Congrats! Tape data loading onto the drum! That’s quite an achievement. It’s really neat to see the data build up on the track at the pokey speed it’s read in from the tape. Such a relaxing computing pace.
Thank you! Interestingly, what we're seeing on the scope is incrementing four words at a time, so that's actually 116-bits being transferred over at a time, which is decently fast!
You have the best ability to convey complex computer jargon that I could never understand on my own. I appreciate you saving historical items like this and sharing. Many thanks. Always enjoy watching you.
Thank you so much! The G15 is quite a complex system, even something as simple as loading in paper tape data is a fairly difficult system to understand, so I definitely wanted to take the time and explain it as clearly as I could!
Unfortunately it can't. I spoke with David when he was at Systems Source removing the good drum from the Bendix they have on display. I said the very same thing, "I cant wait for it to print out 'Hellorld'." David said it cant, since the Bendix is a numerical computer only. It can't do alphanumerics. Therefore no letters. The closest he could get would be printing images of the Hellorld letters. I hope he can get that far.
That single scope shot made the entire input system click in my head and really helped me understand what was going on! Also, it was just super cool to see the data push around the drum line!
First let me say: Lloyd, you are a bloody legendary hero to keep coming to restore this old beast from your working days back to life and we totally appreciate you! Grats Dave! PLEASE keep focus on this machine until it's 100% rather then cycling through projects if possible (and if Lloyd's schedule allows) because it's SO close I am all but dancing on my toe-tips here. I cannot believe how invested i've gotten into seeing the Bendix run, it's worse then with the Centurion that got me hooked on your channel in the first place, LOL.
He does need to wait on other people too, plus he needs to get some mental breaks to work on something else or this thing can burn him out a bit. I'm glad it's getting more attention lately, though! Bendix is like the new Centurion, the big project, and after that we can get on with a new big overarching project while the others get to be the fresh air between.
@@UsagiElectric That's not just 10 years! You look like you're in your early 20s almost when clean shaven and in your 50s or later when you're about to shave lol
I've been watching a lot of this and I'd like to thank not just you but Mrs Bunny and all the little bunnies. There's a lot of sacrificing going on making these videos for us and working on this hardware. And I want you to know it's greatly appreciated. 😊
Excellent work. Won't post any spoilers here, but your solutions to problems are innovative, and, more importantly, _functional._ Well done, both of you!
Absolutely fantastic. I was quite Skeptical that it would ever live again, "Bit more then you could chew" type of thing. But you and the community has shown, there is bite to that bark. I am in awe!
Fun fact, an ordinary LED can act as a photo-diode that’s most sensitive to light with the same wavelength as it would emit (does NOT apply to blue/white LEDs). Since that bulb probably emits a good deal of light across a broad portion of the visible spectrum, a yellow SMD LED glued to the end of a light-pipe with the same diameter as the bad photo-diode should work; just keep in mind that LEDs of any color have a low PIV (Peak Inverse Voltage), should you try to use them as photo-diodes in a tube circuit! 😊
I was wondering about that myself, in fact it was the first thing I thought of when the germanium photo-diode was confirmed toast. Not even sure where I heard this fact but it clearly stuck in my subconcious somewhere!
Just gotta say, I love the editing in the crosstalk between Dave and Future-Dave. You are approaching Jedi level Youtubing skillz. The self-deprecating comments are fun, too. 'Uncoordinated monkey' had me snorting my coffee.
"The sun is going down. Lloyd is only here for a limited amount of time."... Message received. Lloyd is a Werewolf! And you need to get stuff done before he transforms from the moon light and wreaks havoc on small villages in Europe. 😂
Seeing the blocks of bits march across the scope as they were shifted onto the number track gave me goosebumps. It's incredible that you've even gotten this far on a machine that is pushing 70 years old! You're a restoration legend, and Lloyd is too for contributing his first-hand experience with this machine! I can't wait for the next video.
Wow wow wow, thought I'd be one of the first to say congratulations, been following this with interest from the start, now I'm watching the full video.
With a function generator, a resistor and your scope, you can create a curve tracer for the diodes. It may be that the breakdown voltage on the bad one is very low. BTW: I have found that a 1N5711 works OK as a replacement for a 1N34 in some circuits. The signals you have a large enough and slow enough that it is likely that the details of the diode don't matter much. The BAT41 is a 100V diode. I think you are only running about 25V on those diodes so that should be OK A conventional silicon diode would likely work for you if you need a higher voltage part
Maybe an octopus curve tracer would be an easy way to test diodes under higher voltages? I'm a little sad to see the beard go, I don't know if you've ever watched the anime Gintama, but with the beard you always make me think of Makoto the mascot. I have a mental image of you as Makoto with a G15 slumped over your back.
Seeing the values populate on the drum and being read back correctly was just so awesome I wished there was a double-like button. What a thrilling experience. Well done!
If you cannot find an exact photo diode you could put a fiber optic cable in its place and run it to another location where you can then use some other photo diode that is larger.
Not a bad idea. You should probably hide that construction behind the metal case, to keep the original appearance intact. But it would definitely solve the size problem, and you could use the more available versions of light-sensitive Germanium diodes.
This is quite an odyssey! But I'll follow your Bendix exploration as long as it ever may take. This machine has something incomparable to it. You're doing a fascinating and invaluable job by restoring this machine, your patience and resilience are admirable! All the best of luck to you and your mates and family!
You probably know this already, but .... The diode test at 27:42 is only testing for opens and shorts. You need to reverse the meter leads to test for leakage. However, it's possible that it needs higher voltage for the leakage to appear. You may need to construct what would be a high voltage ohm meter to test the diode's reverse leakage at 100V (instead of the ~3V your meter uses).
Undoubtedly, your most interesting and captivating video yet. I'll watch again to verify what I thought your troubleshooting was actually fully going. EXCELLENT. Any video where the viewer gets to burn his or her brain energy/thoughts is my kind of video. Reminds me of 3:00 AM nights when I was with DEC and trying to get a system going. No sleep and running on just coffee and a scope probe !!!
🎉 Nice! Maybe the diodes are bad in the sense that their breakdown voltage has decreased to something lower than the one they need to block. So the meter's low voltage test doesn't trigger the failure.
or gone too resistive/leaky in reverse, germanium diodes do have more reverse leakage current than silicon anyway, this wont show up on a multimeter diode test, you need a 'semiconductor analyzer' like a Peak dca55/dca75
I can only imagine how exciting it must have been watching that data buffer build up on the drum. The moment that it turns from a big mess of saturated analogue circuits into a genuine digital computer!
Ordinarily I watch 5 mins on Usagi on Sunday, and the rest on Monday morning during one the first of my day's wretched meetings. Sadly I was so engrossed, I watched it right through to the end! ARGH!!!
Looking at those cards jogged an old memory... I'm pretty sure I encountered one of these machines way in the back of a desert junkyard in the early 1990s. Man, if I could just go back in time and have another look
To test the diodes , you need a curve tracer because with a multimeter you only see one see point and with a curve tracer you will see all the carateristic of the component between a whole area on a I V plan . So you will see if there is leak and very fine detail you can't see with a multimeter .
Thank you for again an exciting episode. It feel like a real thriller to me, with a positive ending every time. I love you man, and am looking forward to the new episode every sunday. Thanks again!
As an electrical engineer...I would love to see you at some point get this to the point of performing the differential equations calcuations you mentioned in the beginning of the video! Even a relatively "simple" one would be amazing to see!
So cool to have friends in museums 😁 It's amazing to see such stuff for the first time ever, and in working condition to boot I have only seen mainframe emulators but this is the real deal, old school
AWESOME!! Now all you need to do is to restore a Tektronix 575 curve tracer and test the diode under load! (This only uses 1/4 the number of vacuum tubes that your Bendix uses!) Good luck!!😁
I'm wondering what future future David will be like, and how him meeting past past David would go like. I find it pretty crazy you're able to figure this old machine out! Can only imagine when this is all fixed up and learn even more... and to show off that knowledge to you before you even saw this machine. Would be fun.
Lol, the concept of a separate I/O system interacting with another that handles actual computation is what I used to make my first "computer" out of multiple RP2040s (each handled a pool of tasks/jobs, one was video/audio, another I/O with keyboard/storage, another with handling computation and syncing up/making everything interact correctly). Kind of reminds me how Commodore just made disk drives basically an entire computer, instead of following Apple's much simpler example at the time. I've streamlined a lot of stuff on my fantasy computer running off RP2040s and hope to shrink it down with the RP2350 based design.