Apollo Guidance Computer Part 20: electroluminescent panel for our DSKY, making an accelerometer

Now that our AGC works, we rush to integrate the splendid electro-luminescent DSKY display that Ben Krasnow made for us. I also build a rudimentary PIPA accelerometer device so we can try to demonstrate the entire landing programs sequence on our AGC before the Apollo 11 landing 50th anniversary festivities.
Some relevant links:
Playlist of the restoration series: • Apollo Guidance Comput...
Block I AGC period documentary: • MIT Science Reporter-"...
Inertial navigation system documentary: • Video
Schematics: www.ibiblio.org/apollo/links....
and: archive.org/details/agc_handb...
Mike's AGC backplane viewer: apolloguidance.computer/200310...
AGC software repo: github.com/virtualagc/virtualagc
The Ultimate Apollo Guidance Computer Talk: • 34C3 - The Ultimate A...

[@AppliedScience]

You all did an amazing job with the restoration! Incredible effort, and also great video documentation. I'm glad my DSKY found a place in your project.

[@Albertkallal]

I had watched that video of yours and appreciated your crazy work and efforts. Can you make a video of how your work wound up hooking up with this project? Quite certain this was not your original goal. An astounding example of accidental work by you now winding up at the right place and time in history! There's a interesting story here even if you don't think so! So, your crazy idea and work of creating this panel from absolute scratch is now found a incredible landing spot!!

[@ScienceANDesign]

Applied Science your DSKY is pure gold. I’m sure a lot of people would like to have one of them on their ceiling, so they can see it before falling asleep... and with a little bit of luck, one can dream about being an astronaut on a mission to the Moon 😊

[@nmccw3245]

Excellent work on the EL display

[@SuperAWaC]

you really outdid yourself on this one ben, outstanding work

[@Albertkallal]

@@ScienceANDesign Or how about a alarm clock in that format. Now what verb to set the wake up alarm? Oh right - when the alarm goes off in the morning it can show 1201! So a alarm clock in this format would be great!

[@kbates666]

I can't understand why these videos are not more popular than they are. Every nerd would love this series!

[@_Synthesize_Me_]

This has been the best computer related series that I’ve ever seen on KZbin, and watching such a complex project get completed over time is incredibly satisfying. Great job to everyone involved and I’m looking forward to more stuff like this from you, Marc!

[@Galaxius2117]

I am just fascinated by the fact that the Apollo Guidance Computer seems to have a mind of its own. I mean.. it gets angry if it doesn't sense stuff, and whatnot!

[@ml.2770]

...at 15 seconds, guidance is internal... This video series has been just fantastic to watch over months and months leading up to the Apollo 11 anniversary. Thank you so much, everyone involved, for making it.

[@canadianman000]

I knew it. I knew we would see Applied Science. And I knew from the beginning you wanted to reverse engineer those gyros for the AGC :)

[@Ghozer]

YAY Still posting, when the previous said "Restoration complete" I was worried there would be no more on the AGC!! :D

[@pauldorman]

Wonderful! I'm such an idiot. Failing to put two and two together I actually suggested pairing the restored AGC with Ben's display in the comments of the previous AGC video. Doh! I was both right, and a complete dope for not considering that the electroluminescent display was intended for the AGC all along!! It looks fantastic. Great work and collaboration. What an achievement!

[@digitalshackonthelane]

Marc. You and your team have done something amazing and important for all our common shared human heritage. You all should be extreamly proud and chuffed about this achievement. The private collector is a pretty awesome guy it seems and he seems to be in it for the good. However...He must be made aware of how important getting this piece of human history running is and how it needs to be premantly and proplerly preserved for the future of all man kind.

[@TheRealColBosch]

He let them take a milling machine and soldering iron to his memory module. I think he fully understands and is on-board with this project.

[@digitalshackonthelane]

@@TheRealColBosch Oh that I do not disagree. My concern in the end is its final preservation state. Fundamentally it's not my business.. I accept that.. I guess I was more publicly commenting on how monumental it is to have this group of guys working on it, get it to a ground opera table state, and to what will or with whom the AGC will finally end up. (More from a curator / museum preservation pov/concern)

[@jlwilliams]

Wow, I feel like I'm 14 years old again, totally geeking out over everything I could read about Apollo. Thanks!!

[@lmiddleman]

Loving all the antique technology... AGC, DSKY, gyro, accelerometer, MS-Windows.

[@UnitSe7en]

I suppose you think they should fire up the SPARC, right?

[@michaelriley1118]

This was a very fascinating video. The micro-controller not being able to keep up was an interesting observation. Thanks for all the detail.

[@jaymercy224]

That's why I like of analog technology. My 1980´s TV goes on 5 times fast as my 2017 TV because it's dont have to read giga byte code before it start working.

[@marknn3]

A properly selected Microchip PIC would probably be able to keep-up, because it can do a lot of timing and pulse generation directly in hardware. Though this will require a deep-dive into the datasheet.

[@tryggbert]

Incredible work done by this skilled team, Was lucky enough some 10 years ago, to se the AGC in an Apollo exhibition in tecnical museem in Stockholm Sweden. Did not understand then, what this big box this was.

[@Cougar333]

"Oh--OK--K-K-K-K. I'm-I'm-I'm coming." - Marc Us as well, Marc.

[@SomeGuyInSandy]

Thank you for illustrating the cause of the 1202 so well! This is such an awesome project! It is history in the making!

[@ScienceANDesign]

Yeah..!! Excellent work! These videos are like candy to me... always want more!! Thank you for sharing this adventure with the world..! Greetings from Buenos Aires..!👏👏

[@dwarf365]

Man, this is so cool! Congratulations on all of your success in bringing back this historical hardware!

[@darkwinter6028]

Cool... just doesn’t begin to describe it. I’m afraid I’m running out of superlatives... 😎😎😎

[@idiotwithasolderingiron]

This has been an incredible journey. I am glad you brought us along. Thank you

[@brettatton]

Ha ha! Now I can't stop hearing Marc as Brandon with a French accent! I love it! "I didn't even get to the relevant conundrum!"

[@rthefish]

Perhaps one of the surviving "Moonwalkers" would like to see it, too

[@johnzaleski5182]

Can we just get these guys a Saturn V so they can simulate everything in real-time? I mean, this stuff is just next-level amazing!!

[@radio655]

What a masterpiece rebuild to honor the 50 year anniversary of mans greatest adventure. Congratulations on the great success.

[@gk10002000]

love the punch card programming. I learned fortran in my first computer class in 1976 using punch cards! you learned to not make any errors because you submitted your program to the computer center and it may run in some batch that day or the next and you did not have time to redo things

[@sbruce899]

I've been watching your progress on the refurbishment of the AGC and I have to say you guys are my heroes. Just think of all the people, time and money that went into the development of that AGC. And you are bringing it back to life. thank you.

[@RolandElliottFirstG]

I was suprised to see Ben delivery it in person.

[@tristshapez]

Shout out to the Galaxy Quest reference :D

[@Anamnesia]

Nice that *_Applied Science_* has gotten involved in the project. I've been watching his channel for a few years now - well before the vinyl/needle scanning electron gun video!

[@volo870]

16th of July. It's the 50th anniversary of Apollo 11 launch. Congratulations!

[@tachelesreden]

Again a great video. The way you explain how the "slow and underpowered" AGC was able to execute realtime data is just brilliant!

[@Kae6502]

This just keeps getting better! Thank you SO much! :)

[@wildstar1063]

Great job, I remember watching the Apollo 11 moon landing as it happened on my grandmothers black-and-white TV. It's great to see someone resurrecting this old tech. Hopefully this will contribute to the accuracy of some of the emulators people have been building.

[@scotty3034]

Just mind blowing how the original engineers came up with this stuff.

[@robertselectronichobbies9507]

Great video! Thanks for the technical explanations. It is just fascinating to see how all this was done.

[@donmoore7785]

Very neat- the pendulum accelerometer is cool and of course so is your simulator. An amazing amount of research was required to achieve what you have.

[@jerryfraley5904]

Um, holy sh*t. I was kinda expecting this meeting, but wow, thanks for making a vision true for so many. You guys are stellar!

[@derekfronek3094]

Fantastic work! Would really like to see if this and project egress could come together, but since they are completely different parts of the overall rocket I don't see how. I have really enjoyed learning about the architecture of this amazing computer, thank you!

[@BaumInventions]

Looks like rocket science ... oh wait... it actually is rocket science :D love it like always

[@UnitSe7en]

Nope. This is electronics. Definitely not rocket science.

[@rkan2]

More like computer science (the software parts from Mike) and some electronics

[@BaumInventions]

Have you ever heard of the concept of a joke? Its when you say something that is some how amusing but not always 100 Precent correct in all ways... Its like bit 15 is stuck and you have to make up your own solution to a incomplete Information. Thats difficult too.

[@johanveltheim4331]

@@BaumInventions Well "This is electronics. Definetely not rocket science." could be considered to be playing on the same joke, at least that's the way I read it. :D

[@UnitSe7en]

@@BaumInventions I understand that you were making a joke, but your punchline was based on an invalid statement, ergo, this thread.

[@kevincozens6837]

This is a great series of videos. It is amazing to see how much work has been put in to restoring the AGC. Fran Blanche had started work on a faithful recreation of a DSKY but stopped work on it after a while as someone else had made one. Perhaps it was the one seen in this video that caused her to stop work on the one she had started. I used to use one of those keypunch machines seen to the right of the one being used in this video at 2:16. Like the person doing the narration I also use ATMEL AVR parts without using Arduino. :)

[@stitcherlives]

Fantastic as always

[@sheep1ewe]

Awsome Ben could came ower, i think he's the exact right person to participate in this.

[@corear]

Marc, tres cool extraordinaire!

[@scowell]

Love this stuff! Perhaps use the uP's counter-timer circuits? It probably doesn't have six of them available!

[@brendoncassidy1659]

So hard not to say Beryllium Sphere after watching Galaxy Quest, isn’t it? Happens to me all the time. ;)

[@MostlyPennyCat]

You should build a complete simulacrum rig for the AGC. Feed it's control outputs into a physical model of the space craft and feed the simulations data back into the AGC. Then have a rendered representation of the LEM landing on the moon. You could even fly it, with controls.

[@MostlyPennyCat]

Oh my god... _you could literally write a modern 3D version of 1979's Lunar Lander by Atari_ With the AGC being offboard simulation code.

[@yorgle]

Love you guys and this series!

[@Albertkallal]

Very cool. The same toggle and count is rather simular to how the cassite inputs work on a apple II. It's absolutely brilliant how that accelerometer works. Just count the pluses or ticks. Even if computer is busy a single pluse from that device will trigger a interrupt and a wee bit of code will run on the AGC. Such interrupt subroutines thus have to be very tight and short code. Amazing that an Arduino has difficulty creating the required pluses for this task. I suggest you use the Arduino tone generator library. This would allow you likely to easily generate the multiple pluses you require here. In place of say code to generate x number of pluses over a given time interval, with the tone library you could send one command and thus generate a steady stream of pluses. Or hand code the assembler code for the Arduino just like those amazing folks did 54 years ago!

[@sinebar]

The technology back then was astounding even by today's standards I think.

[@nmccw3245]

You guys rock.

[@zachz96]

It's done just in time!

[@siberx4]

A note, Marc, that the Arduino library's standard functions for I/O (digitalWrite() and similar) are *very* slow and have substantial overhead. Using interrupts is a good start, but if you aren't already then directly accessing the I/O registers/ports for your microcontroller in question (such as directly writing to PORTA/B/C/D for AVRs) can be SUBSTANTIALLY faster (as in, 20 times or so faster). A 16MHz AVR should be able to toggle I/O pins at multiple megahertz and getting into and servicing the interrupt routine shouldn't take more than a few clock cycles.

[@siberx4]

A bit more reading has revealed that even using arduino's attachInterrupt() routine has excessive overhead (about 45 cycles before the routine starts, very similar to your reported initial delay!). Again here, directly calling the underlying C routines to register an interrupt service routine can shave that down to something like 18 clock cycles: billgrundmann.wordpress.com/2009/03/02/the-overhead-of-arduino-interrupts/

[@CuriousMarc]

Of course. I don't use the (terrible) libraries.

[@AmishFan]

awesome stuff as usual :)

[@littlejason99]

I was hoping for at least one more video!

[@TheBypasser]

Great job! Btw I recall our old discussion of the "LED lightbulbs" and from what I see they look really great, by the video I'd never tell they are "fake" before you opened the device! P.S. do not let the Arduinos pollute such a beauty, better use stock AVRs :) From what I can tell after years with those chips, the ASM code for those can be written in no time (also the instruction set is pretty slim on those), and no compiler, strangely, will ever compete with it (which was pretty much surprising as the new x86/x64 and GPU-targeted compilers can produce very clean code, especially when you type things in carefully).

[@AsbestosMuffins]

could you guys get the AGC to fly a rocket in kerbal space program through your interfaces? scratch that, you can play real space program with this already

[@KingdaToro]

Orbiter would be better for that, using either the AMSO or NASSP addon.

[@Turboy65]

How about a complete Apollo Mission mod pack for KSP?

[@user2C47]

Kerbal Space Program could easily have a mod written for interfacing the AGC, and you can probably find mods that improve physics and allow for an exact replica of the Apollo spacecraft.

[@boris3320]

Incroyable

[@ScottLahteine]

Maybe this is coming a bit on the late side, but there is a FastIO library that is significantly faster at setting pins than the standard Arduino pin routines. We use it in Marlin to get the fastest possible performance in a printer driver from a 16MHz AVR.

[@orinokonx01]

Galaxy Quest reference! Yessss!!!

[@rsmrsm2000]

Amazing

[@tanmaywho]

The first question I was gonna ask was if you are friends with applied science after watching that display in the thumbnail.

[@stuartmcconnachie]

Ben gave the game away when he came down the stairs. Obviously just brought it on Amazon.... 😉 Props for all the hard work Ben. I saw your videos previously, it was not easy making your own EL. For the accelerometer could you not just output either a continuous ‘positive’ or ‘negative’ channel signal, and not bother with doing any of the pulse shaping in the arduino. Then in external hardware logical AND that signal with the incoming AGC pulses to get your required signal pulses? Probably no need for a FPGA then as arduino code can toggle the ‘positive’/‘negative’ signals when the incoming AGC clock is low. That would seem to give you a much longer window for your processing.

[@randalltufts3321]

Oh you little tease you. LOL Waiting with extremely baited breath. Haha Excellent job gentlemen 👍

[@poisonbiting5639]

wow,Looks scientific awesome

[@DKTAz00]

I was wondering what he was going to use that display for ! :D

[@danielprimostuart]

I've "Ben" waiting for it!

[@CG-vd4rh]

Ben is the man

[@DrewNorthup]

Use the right micro + language platform and you won't have timing issues like that. My suggestion: Parallax Propeller in assembler mode, as it has 8 cores each with direct I/O.

[@williambello4089]

You had me at "little holes of goodness" @2:24, (which also is the nickname of my ex-girlfriend). I still don't understand why she left me... :/

[@LeadingEdgeHD]

Did you have a "little pole of goodness?" She probably wanted the "big pole of goodness."

[@PaulJoy]

Amazing work. Would it be possible to create an interface that translates the IMU and thrust response from something like a drone and have the AGC land the drone?

[@ShopperPlug]

0:52 - He brought it in an Amazon box lmfao

[@transit2digital16]

Use an cypress psoc 4 or 5 . These have ARM Kernels plus some sort of fpga , which can be programmed very easy Not only verilog but also like graphical wiring good old SN74..... parts

[@digitalblasphemy1100]

awwwww yeah

[@grassulo]

I am amazed at what you have done making an apollo agc work again but I want to see an alto display bad apple by touhou I know you love anime Marc and if your alto has and an adda (analog to digital and back it's out there!) card you could have the music in 4 bit sound like coleco vision with much faster graphics...

[@mb5058]

The team and the project was just written up in the wsj, Congratulations Marc and team! www.wsj.com/articles/an-apollo-spacecraft-computer-is-brought-back-to-life-11563152761

[@mariodistefano2973]

*FANTASTIC RESTORATION !!!* But what happened to the faulty core memory B12? did you repaired after it? (but maybe I'm just missing something... maybe, again, the language barrier...? )

[@shifter65]

0:50 jaw drops!

[@Eo_Tunun]

I never heard the AGC was so fast at executing its job, it still outperforms a lot of modern hardware. All you ever hear is "It had less memory than a programmable pocket calculator" nowadays.. I'll do the Wayne's World notworthy thing to you lot for the next quarter of an hour. It's due. :oD

[@aut0maticearth]

I just listened to 13 Minutes to the Moon podcast today and there was mention of a Rendezvous Radar being active that contributed to the 1202 alarms. But the alarms only occurred during a 16 68 to find delta h. So then was the full story of the 1202 alarms that the hardware counter couldn’t keep up with both the radar and the pipa? Edit - oh and a follow up question... will you be able to get the electroluminescent display into Carl’s DSKY??

[@yorgle]

That's an excellent series. I just finished it today. :D

[@carlclaunch793]

The problem was a design flaw. When the rendezvous radar was switched to the abort guidance system, a special purpose backup computer in the LM that performs an abort back to orbit, the position data was still routed to the AGC Further, since the radar antenna was using a different and unsynchronized power supply, it was possible for the relationship between the AGC 800 Hz supply and the other 800 Hz supply to be so far out that the analog to digital decoding of the radar antenna position couldn't resolve a position. Instead it flooded the AGC with random position changes at the maximum rate possible for a counter circuit. This added a few crucial percent of CPU load to software that was already running above 90% busy at critical points in the landing. It was not because the radar was on or off. It was because, when the flight plan had them switch the radar from the AGC to the abort system, instead of avoiding an impact on the AGC it actually developed a pathological load. Later missions would zero out the Analog to Digital circuitry when the radar was switched to the backup system. In fact, that was apparently the safest choice - have the radar lock onto the command module and provide the data to the abort guidance system to make an abort even safer. The fix they gave Apollo 11 for liftoff from the moon, to disable the radar circuit breaker, actually had no impact on the problem. If the radar switch was set away from the AGC and the relationship between the power supplies happened to be bad, then the same load would have hit the computer. Ascent presumably ran at a lower CPU load so it might not have ever caused a symptom even with a bad power supply phase relationship.

[@aut0maticearth]

@@carlclaunch793 Thanks for the reply :) Wow! That's incredible that all that was going on and yet an abort was able to be avoided. That priority scheduling really was something.

[@HeathLedgersChemist]

Please let Ben know that I seriously wish to buy one of his EL displays. Seriously. I've messaged him, but not heard back.

[@edgeeffect]

Wouldn't it be great if some private collector somewhere had a real DSKY that they were willing to donate. I mean, your replica is great... but the more real hardware you can use the more cool points the project gets.

[@MrWoohoo]

Would love to see an episode about the historic 1202 alarm.

[@SweetBearCub]

Amy over at Vintage Space did an episode about the alarm(s), and a ton of other Mercury/Gemini/Apollo videos! See her 1202 alarm video here: kzbin.info/www/bejne/oXinYa17l853hrM

[@CuriousMarc]

You bet. It may take more than one.

[@robina.jensen6114]

Instead of using 16 MHz Atmel, you could use Teensy 4.0 board with ARM Cortex-M7 processor at 600 MHz.

[@lorenzo42p]

you can set multiple io pins at the same time on the arduino. would that help?

[@muhdiversity7409]

The game is Lunar Lander

[@andie_pants]

4:29 AC Delco! I wonder if that was made here in Dayton, Ohio.

[@pikachu.922]

I had a feeling this was why he made that

[@craighalpin1917]

Could you possibly attach the output of the oscillating crystal to the computer and use the arduino only as a sort of valve arangment controlled using pulse width? It probably would not be as exact but maybe it could work. I realize that I'm perhaps naive. It would still be required to keep track of pulses but it would not need to generate as many. The initial 3 pulses could be build in hardware... And after those 3 it could switch over to raw pulse input controlled by the arduino... (It would be better on an FPGA but in a pench I think that this could this work) Of course you would not want to risk the computer with this

[@DanafoxyVixen]

Now that the AGC's working, whats its future.? will it end up relegated to be a static display or back into storage? Hopefully it will be powered up and used for more years to come

[@fostercathead]

Did the gyro come with an extra beryllium sphere?

[@USWaterRockets]

This is KSP 1.0 from 1969!

[@roberthayes6329]

It skipped over this minuet to instantly arrive at this moment in time. 🤔

[@roberthayes6329]

All I could think about while Mark was explaining the pulse timing.🤤

[@rich1051414]

Hardware counter where the pulse itself triggers the counter. If you blink, you miss the pulse, and unfortunately, the arduino blinks.

[@b2gills]

I wonder if a Propeller chip could work as a Pipa since it has 8 cores.

[@tomservo5007]

an arduino is fine, the problem is he is using the Arduino IDE to program it ... he needs to switch to Atmel Studio.

[@mzaite]

Tom Servo I’ve been dealing with some of my own Arduino latency issues in an In car data unit. What does IDE do that drags on the Arduino?

[@djmips]

@@tomservo5007 He said he's just using the Atmel chip. 8:15 - Doesn't sound like he's using the Arduino software library / IDE.

[@CuriousMarc]

Of course not. I am not using Arduino anything. It's programmed in the raw. It's just that the underlying Atmel chip is quite slow. Slow at interrupt detection, slow at context switching. The whole thing (with zero program instructions) is already over 2 us. Then all it takes is a few instructions to keep track of pulses count in 16 bits (slow too, multicycle) and my whole time budget is used up.

[@mzaite]

@@eskomattila4468 Ahhh I see, I'll look into that! Thanks!

[@williamsquires3070]

I wonder how the specs for the PIPA compare to a modern 3-axis accelerometer as found in a modern smart phone?

[@Antoniobeta]

Anyone knows with which program they made the graphical interface on the computer?

[@hoppend]

Is the luminecance display actually going to be installed in the DSKY unit?

[@alexandrebustico9691]

For the problem of performance with the microcontroller you use : by the 2010 standard, AVR 8 bit is already an antic technology ! Without the burden of programming an FPGA, a modern 32 bit microcontroller like the 2013' STM32F4 can drive 2 timers (each 4 channels) in PWM mode and DMA to output the needed 6 channels of pulse with the timer accuracy. Ok, would have been useless since landing was perfectly done anyway !

[@SuperAWaC]

are you going to be able to plug it into a simulator on a computer like orbiter or something and have it drive to the moon with no other inputs?

[@ferrari2k]

Who is watching this while having the Apollo 11 livestream running? :D

[@Stepclimb]

7:33 you state 100 PPM is “parts per million”. Even though that is the usual acronym for PPM, I would think the PPM spec in this case is 100 “Pulse Per Meter/sec” of delta V. At the bottom of the drawing sheet you can see the LM spec of 1 pulse per cm/sec of acceleration. This is of course for the LEM. The CSM spec was less accurate, because it didn’t need to be as accurate.

[@lollandster]

That makes sense. 100 pulses per meter/sec is equivalent to 1 pulse per cm/sec. 100 parts per million of 32G (full scale) is equal of 0.319 pulses per cm/sec if my math is correct. Of course, accuracy and resolution is not the same and it is normal to higher resolution than accuracy, but PPM meaning pulses per meter/sec does make more sense in this instance.

[@hackbyteDanielMitzlaff]

*❤❤❤❤❤❤❤❤❤❤❤❤❤* ;)