe you managed to find another memory module to test and confirm the fault! I think the work you are doing with the AGC is incredible and I'm learning a lot more about the nuts and bolts of how a computer really works than I ever expected. Thanks for taking us all along for the ride.
Heres a thought. You know which bit is bad and where. Instead of repairing it just find some similar core memory and connect it to your existing module in such a way to make them work together like masked memory.
It's so much fun watching trained engineers dealing with a fault. Very methodical, very precise. It's unfortunate that the construction of the module precludes repair, but it's fascinating seeing you peel back the layers and investigate the issue. Sort of like C.S.I. but with less blood! :-) Looking forward to the next one.
I am always incredibly impressed by your videos Marc. Also I would like to point out how brilliant your speaking manner is in the narration. I really hope your passion for making this kind of content for us all never fades.
Seconded. Love the thought process and theory of operation narration. Makes a much more engaging video. Really allows you to compare how you yourself would approach the problem and compare.
The industrial design of the AGC, and especially of the modules in their slots, is so beautiful. It's almost difficult to believe that this was a product for industrial use, intended for a tiny number of users and never meant to be offered commercially at all.
If you even think about it, it’s a miracle the thing got off the ground with that technology not to mention flying to and landing on the moon! Absolutely amazing!
I know they talk about it a bit in the video, but look at all those tiny wires then think about them under 3g of load and vibrating all over the place. It's amazing that the engineers managed to make something so flimsy be able to survive such harsh conditions.
And if you watch Marc’s other videos about the IBM 1401 he says they have issues sometimes keeping that running for periods of time and realizing the AGC basically uses the same tech in a smaller footprint it really makes you realize more what an engineering feat this thing really is!
maybe also a reason that program was terminated... they just were ultra lucky.. imagine they had ´lost one or two Apollo Crews... the memory / AGC Is just a small part... all the other stuff was just as fragile
The museum should have let you simply swap modules, even if not permanently but at least until you got your AGC working. I'm pretty sure the museum doesn't plan on using their AGC for anything other than display purposes.
Come on, give them the module... at least for a while to run the AGC ... you can put the broken one at display, who cares...And why is there a classic Mercedes in the computer museum ?
The problem solving, the detective work and all the help from so many people is so awesome to see. Fantastic work and thanks for sharing this incredible series for free!
This is brilliant. Your plan to figure out how long and where the break was, was seriously awesome. I work in Aerospace and just got into Engineering not too long ago... Love your videos, they inspire me to never stop learning.
I think I'd go with what appears to be your gut feeling that trying to repair the existing core module isn't in the best interest of preservation at this point in time. There's no crime in leaving that work to a future preservationist with access to more advanced methods. If you document what you've done well, which it really looks like you have, it will give the next generation a leg up on solving the problem. I think a workalike module with modern components is a perfectly reasonable solution to the problem for the time being. It's not like the entire unit is 100% period anyway - the replacement DSKY is an obvious example of that.
That's a good point to leave it to the future to attempt to repair, but my apprehension on that is two fold: 1. We still have some of the people alive who developed it and can advise to a point and 2. We don't know how much longer the hardware will be serviceable. It would be awesome to have it functional again while the people who developed and also the people who experienced its contribution are still alive.
Was there a Parity bit? if so what you can do is read the memory out and use the parity bit to fill in the broken line in the core It going to be odd or even parity. Remember working with core memory. I may have an old DEC PDP-11 Core Module around
Another awesome video! Thanks for doing all of this. I don't know what your plans are. I would guess that one plan would be to make something from modern stuff that emulated this memory, which would be okay, but I'd really, really, like to give you the encouragement to attempt to build a duplicate of this memory somehow. With modern CNC machining (and connections to Samtec), it seems to me like the task of weaving a new core stack would be time consuming and meticulous, but I bet you that 1000 people would volunteer to weave the cores by hand. I think you could probably make some jigs to aid the process and get it done. Whatever you decide to do, thanks again for all the work. I love this series!
Has anyone considered making the memory. This stuff was originally hand made. With the benefit of 3D printing to make assembly jigs and perhaps some bits from you sponsor it's entirely possible to just make a new block of memory.. This is something a person with good eyes and a steady hand can do. You may even be able to find some good core memory from a different system that can be re-mounted for use in the AGC. Although it may be hard to find someone willing to let you cut up their old IBM core memory cards.
I wish they'd let you borrow the working module in exchange for your broken one - it's not like it matters if it works or not on display in a museum...
That'd be impossible as far as a historical value is concerned. Kinda like asking The Louvre to give a flake off The Gioconda in exchange for another off the Lady with an Ermine. Neither collection would agree.
This is a real cliff hanger! I just love to see all your mad equipment and extraordinary skills come into play trying to solve the problems and unearth the details of this beautiful machine. Looking forward to the next episode! 🖖
@Brian and @Colin: unfortunately it's a bit more involved than just swapping items. a) It's not entirely up to them. Items are donated to their trust with strings, usually restrictions on what can be done with it. Some can't even be shown, much less used. b) This is a historically significant and complete AGC donation: the first prototype Block II unit from Draper Labs. Every module on it is RAY #1 or RAY #2. You cannot just "swap things out" and keep its historical value intact. c) There is unavoidable tension between restoration and preservation. It's hard to do both well at the same time. The CHM Museum's charter is very clearly preservation (like most other museums, actually). On the other hand, the LCM (in Seattle) charter is clearly restoration (which is more unusual). Both are helping computer history in many ways, but both are following their charters in ways we don't always like: the LCM rips through historical machines and replaces out old components for new ones in ways that makes us cringe; the CHM is very uptight about preserving their artifacts absolutely original and unpowered. That said the CHM is doing restorations on items they have several of (like the IBM 1401 and the Xerox Alto's), and occasionally on unique items (the PDP-1). They have only one AGC, and it is slated for preservation. d) The CHM and its curators are very aware of the significance of the 50 years Apollo anniversary, and are quite supportive of our project. If there is a way to help us out while not endangering - or even better, while enhancing - the preservation charter, they will support it. e) We have several collaboration proposals in the works, stay tuned!
@@CuriousMarc well... I didn't mean to make it sound so simple - but it does seem like an obvious solution at some point. Good Luck on the project - this series is easily one of my favorites on YT these days! I mean you could totally cheat and pop an arduino and a few AA batteries as a simulator in there an nobody would know but of course that kind of defeats the whole purpose in the end. Anyway - it's all very fascinating to me!
@@CuriousMarc Does NASA have any spare modules floating around? Its possible there are others in the hands of collectors too. Amazing this was rescued from the scrap!
Not sure if NASA has any, but the Smithsonian and maybe some other museums and private collectors. They probably did not have many extra of these around even at the time, as contruction was awfully difficult - by far the most difficult item to manufacture in the AGC.
That'd be fun! I'd make a one-bit memory module that intercepts those two pins (and related sense/inhibit) - then work really hard to completely conceal my work.
I was thinking would it be possible to rewire a 16 bit plane from a PDP-11 though the amount of effort it would take might be easier to just use static ram.
Yes, this seems like the best way forward. The electronics don’t seem that hard - making a pin compatible physical plug in seems to be the hardest part.
Even though I've been following along and the first part of the video was basically review its all presented in a more documentary instead of blogging format. Its quite a nice production.
Of all the brilliant CuriousMarc videos, this one is truly way beyond my technical understanding - and yet, I am so engrossed I almost don't dare to blink :-)
Always nice to have access to all that test gear isn't it? I've been watching your channel for some time and am fascinated by all the vintage hardware.
With all the expert knowledge available between you all and with all the test equipment at your disposal, i am sure it would be feasible for you to build your own memory module using modern day parts. You have all the design schematics available, someone could design the memory housing with connecting pins, someone else could design a pcb to replicate what the memory module does and then wire it to the connecting pins. At least this would get around the problem your having of having to try and find an original working memory module. Just build a modern day equivalent, that will at least then allow the system to run, then once it's all working as it should be, then i am sure the computer museum would allow you to use their original memory module.
Another idea is using a high voltage/low current signal generator on each end of the broken wire while scanning with the X-ray machine. At some voltage I would expect an arc which should be detectable. There is some risk of causing additional damage but it may be a good option “P”.
Actually, you wouldn't even have to swap anything. Just replace the broken bit with a parity generator -- it will reconstruct the missing bit from the other data.
I never expected it to be repairable, but that was an impressive attempt at it. The xray machine is so cool I would love to get my hands on that haha. I can't wait to see what you guys come up with, this is a really cool project.
Curious marc I love what you do for us and for the history as a whole. And your team is pinnacle in saving that history. Thankyou from just a fly on the wall.
We once did a really really bad hack for getting some broken wire to "work" in a similar situation. The circuit we had was current driven and everything was potted, hence very well insulated. We had someone come up with a high impedance high voltage current source that was driven so high (several hundred volts) but accurately enough that it worked for a while and we could do what we wanted, after a while though it stopped working, despite dialing the voltage always higher, likely the wires eroded. Someone once then jokingly said that with a high enough voltage the spark would be bright enough to actually see thorugh the potting where the break is...
So wish had kept the Dsta General Nova minicomputer I had with core memory....the choices of youth eh? Excellent series. Shows how to cope with faults, provide solutions/ plans AND deal with complete failure in an avenue of approach.
I'm enjoying watching this series, very interesting and challenging. Although much more complex, this AGC reminds me of Wang's 300 series calculator systems made in the same time period. Love the Tek 7854 scope w/waveform calculator, a classic!
emulation is like having a ultra great classic muscle car with a V8 engine that´s like porn... the engine is broken and instead of fixing it you put in a 2018 4 cylinder diesel engine just to drive the car
@@bobl78 Having said that it is still better then having it not running. I can't imagine them making a core memory module so in the end whatever they come up with will be an emulation.
Yeah, I don't see why you couldn't swap them - yours is broken, there's just sits on the wall non-functional anyways. Put the broken one on display, use the good one.
If the goal is to power up and run the Apollo software on the AGC, you could build new wires, ferrite core rings and run them outside the memory module hooked up in parallel to the broken wires. Chances are the broken wire is only one and you won't have too many new ferrite cores to hook into it.
Could you perhaps piggy-back a small circuit to just those pins to mimic a working section of memory? Another idea perhaps is to fabricate a module to mimic the look and behaviour of the original and use that just for demonstrations. That will leave the original module in good condition and may be easier than repairing it.
Loving this content... just finished the AGC book by Frank O'Brien, reading for the second time... stoked about the 50th anniversary (ongoing, if you look at the whole program). Dumping the historical memory will be extremely interesting!
Doesn‘t the higher dielectric constant of the epoxy resin in comparison to air lead to a much shorter length to failure? I guess the Er is at least 3 to 4 in epoxy so the failure should be nearer to the connector
Thoroughly enjoying this journey. It will be interesting to see the design and build of a substitute for the non-working AGC. I wonder if you can use modified "donor" core planes from an old IBM or DEC computer.
Would building some kind of emulation for the erasable memory be possible? something that plugs into the physical slot and responds in the expected manner with more modern materials and techniques? It might even be a good proof of concept on retrieving any residual data in the erasable memory (if any, still not 100% understanding how the memory lattice in that works) of both modules.
The problem would probably be power. If the module can be given a constant amount of power (let's say 5v), you could probably use something like a stm32F103C6, dirt cheap microcontroller at 37 GPIO pins with 20kb of ram. But that's no fun. Lets see a new build of the old design!
01:25 Mike.... Don't bend over like that... sit down man! I've just moved house recently and my body just emphasized with the pain of being bent over like that for long periods!
@@TheErador Not necessarily. It would not be easy but I suspect there would be a solvent that would soften the glue and allow access without killing the wire insulation.
@@thedamnyankee1 That is the 64K question. But IF as in this case there ARE others of this module, then possibly it could be an option worth looking at.
Perhaps try tracing the broken wire with a wire loop antenna? Feed a couple of 100MHz into pin with broken wire and use small wire loop antenna on HP8566B to trace the RF emission of the wire.
I wonder what if heat up or cool down the module with monitoring via connected tester? Is there a chance for temporary restoring the contact by thermal effects? If it happens, you can try to weld the wire by high current through it. But of course it's extremely risky and need some experiments with same wires...
Absolutely amazing video today!!! Keep it up Marc! I guess the next step is rebuilding or emulating the memory? (given you mention samtec sponsoring the connector rebuild). Cheers!!
Is there a way to inject a short wavelength signal on the wire using the estimated length for the optimal frequency and construct a tuned probe for detecting the localized signal strength? Might that help isolate the location of the break?
I tracked down the point of breakage one time with high voltage. I put the broken wire on a sensor input and a needle with oscillating high voltage to probe it. I used a flatbed plotter to move the needle and wrote a small program to scan the hole board. I didn't had the diagram of the board only the pin-out of the chips. I doubt it would work with such a tree dimensional construction. The distances are just to great for detection. Maybe the area can be tracked by using the wire as an antenna and get at least the region the signal stops.
i love the repair Series of the AGC. Have your think about Chemical desolving the Resin in the Module ? Ich think it is a posible way Also i think you have only another option that calls Rebuilding with old (magnetic core) or new (smd) technic. Eventulay a 3D Xray is a way but i think this machins are not accurate enoth. i wisch you good luck its awsome !!! Best Regards an All from Germany
@@TinkerLynx jes that is a prooblem for a chemicist maybe it is possible. I think the easyst way ist a retrofitted pcb with a microchip tha emulates the core memory. I hope the best and press both thumbs. This ist a awsome piece of history.
@@tollerteppich Ya, I looked into it a bit, an FPGA such as the flash based Intel MAX10 could do the job. The issue is latency, core memory is actually pretty quick. But I think the internal memory on an FPGA can do it. probably under $100 if someone would volunteer there time to do the coding.
@@TinkerLynx i have not the gackgrund to Programm an fpga. I can a litle bit C on a mega8 XD but i think here are experts on fpga that can help. Also i think with a lot of hand work it is possible to rewire the module after stripping with the original ferite cors
Can you simulate the broken bits externally? i.e. run the module as is - decode where you know bad bits are and use a modern circuit to replace them (bad block mapping style approach) ...and watched to the end - sounds like the above is approach D, E or F :)
Is it not possible to dissolve the podding in some way? Or machine it away... Curious to see what you'll come up with. Maybe some other module will show up or you'll end up with some FPGA substitute, who knows.
Great series an very impressive work. Is there a Chance you could make a video about the Tek 7854 ? It is a fascinating piece of Computer History by itself.
16:53 I love the path to the directory being displayed. D:\Documents and Settings\Marc\My Documents\Make\Projects\Apollo Guidance Computer\Memory\XRay Pictures.
Since the history museum module is just a display, maybe they'll let you swap it with yours. Otherwise, you might try using Methylene Chloride to dissolve the potting. Last resort, maybe Mike can gut your module, build an emulator board, then install it in the module (it would make a great video series). Good Luck.
Huh. I saw the Polish-sounding name on the tag at 20:08 and looked this guy up. He had an interesting life! :o Fought in the WW2 during the German invasion and the Warsaw Uprising. Escaped from Warsaw after the Uprising through the sewage canals, then emigrated to USA to work in RCA and NASA. He studied in 3 different universities in Poland, Germany and USA (MIT) and somehow got the Apollo memory module on the way. :o
I commend you for even attempting to fix the memory module at all. When I realised it was potted I thought any repair would be impossible, I can't believe you managed to find another memory module to test and confirm the fault! I think the work you are doing with the AGC is incredible and I'm learning a lot more about the nuts and bolts of how a computer really works than I ever expected. Thanks for taking us all along for the ride.
Heres a thought. You know which bit is bad and where. Instead of repairing it just find some similar core memory and connect it to your existing module in such a way to make them work together like masked memory.
Is there any solvents i wonder that could be used to dissolve away the potting?
Yes, but it will probably eat the insulation off the magnet wire as well.
A channel that truly shows the complexity of computer science. This is just art made out of science.
Love this series, Although I can't 100% follow all the technical things 80% is enough with the clear explanations.
It's so much fun watching trained engineers dealing with a fault. Very methodical, very precise.
It's unfortunate that the construction of the module precludes repair, but it's fascinating seeing you peel back the layers and investigate the issue.
Sort of like C.S.I. but with less blood! :-)
Looking forward to the next one.
I am always incredibly impressed by your videos Marc. Also I would like to point out how brilliant your speaking manner is in the narration. I really hope your passion for making this kind of content for us all never fades.
Seconded. Love the thought process and theory of operation narration. Makes a much more engaging video. Really allows you to compare how you yourself would approach the problem and compare.
It's just a computer repair, not rocket surgery.... Oh wait...
Technically both.
Thank you for showing all the process and not just final result, this is really of huge interest !
6:47am in the morning here and a new video about the AGC. There is nothing better to start the day. :)
The industrial design of the AGC, and especially of the modules in their slots, is so beautiful. It's almost difficult to believe that this was a product for industrial use, intended for a tiny number of users and never meant to be offered commercially at all.
If you even think about it, it’s a miracle the thing got off the ground with that technology not to mention flying to and landing on the moon! Absolutely amazing!
I know they talk about it a bit in the video, but look at all those tiny wires then think about them under 3g of load and vibrating all over the place. It's amazing that the engineers managed to make something so flimsy be able to survive such harsh conditions.
And if you watch Marc’s other videos about the IBM 1401 he says they have issues sometimes keeping that running for periods of time and realizing the AGC basically uses the same tech in a smaller footprint it really makes you realize more what an engineering feat this thing really is!
maybe also a reason that program was terminated... they just were ultra lucky.. imagine they had ´lost one or two Apollo Crews... the memory / AGC Is just a small part... all the other stuff was just as fragile
The museum should have let you simply swap modules, even if not permanently but at least until you got your AGC working.
I'm pretty sure the museum doesn't plan on using their AGC for anything other than display purposes.
Come on, give them the module... at least for a while to run the AGC ... you can put the broken one at display, who cares...And why is there a classic Mercedes in the computer museum ?
I was wondering the same thing... first car with modern ABS perhaps?
I agree, you would think they would be interested to see it work lol
Agreed. It's a waste having a working module sitting on that display
Exactly. Come on, CHM, do the obvious right thing here.
compu85 Bingo!
I love those old scopes where "plugins" literally ARE "plug-in".
Ohhh my Goshh .. I love those big tektronix !!! What a battle tank are those !!
And they have buttons! Not nested menus. It's at the same place every time, and actually does what's written on it ;-)
The problem solving, the detective work and all the help from so many people is so awesome to see. Fantastic work and thanks for sharing this incredible series for free!
This was truly amazing ... the fragility and complexity of core rope memory makes me wonder how it ever worked in the first place.
This is brilliant. Your plan to figure out how long and where the break was, was seriously awesome.
I work in Aerospace and just got into Engineering not too long ago... Love your videos, they inspire me to never stop learning.
I think I'd go with what appears to be your gut feeling that trying to repair the existing core module isn't in the best interest of preservation at this point in time. There's no crime in leaving that work to a future preservationist with access to more advanced methods. If you document what you've done well, which it really looks like you have, it will give the next generation a leg up on solving the problem.
I think a workalike module with modern components is a perfectly reasonable solution to the problem for the time being. It's not like the entire unit is 100% period anyway - the replacement DSKY is an obvious example of that.
That's a good point to leave it to the future to attempt to repair, but my apprehension on that is two fold: 1. We still have some of the people alive who developed it and can advise to a point and 2. We don't know how much longer the hardware will be serviceable. It would be awesome to have it functional again while the people who developed and also the people who experienced its contribution are still alive.
Was there a Parity bit? if so what you can do is read the memory out and use the parity bit to fill in the broken line in the core
It going to be odd or even parity. Remember working with core memory. I may have an old DEC PDP-11 Core Module around
Aha! Getting warm here...
Another awesome video! Thanks for doing all of this. I don't know what your plans are. I would guess that one plan would be to make something from modern stuff that emulated this memory, which would be okay, but I'd really, really, like to give you the encouragement to attempt to build a duplicate of this memory somehow. With modern CNC machining (and connections to Samtec), it seems to me like the task of weaving a new core stack would be time consuming and meticulous, but I bet you that 1000 people would volunteer to weave the cores by hand. I think you could probably make some jigs to aid the process and get it done. Whatever you decide to do, thanks again for all the work. I love this series!
Has anyone considered making the memory. This stuff was originally hand made. With the benefit of 3D printing to make assembly jigs and perhaps some bits from you sponsor it's entirely possible to just make a new block of memory.. This is something a person with good eyes and a steady hand can do. You may even be able to find some good core memory from a different system that can be re-mounted for use in the AGC. Although it may be hard to find someone willing to let you cut up their old IBM core memory cards.
Best series on UA-cam :) So wish I could get involved!
Maybe You can ower some form of textmessages?
I wish they'd let you borrow the working module in exchange for your broken one - it's not like it matters if it works or not on display in a museum...
No kidding , must have a real ass in charge there.
That'd be impossible as far as a historical value is concerned. Kinda like asking The Louvre to give a flake off The Gioconda in exchange for another off the Lady with an Ermine. Neither collection would agree.
Would love to see you come up with a modern day alternative that you can fit right in place to do the same job. Cool stuff.
This is a real cliff hanger! I just love to see all your mad equipment and extraordinary skills come into play trying to solve the problems and unearth the details of this beautiful machine. Looking forward to the next episode! 🖖
Why not just swap modules? The museum obviously don't need a working one and the broken one you have is more complete anyway.
@Brian and @Colin: unfortunately it's a bit more involved than just swapping items.
a) It's not entirely up to them. Items are donated to their trust with strings, usually restrictions on what can be done with it. Some can't even be shown, much less used.
b) This is a historically significant and complete AGC donation: the first prototype Block II unit from Draper Labs. Every module on it is RAY #1 or RAY #2. You cannot just "swap things out" and keep its historical value intact.
c) There is unavoidable tension between restoration and preservation. It's hard to do both well at the same time. The CHM Museum's charter is very clearly preservation (like most other museums, actually). On the other hand, the LCM (in Seattle) charter is clearly restoration (which is more unusual). Both are helping computer history in many ways, but both are following their charters in ways we don't always like: the LCM rips through historical machines and replaces out old components for new ones in ways that makes us cringe; the CHM is very uptight about preserving their artifacts absolutely original and unpowered. That said the CHM is doing restorations on items they have several of (like the IBM 1401 and the Xerox Alto's), and occasionally on unique items (the PDP-1). They have only one AGC, and it is slated for preservation.
d) The CHM and its curators are very aware of the significance of the 50 years Apollo anniversary, and are quite supportive of our project. If there is a way to help us out while not endangering - or even better, while enhancing - the preservation charter, they will support it.
e) We have several collaboration proposals in the works, stay tuned!
@@CuriousMarc well... I didn't mean to make it sound so simple - but it does seem like an obvious solution at some point. Good Luck on the project - this series is easily one of my favorites on YT these days! I mean you could totally cheat and pop an arduino and a few AA batteries as a simulator in there an nobody would know but of course that kind of defeats the whole purpose in the end. Anyway - it's all very fascinating to me!
@@CuriousMarc Does NASA have any spare modules floating around? Its possible there are others in the hands of collectors too. Amazing this was rescued from the scrap!
Not sure if NASA has any, but the Smithsonian and maybe some other museums and private collectors. They probably did not have many extra of these around even at the time, as contruction was awfully difficult - by far the most difficult item to manufacture in the AGC.
It's amazing to see how much tech has evolved to help us track down these issues. Love the videos.
Time to make a pin compatible static memory module in a fancy box... (without potting!)
That'd be fun! I'd make a one-bit memory module that intercepts those two pins (and related sense/inhibit) - then work really hard to completely conceal my work.
I was thinking would it be possible to rewire a 16 bit plane from a PDP-11 though the amount of effort it would take might be easier to just use static ram.
You only need to make a pin compatible adaptor for the broken bits - if they can be isolated
Use a flash based FPGA with the correct pinout and memory contents. I'm sure Lattice or Microsemi would happily give you the parts.
Yes, this seems like the best way forward. The electronics don’t seem that hard - making a pin compatible physical plug in seems to be the hardest part.
Even though I've been following along and the first part of the video was basically review its all presented in a more documentary instead of blogging format. Its quite a nice production.
Of all the brilliant CuriousMarc videos, this one is truly way beyond my technical understanding - and yet, I am so engrossed I almost don't dare to blink :-)
I was waiting for the next part of this series. This is great. Thank you!
Memory will just have to be dumped and emulated I suppose. I love this journey! It's so exciting to see when a new video has been uploaded.
Always nice to have access to all that test gear isn't it? I've been watching your channel for some time and am fascinated by all the vintage hardware.
Apollo repair serie deserve the title of the best tv show of the year, i like it more than game of thrones :-)
Yay ! 6.20AM here and this gem to eat my breakfast by ! Thanks as always Marc & the team
With all the expert knowledge available between you all and with all the test equipment at your disposal, i am sure it would be feasible for you to build your own memory module using modern day parts. You have all the design schematics available, someone could design the memory housing with connecting pins, someone else could design a pcb to replicate what the memory module does and then wire it to the connecting pins.
At least this would get around the problem your having of having to try and find an original working memory module. Just build a modern day equivalent, that will at least then allow the system to run, then once it's all working as it should be, then i am sure the computer museum would allow you to use their original memory module.
Really enjoying this series, I was kind of hoping the museum would just lend you the working module to test with. Good luck!
Another idea is using a high voltage/low current signal generator on each end of the broken wire while scanning with the X-ray machine. At some voltage I would expect an arc which should be detectable. There is some risk of causing additional damage but it may be a good option “P”.
Your videos are so bleeping cool! Maximum geek - and that's about the highest compliment I can give you!!!
you can swap broken data bit with parity bit, then disable parity check.
Actually, you wouldn't even have to swap anything. Just replace the broken bit with a parity generator -- it will reconstruct the missing bit from the other data.
I never expected it to be repairable, but that was an impressive attempt at it. The xray machine is so cool I would love to get my hands on that haha. I can't wait to see what you guys come up with, this is a really cool project.
We have that hp spectrum analyser in our lab, best keypad I've ever felt in my life.
Time to get a loom, and start weaving. lol
Worth the wait every time.
Curious marc I love what you do for us and for the history as a whole. And your team is pinnacle in saving that history. Thankyou from just a fly on the wall.
I just love watching your videos. You're great at explaining what you're doing and it's SO fascinating!
This is just pure awwwsome. I wish you guys best luck. Please continue making quality content.
We once did a really really bad hack for getting some broken wire to "work" in a similar situation. The circuit we had was current driven and everything was potted, hence very well insulated. We had someone come up with a high impedance high voltage current source that was driven so high (several hundred volts) but accurately enough that it worked for a while and we could do what we wanted, after a while though it stopped working, despite dialing the voltage always higher, likely the wires eroded.
Someone once then jokingly said that with a high enough voltage the spark would be bright enough to actually see thorugh the potting where the break is...
What an epic journey. And just amazing to see what they achieved and built
So wish had kept the Dsta General Nova minicomputer I had with core memory....the choices of youth eh? Excellent series. Shows how to cope with faults, provide solutions/ plans AND deal with complete failure in an avenue of approach.
You got rid of a DG Nova?!? Aaaargh!!!!
I could totally see the plotting and scheming gears turning in Kens head there at the end of the video 🤣
I'm enjoying watching this series, very interesting and challenging. Although much more complex, this AGC reminds me of Wang's 300 series calculator systems made in the same time period. Love the Tek 7854 scope w/waveform calculator, a classic!
does plan E involve weaving your own core memory module? I imagine the last resort would be emulating the module.
emulation is like having a ultra great classic muscle car with a V8 engine that´s like porn... the engine is broken and instead of fixing it you put in a 2018 4 cylinder diesel engine just to drive the car
@@bobl78 Having said that it is still better then having it not running. I can't imagine them making a core memory module so in the end whatever they come up with will be an emulation.
Would be really cool if the museum agreed to letting you guys use their memory block.
Yeah, I don't see why you couldn't swap them - yours is broken, there's just sits on the wall non-functional anyways. Put the broken one on display, use the good one.
If the goal is to power up and run the Apollo software on the AGC, you could build new wires, ferrite core rings and run them outside the memory module hooked up in parallel to the broken wires.
Chances are the broken wire is only one and you won't have too many new ferrite cores to hook into it.
Could you perhaps piggy-back a small circuit to just those pins to mimic a working section of memory? Another idea perhaps is to fabricate a module to mimic the look and behaviour of the original and use that just for demonstrations. That will leave the original module in good condition and may be easier than repairing it.
I was thinking the same thing as your first suggestion. Since the wire is broken, it should be possible to just bypass it entirely.
Loving this content... just finished the AGC book by Frank O'Brien, reading for the second time... stoked about the 50th anniversary (ongoing, if you look at the whole program). Dumping the historical memory will be extremely interesting!
Doesn‘t the higher dielectric constant of the epoxy resin in comparison to air lead to a much shorter length to failure? I guess the Er is at least 3 to 4 in epoxy so the failure should be nearer to the connector
I second that! And noted immediately while watching.
Love video, hope you can fixed it
I watched it all - what an amazing project and I know understand core memory properly.
Thoroughly enjoying this journey. It will be interesting to see the design and build of a substitute for the non-working AGC. I wonder if you can use modified "donor" core planes from an old IBM or DEC computer.
Wonderful. Hopefully you can borrow/use/trade the museum's AGC module. Keep up the good work.
Would building some kind of emulation for the erasable memory be possible? something that plugs into the physical slot and responds in the expected manner with more modern materials and techniques? It might even be a good proof of concept on retrieving any residual data in the erasable memory (if any, still not 100% understanding how the memory lattice in that works) of both modules.
The problem would probably be power. If the module can be given a constant amount of power (let's say 5v), you could probably use something like a stm32F103C6, dirt cheap microcontroller at 37 GPIO pins with 20kb of ram.
But that's no fun. Lets see a new build of the old design!
01:25 Mike.... Don't bend over like that... sit down man! I've just moved house recently and my body just emphasized with the pain of being bent over like that for long periods!
the world needs your problems.
I can't help but wonder if a solvent would eat the glue, allowing access. Maybe even some way to dissolve the potting?
I'm guessing anything suitable to nom the epoxy will also strip any insulation from the wiring too.
@@TheErador Not necessarily. It would not be easy but I suspect there would be a solvent that would soften the glue and allow access without killing the wire insulation.
That is by definition, destructive. you try not to do that to historical artifacts.
and how do you safely try the 100's of solvents when you only have one and its irreplaceable?
@@thedamnyankee1 That is the 64K question. But IF as in this case there ARE others of this module, then possibly it could be an option worth looking at.
I hope you took the chance to secretly exchange the modules while inspecting them in the museum ;)
ntrasla who would ever be affected is the question? It’s just a visual item being at a museum anyway.
Thank you for the update!
How would you downvote this?
Very impressed with the meticulous efforts!
I love the way you portrait your diagnostics and investigations. Its allmost medical mechanics. (Medichanicism?)
Perhaps try tracing the broken wire with a wire loop antenna?
Feed a couple of 100MHz into pin with broken wire and use small wire loop antenna on HP8566B to trace the RF emission of the wire.
No point since they can't take it apart without destroying it...
The procedure reminds me of the movie 2001 - A Space Odyssey, when Dave and Frank are testing the AE-35 unit.
Stanley Kubrick seemed to be inspired by the AGCs Modules when designing HALs "Logic Memory Center".
I was proud when I troubleshot my toaster. Then I saw this. Immediate humblesub.
I wonder what if heat up or cool down the module with monitoring via connected tester? Is there a chance for temporary restoring the contact by thermal effects? If it happens, you can try to weld the wire by high current through it. But of course it's extremely risky and need some experiments with same wires...
Absolutely amazing video today!!! Keep it up Marc! I guess the next step is rebuilding or emulating the memory? (given you mention samtec sponsoring the connector rebuild). Cheers!!
Do you have access to a 3D x-ray machine?
Is there a way to inject a short wavelength signal on the wire using the estimated length for the optimal frequency and construct a tuned probe for detecting the localized signal strength? Might that help isolate the location of the break?
Apollo memory AND the 7854! Thanks 👍
I tracked down the point of breakage one time with high voltage.
I put the broken wire on a sensor input and a needle with oscillating high voltage to probe it. I used a flatbed plotter to move the needle and wrote a small program to scan the hole board. I didn't had the diagram of the board only the pin-out of the chips.
I doubt it would work with such a tree dimensional construction. The distances are just to great for detection. Maybe the area can be tracked by using the wire as an antenna and get at least the region the signal stops.
This is fantastic!
i love the repair Series of the AGC. Have your think about Chemical desolving the Resin in the Module ? Ich think it is a posible way Also i think you have only another option that calls Rebuilding with old (magnetic core) or new (smd) technic. Eventulay a 3D Xray is a way but i think this machins are not accurate enoth. i wisch you good luck its awsome !!!
Best Regards an All from Germany
The issue would likely be stripping the insulation from those wires with the epoxy.
@@TinkerLynx jes that is a prooblem for a chemicist maybe it is possible. I think the easyst way ist a retrofitted pcb with a microchip tha emulates the core memory. I hope the best and press both thumbs. This ist a awsome piece of history.
@@tollerteppich Ya, I looked into it a bit, an FPGA such as the flash based Intel MAX10 could do the job. The issue is latency, core memory is actually pretty quick. But I think the internal memory on an FPGA can do it. probably under $100 if someone would volunteer there time to do the coding.
@@tollerteppich I'm reading these with a German accent and I'm having a blast :dd
@@TinkerLynx i have not the gackgrund to Programm an fpga. I can a litle bit C on a mega8 XD but i think here are experts on fpga that can help. Also i think with a lot of hand work it is possible to rewire the module after stripping with the original ferite cors
Can you simulate the broken bits externally? i.e. run the module as is - decode where you know bad bits are and use a modern circuit to replace them (bad block mapping style approach)
...and watched to the end - sounds like the above is approach D, E or F :)
Is it not possible to dissolve the podding in some way? Or machine it away... Curious to see what you'll come up with. Maybe some other module will show up or you'll end up with some FPGA substitute, who knows.
Can you run some RF/HV thru it to try and fuse the break back together? Nothing to lose at this point is seems. Seen it done with tube filaments.
You could still fry a bunch of other stuff
Module is already useless, can't even take the cover off without destroying it further... not seeing any other feasable option
WOW! Thanks for Sharing!
Yup, I want to see it running too!
Absolutely fascinating! Thank you for sharing this story!
Great series an very impressive work.
Is there a Chance you could make a video about the Tek 7854 ? It is a fascinating piece of Computer History by itself.
Good idea. It's a long time I haven't done a test equipment video. The Tek 7854 is quite a scope indeed.
It is... got myself one recently :-)
Yay a new video
i wonder if there are any other modules around
Kim był "Bogusław Frackiewicz"?
16:53 I love the path to the directory being displayed. D:\Documents and Settings\Marc\My Documents\Make\Projects\Apollo Guidance Computer\Memory\XRay Pictures.
next is \Projects\Saturn 5 and \projects\lunar module
@11:27 HP-85! Drool! (Well, more drool added to the lake already present).
I wonder if any of the owners of the display-only units would trade a good module for the one with the broken wire.
Love your videos keep them coming please !!
Since the history museum module is just a display, maybe they'll let you swap it with yours. Otherwise, you might try using Methylene Chloride to dissolve the potting. Last resort, maybe Mike can gut your module, build an emulator board, then install it in the module (it would make a great video series). Good Luck.
Je disais hier soir que l'on avait pas eu de nouvelles depuis longtemps :-) Encore une vidéo très intéressante. Vivement la prochaine. Merci à tous.
Plan D: FPGA simulated memory module?
Sounds like a Good Idea. You might be on to something...
Great troubleshooting logic😳😳🧐🧐. Idea: 3d printer for a hardware clone with more young electronics inside? However you are great team 🤗
Give them a sales pitch about how yours is newer. But the one without the potting is a better display piece.
Hahaha. They might see through the marketing smoke and mirrors though...