Hi there! In 1943 the Colossus computers used to aid decryption of the Lorenz cipher at Bletchley Park used photoelectric tape readers. They could read 5 hole tape at 5000 characters per second, with the tape moving at 30 mph. The tape was advanced by pinch wheels rather than driving it with the sprocket holes to avoid tearing the tape. The sprocket holes were also read by photoelectric detectors and was used to clock the whole machine so its operation was completely synchronous with incoming data.
When they ran the tape at 60 mph it disintegrated so they had to slow it down to 30 mph. Colossus had no trouble reading the tape at 60mph though. It is estimated that it could have processed 10,000 cps with ease.
Not just computers: "After amplification, the sound from the microphone was fed to an electromagnet that vibrated a pen with black ink, under which a paper ribbon was stretched. After that, the tape was passed through a photocell, directing light from a powerful lamp onto the paper. The recorded vibrations caused voltage changes at the output of the photocell, were amplified and fed to the loudspeaker, which reproduced the recorded." th-cam.com/video/wMBla0HeNtI/w-d-xo.html
Thank you for the insight! Interestingly, the Colossus uses vacuum photocells, which are quite a bit different than photodiodes. I think we may have two firsts here: the Colossus is the first for an optical tape reader, and the G15 is the first to use photodiodes for an optical tape reader!
"Who on earth let me get my hands on one?!" Well, in my humble opinion, there is simply no safer place in the world for a vintage computer than in your capable hands. You deserve every inch of that piece of hardware!
@@UsagiElectricand the Elliott 402 of 1955 had an optical five-hole tape reader. Heck, even Collossus code-cracking machine from 1943 read its input tapes optically.
A lot of of engineers put their life into designing and constructing these machines - I really love how you bring those old "wonders" back to life. Those engineers are really unsung heroes IMHO.
The original team that designed and built this thing is on another level. I'm only barely scratching the surface of their brilliance, and even 70 years later, they're still taking me to school and teaching me all sorts of new things!
The 2D21 was a very popular tube, somewhat similar electrically to the 2050. It was extensively used in computers, primarily in reader and punch applications. The 5727 is a special version, ruggedized and subjected to 500-hour batch tests to ensure its reliability and precise operating characteristics. You will also find them in electronic metronomes, namely later versions of the Crystalab Metronoma and a number of those produced by Seth Thomas.
I can't trust selenium rectifiers at all, so glad you took them out of circuit I have been working on restoring a super rare piece of arcade equipment that suffered an electrical fire from a selenium rectifier that went up in smoke (glad I didn't have to experience that smell, it happened to the previous owner) and I am many hours into completely rewiring and overhauling it (the entire wiring loom and transformer were destroyed in the fire)
The nice thing about the 1N4007 mod is that since the forward voltage is almost an order of magnitude lower, so will the power dissapation through it. Those big chunky fins on the selenium devices were mainly there b/c their forward voltage drop was so high so they wasted tons of power. I'm enjoying your channel, your enthusiasm for old tech is contagious, and some of the machines you're showing are really amazing!
Hi, I am often baffled by the technology in your videos but always watch, learn and enjoy! The enthusiasm that you bring is entertaining and carries me along, My wife watches with me, mainly for seeing the rabbits and cats at the end! thanks again.
Nice work on the reader; one teeny tiny 1N4007 bypassing the stinkelenium wrecktifier can really do a nice job, doesn't it? Wishing you total success on your journey. It's good to know that Bendix may have been the pioneers of photoelectric tape sensing. I see you appreciate the inconspicuous medical instruments too... Use the Forceps!
Thanks for showing us the next step of this journey! I loved the comment directed at those who want to replace the capacitors at all costs! I'm with you on this, it shouldn't be a routine task!
some early mullard phototdiodes/phototransistors were actually how some early glass encased germanium transistors started off, if they were found to have good light response, they were encapsulated as photocells, if poor, they became a OC70 or OC71 transistor, clear silicone gel for light use, white opaque for transistor, although you can find some transistors with clear silicone gel if you scrape the black paint off
I remember somewhat later cutting the top off a salvaged BC108 with a razor saw to make an NPN phototransistor, because proper phototransistors were expensive and not generally found on scrap / surplus boards. A drop of clear resin or similar was recommended for protection.
Back in the CK-722 era I made a code practice oscillator. With a socket, so I could check various transistors. I discovered that a transistor with its case removed was light sensitive. It would oscillate or not, depending if the light was hitting it.
Hello @UsagiElectric, i caught a minor error in the video, which has gotten me several times with chemical elements, at 4:42 you have the total ore production of Thallium TI which was used for rat poison and insecticides etc and Titanium (Ti) which for the longest time the major producer was the USSR, which if your bendix is using a Ti drum would make it not only pricey, but VERY hard to source without going through intermediary countries, who were friendly with the Soviet Union. It was also a MAJOR issue with trying to build the various versions of the SR-71 family of planes, whose titanium requirements used at times apparently almost all of the USSR supply. Loved the video, thanks for keeping these bits of history alive, and available for all to see. I will likely never be able to travel to see these mainframes due to immune disorders, but your videos scratch my itch for this era of computing.
Good catch! The reason I showed Thalium is because Titanium comes after Thallium alphabetically and I accidentally pulled the wrong page from the PDF in. In my sleep deprived fog of editing, I totally didn't even realize until it was pointed out to me. Whoops! However, I think everyone is slightly off the mark with their timeline of Titanium prices. Everything I've seen shows that by the late 1950s, US production of Titanium was ramping up dramatically. The US produced more Titanium in 1955 alone than Bendix would have used in the entire 8 year run of G15 construction. One thing to remember is that each G15 only used around 45 pounds of Titanium total. Sure, the SR-71 had to source Titanium from overseas, but it would have used several orders of magnitude more Titanium in one plane than Bendix used for the entire G15 construction. Here are the sponge prices for Titanium, and what's most important to note here is that they drop precipitously as the 1950s come to a close. (Source: pubs.usgs.gov/sir/2012/5188/sir2012-5188.pdf ) 1956: 2.75 1957: 2.25 1958: 1.82 1959: 1.60 Our own density measurements line up almost perfectly with using Titanium, and when you consider stability, clearances, thermal properties, etc. etc., Titanium makes perfect sense. The only counter-argument I've heard against Titanium is that it was too rare, but if anyone could have swung it, Bendix absolutely could have!
@UsagiElectric OK good to know, I know bendix was involved with aerospace stuff, but I was under the impression the USA was not a major producer of Ti til later in 60's , either way Bendix would have absolutely been in line to get Ti as their mainframes likey ran calls for military research etc. Thanks again it is great to see such technology lovingly restored by you. Can't wait for the DC switch to be flipped
10:12 those are certainly not electrolytics. They're likely just paper capacitors in a metal case. The right way to do it, is to check if electrical leakage will result in overbiased tubes (leading to destroyed tubes and transformers), and/or if one shorting out will result in burnt out solenoids etc. A little bit of leakage may be tolerable in some applications, but in others it will result in damage. So verify that they're not in some critical DC blocking configuration to the punch solenoids or something. I literally had one short out on me just a week ago, taking an easily replaced resistor with it. A similar one in the same configuration shorted out upon HV testing it. Most often, they don't short - just leak electrically. As a snubber, like i think i see in the circuit, they won't really do damage - just cause a fault as if the contacts are stuck closed. Sometimes smoke, if it's a HV switch.
It appears those are the capacitors used for filtering the power. (Frankly, I think its less about protecting the reader from noise, then about protecting the rest of the computer from the reader.) I would advise replacing them. If they become leaky, they're going to go BANG! or possibly start a fire. Being across the power line, they have virtually unlimited current to do bad things. (or create shock hazards.) They should be replaced with modern X and Y rated caps. This is very similar to replacing "death caps" in vintage guitar amps.
@@russellhltn1396 Yes, any capacitors that are across the line or between line and ground should be replaced with proper X and Y rated capacitors regardless of whether they are working or not. That is a major safety issue.
It did not have the first optical tape reader, but was probably the first to use photo diodes for it. Colossus had an optical tape reader in 1943-1945. Colossus was a set of computers developed by British codebreakers in the years 1943-1945 to assist in the cryptanalysis of the Lorenz cipher.
When I was working as an engineer, our model shop had an old Agie wire EDM that ran off punch tape that was loaded on what resembled film reels. The way it worked was the NC program was ran through a punch machine and the resulting paper punch tape was then loaded onto the reel and the wire EDM made the part. It was all analog but very accurate. They eventually upgraded it to run off I think a Greco box so they could disperse with the punch tape which was prone to damage.
In the mid 70s I used an HP "mini" that had an optical tape reader. It was designed to "NOT' transfer the tape to a 2nd reel but to just dump it on he floor. Amazingly, it wound up with no problems every time!
I have almost no knowledge of any sort of computing tech before the 90's and I find this so immensely fascinating learning about these old monsters from before my time. Love this content!
The 2D21 was also used in the first model (V200) of Seeburg jukeboxes with the 'Tormat' selector system. Tormat was actually a magnetic core memory, in 1955 (!). They went to the 2050 with the next machines. The 2050 tube was used to trip the machine from 'scan' mode into loading a record to play. They used Tormat from 1955 to the late 70's It was a neat system, when it worked, and a bear to fix when it didn't.
I still remember the smell of this kind of old electronics, the thick layers of dust inside and the oil used to smooth the rolling of axes and motors. The non-metric, kind of stainless steel, screws that could only be used there, at that specific spot and nowhere else. Massive plates of metal for the casing within another metal case. Huge connectors that were one of a kind, sometimes I think what got into their minds to produce stuff like that, so heavy, so large, so unliftable.
I can't help but wonder just how much of that price-tag is *labour hours* of fabrication and wiring all those backplanes. There's a lot of work in that big heavy teal beast!
I would say a considerable bit! My guess is that it's about half and half labor/R&D recoup and materials cost. But even then, $250k equivalent materials cost is pretty impressive!
#PUSHTHEGREENBUTTON! The suspense is killing me! I don't know how but I have been sucked into your videos and the way you explain everything is very good. I like videos that don't hide all the troubles they had getting to the point they are at.
It's amazingly small and compact considering most computers filled a room in those days, the innovation is amazing with no expense spared, this was absolute cutting edge technology even down to the optical tape reader and it was certainly built to last, it's incredible how this can be in such good condition after over 65 years, most of which it's been redundant and left to deteriorate with time. I can't wait until the day you get this processing code, still a long way to go, almost certainly with some unexpected challenges ahead, but you'll get there.
This fancy gas-filled tubes actually are called thyratrons - similar to semiconductor brothers - thyristors, acts same way. Also was used as switching devices for large currents. Selenium rectifiers are durable when are sealed, moisture can destroy it within hours. Check coating on these modules, if the paint is intact and whole computer was stored in dry place everything should be fine.
there's a really easy way to see when a machine is made: when the wires are knotted together you can be pretty sure it's pre space age tech, since zip-ties were invented during this era so everything before was beautifully tied together with special wax-wire
Wow, you have a collection of round-tuit's at 8:19. Now you have these, there is no reason to say "when I get a round toit" and many of those jobs that have been waiting can now finally be done :-)
Ferranti computers from the early 1950s Mark I, and Mercury had also optical readers, they were very fast and pretty common in UK. In fact you can even see here in YT a few videos of those days, and see the light from the reader.
I might investigate replacing the vinyl tubing on those rectifiers. Such flexible PVC has plasticizers in it, and as it leaches out some of them can be rather bad for other stuff in the area.
I found your channel a week or 2 ago and sadly i watched the most recent video of the Bendix 15 first so I've been going back and watching all the other ones now! Honestly, ever since learning about Engineering Research Associates in St Paul last year, I've been fascinated with old drum computers!
I would check the ESR and leakage on the capacitors for one. If they are the paper-wax type I would replace them for sure. I would also make sure they are not shorted.
08:45 Ha! I totally misidentified those components. I looked and said to myself, "Those aren't capacitors. They're clearly 93A6 diodes in 66Y2-029 yellow vinyl tubing." But of course I was wrong; they're in _66C2-029 clear_ vinyl tubing!
17:21 The reason the forward voltage drop on the selenium rectifiers is so large is because they're stacked in series - as you can see just by looking at them - to give them the high voltage rating they need in that circuit. If you stacked 6 modern silicon diodes the same way, the voltage drop would be about 3.6V. Of course now you can get modern silicon diodes that can easily handle an Amp, & much higher voltages for just a few cents each, such as a 1N4007. They're tiny enough that you could just solder them next to the old rectifiers & move the top wire, leaving the antiques on place for nostalgia & authenticity's sake. :)
Inspiring as ever. All your babies get the best of care including the electro-mechanical ones. (I know the Bendix isn’t “your” baby but. 😊) Absolutely gotta see System Source when I’m next in USA. No visits planned but … this decade!
Colossus in WWII use for cracking the Lorense code use by the top brass of the Narsi party. Used an optical reader for the punch tape. I think it used CDS cells as diodes had not been invented at the time.
I'd love to see a breakdown of that paper tape, like how many holes, speed, how it loads that onto the ram disk, you know clocking and all that. I'm really enjoying this series BTW.
Back in the day, there was a hack you could do to use an OC71 germanium transistor as a phototransistor, which involved scraping the black paint from the right part of the glass encapsulation to expose the junction to light.
@usagi Electric The code cracking COLOSSUS computer that was built in WWII used an optical tape reader too. From the wiki "A tape transport with an 8-photocell reading mechanism."
When I was exposed to this in high school, they ran a program that played Tic-Tac-Toe on the typewriter. I hope you can find that somewhere. It would be fun to see it happen again.
Well, you are correct in that the banded end of a diode is "more negative" when the diode is conducting. But in both the selenium and silicon diodes the markings are on the "cathode" end. In quotes as they have no actual cathode.
I double/triple checked the diodes were installed correctly according to the Federal datasheet on those specific selenium rectifiers. For whatever reason, the "+" mark on the selenium rectifiers is equivalent to the bar on the diode. Check out the Federal datasheets here: www.worldradiohistory.com/BOOKSHELF-ARH/Technology/Federal-Selenium-Rectifier-2nd-1953.pdf
Not just old rectifiers! Modern bridge rectifier packages are often marked with "AC" or "~" for the legs that connect to the transformer, and "+" and "-" for the DC outputs.
Found very similar looking actual capacitors in an (ultimately deemed beyond economical repair for different reasons) 1960s lab power supply - *they weren't leaking and measured fine* (capacitance, ESR, leakage current at nominal voltage), it was just the insulating plastic that had degraded in an incredibly icky way. Can be slit with a utility razor and removed - though the glue residue on the capacitor body is hard to get off, stoddard solvent and paper towels will get most of it off (likely including the printing :( ) but not quickly, then the parts can get a new heatshrink...
Particularly on a part of the machine that is mechanical so will have some inherent vibration. Plus the machine needs to get transported back when it's working again. Definitely should be insulated IMO.
One thing regarding the RF decoupling filter, those old style capacitors in that unit can fail short and make the chassis live. Might be worth a look. As always a great video.
If anything, I would probably just give that grungy looking sleeving on those diodes and chokes a misting of isopropyl alcohol, just to take care of whatever fungi has taken root in the sleeving. The only small concern with the selenium diode replacement is the difference in the voltage drop (7V on the selenium rectifier vs 1V on the 1N4007). Hopefully other portions of the circuit can compensate; otherwise, R = E/I and P = EI should be what you need to figure the value and size of the series resistance needed to soak up the remaining 6 volts. But otherwise, looks good and good to see things are slowly and steadily moving along on the Bendix! :)
As he said, it is already in series with a big resistor, and since it is used as a flyback diode, clamping the flyback pulse to 1v instead of 7v would be better anyway
As the selenium rectifiers have a voltage drop of several volts you could also have let them in circuit, they wouldn't get any current with the 1N4007 in parallel. Those dangling cables give me the creeps xD
The typewriter looks like a Friden Flexowriter. In the 1960's I used one which was the main control device (apart from the control panel) on an Elliott 803.
The voltage drop on the selenium rectifier is about 7 volts, quite a bit more than the 1.1V max drop (.6 typical) of the 1N4007. I don't know if that is going to matter or not, but it might. Thyratrons are vacuum tube versions of a SCR. Once they switch on, they continue to conduct until the voltage drops below a threshold voltage. The lower voltage drop may alter when they switch off.
Hmm I would put a resistor as well as a diode as in this case it would increase you plate voltage so those 2D21s might not turn off/on as expected. Looking at the Datasheet at 160v plate the operation curves are all quite non-linear all depends on the voltage on the grid (1) if it Negative then you have good linear curve a positive value get you into funky town. The old guys at the radio club swear that the high voltage heavily stressed seleniums rarely blow it is the lower voltage (30) are the ones that go. I would check the voltages first but you are most likely right there is no need for a resistor.
yep, although a lot say they should be replaced on sight as can fail disastrously with a horrid stink, as long as not overloaded and run reasonably cool, too much heat is a big killer for them, they'll likely be ok , but i'd advise fitting a fuse in its feed, if it doesnt have one or a mains input fuse
@@andygozzo72 Over here in the UK the mains plugs have fuses in them, and the motor has a fan fitted with vents right under it. So It's prety well catered for. But yeah, I've heard you can never get rid of the smell. LOL
@@frankowalker4662 i'm in the uk, you cant rely on say a 3 amp plug fuse popping if you get a direct ht short, it wont, it'll explode the rectifier, i very strongly advise fitting a fuse in the feed the rectifier and/or a more suitable current rated fuse on the mains input, 0.315 to 0.5amp, antisurge should be about right, what make/model tape recorder?
@@andygozzo72 It's a Grundig, not sure of the model number. I gave it a recap and replaced some resistors that were out of spec. I kept checking it with my thermal imaging camera and it barely rose above room temp.
19:30 Consider 3D printing it out of flexible filament like Ninjaflex. I have repaired a vintage mechanical calculator with a 3D printed belt and it worked great!
David, the next time you come to NJ for VCF East you should try to get into the Bell Labs facility in Murray Hill. Nokia Bell Labs is building a new home in New Brunswick, NJ and there is no telling what will happen in the original place where transistors and Unix were created.
Aside from the collosus the only other thing I can think of that was using photodiodes at this time frame was the Soviet Luna 3 which used photodiodes to scan the photos it took of the far side of the moon and send them back to earth via radio. That was 1959
Indeed: My parents built the three bedroom split level house that I and my siblings grew up in and paid $26,000 for it in 1960. ~$50k would have bought a lot of house in those days.
@11:55, I'm pretty familiar with motor control circuits and I see a 'DC Braking' supply right there in the middle. So when both F and R relays are off (via series normally-closed contacts), it applies DC power to one of the windings in each motor. This is commonly used to make the induction motor quickly stop. But I'm confused a bit since this DC power is only supplied with incoming AC when one of the nomrally-open F or R contacts are closed. So I'm wondering if those first F/ R contacts have some time-delay dropout or such?? Also that switch in the upper left, only controls the contacts on the tubes along the left-hand side? Something about those relays and that switch just aren't very clear.
personally i'd say those selenuim diodes wouldnt need replacing, going by the rating of those series resistors, theres not that much current passing in them, theyre more likely to fail under ac power rectification use during the reverse part of the cycle..
if you really must replace them , try to track down some early 'top hat' silicon diodes from the early 60s like BY100, BY105, etc. or whatever equivalents you had over there! as they wouldve likely been used as service replacements back in the day ...
That's because Titanium comes after Thallium alphabetically and I accidentally pulled the wrong page from the PDF in. In my sleep deprived fog of editing, I totally didn't even realize until it was pointed out to me. Whoops!
@@UsagiElectric Great work , please keep it up. And you sure would not want to be handling Thallium. Its very poisonous and its use featured in some high profile murders in the UK in the early 1970 - the Bovingdon Teacup Poisoner Etc. My Chemistry teacher lived in Bovingdon so we were always careful around him!
Was about to mention that! Edit: someone mentioned that and Usagi answered, the selenium diodes have the + markings on the cathode end of the component, he linked a datasheet too.
Hi there! In 1943 the Colossus computers used to aid decryption of the Lorenz cipher at Bletchley Park used photoelectric tape readers. They could read 5 hole tape at 5000 characters per second, with the tape moving at 30 mph. The tape was advanced by pinch wheels rather than driving it with the sprocket holes to avoid tearing the tape. The sprocket holes were also read by photoelectric detectors and was used to clock the whole machine so its operation was completely synchronous with incoming data.
en.wikipedia.org/wiki/Colossus_computer
When they ran the tape at 60 mph it disintegrated so they had to slow it down to 30 mph. Colossus had no trouble reading the tape at 60mph though. It is estimated that it could have processed 10,000 cps with ease.
Not just computers: "After amplification, the sound from the microphone was fed to an electromagnet that vibrated a pen with black ink, under which a paper ribbon was stretched. After that, the tape was passed through a photocell, directing light from a powerful lamp onto the paper. The recorded vibrations caused voltage changes at the output of the photocell, were amplified and fed to the loudspeaker, which reproduced the recorded." th-cam.com/video/wMBla0HeNtI/w-d-xo.html
Thank you for the insight!
Interestingly, the Colossus uses vacuum photocells, which are quite a bit different than photodiodes. I think we may have two firsts here: the Colossus is the first for an optical tape reader, and the G15 is the first to use photodiodes for an optical tape reader!
@@UsagiElectricI almost wonder if there would be an in between tech that used selenium cells since those got a lot of use for camera meters.
"Who on earth let me get my hands on one?!" Well, in my humble opinion, there is simply no safer place in the world for a vintage computer than in your capable hands. You deserve every inch of that piece of hardware!
Thank you so much!
Hopefully it all comes up smoothly and we can get this beast back up to System Source and let museum-goers get hands on with it!
@@UsagiElectric💯 agree
And we'll happily be along for the ride, with a fresh batch of popcorn at the ready. :D Best of luck! @@UsagiElectric
Australia's first computer, CSIRAC, built in 1949 was fitted (around 1951) with a 12-hole optical paper tape reader using photodiodes.
Oooh, excellent catch!
That's a machine I know almost nothing about, but that's awesome!
@@UsagiElectricand the Elliott 402 of 1955 had an optical five-hole tape reader. Heck, even Collossus code-cracking machine from 1943 read its input tapes optically.
A lot of of engineers put their life into designing and constructing these machines - I really love how you bring those old "wonders" back to life.
Those engineers are really unsung heroes IMHO.
Think also about the people who assembled these machines. Strung all that wire. So very tedious. They were artists.
The original team that designed and built this thing is on another level. I'm only barely scratching the surface of their brilliance, and even 70 years later, they're still taking me to school and teaching me all sorts of new things!
The 2D21 was a very popular tube, somewhat similar electrically to the 2050. It was extensively used in computers, primarily in reader and punch applications. The 5727 is a special version, ruggedized and subjected to 500-hour batch tests to ensure its reliability and precise operating characteristics. You will also find them in electronic metronomes, namely later versions of the Crystalab Metronoma and a number of those produced by Seth Thomas.
I can't trust selenium rectifiers at all, so glad you took them out of circuit
I have been working on restoring a super rare piece of arcade equipment that suffered an electrical fire from a selenium rectifier that went up in smoke (glad I didn't have to experience that smell, it happened to the previous owner) and I am many hours into completely rewiring and overhauling it (the entire wiring loom and transformer were destroyed in the fire)
The nice thing about the 1N4007 mod is that since the forward voltage is almost an order of magnitude lower, so will the power dissapation through it. Those big chunky fins on the selenium devices were mainly there b/c their forward voltage drop was so high so they wasted tons of power. I'm enjoying your channel, your enthusiasm for old tech is contagious, and some of the machines you're showing are really amazing!
I love the way you highlight schematics and components in bright colors and smooth animations! Wish more people did this (assuming they know how to).
Hi, I am often baffled by the technology in your videos but always watch, learn and enjoy! The enthusiasm that you bring is entertaining and carries me along, My wife watches with me, mainly for seeing the rabbits and cats at the end! thanks again.
Is there a rabbit channel for her to watch?
Nice work on the reader; one teeny tiny 1N4007 bypassing the stinkelenium wrecktifier can really do a nice job, doesn't it? Wishing you total success on your journey. It's good to know that Bendix may have been the pioneers of photoelectric tape sensing.
I see you appreciate the inconspicuous medical instruments too... Use the Forceps!
Thanks for showing us the next step of this journey! I loved the comment directed at those who want to replace the capacitors at all costs! I'm with you on this, it shouldn't be a routine task!
some early mullard phototdiodes/phototransistors were actually how some early glass encased germanium transistors started off, if they were found to have good light response, they were encapsulated as photocells, if poor, they became a OC70 or OC71 transistor, clear silicone gel for light use, white opaque for transistor, although you can find some transistors with clear silicone gel if you scrape the black paint off
Ha! I'm glad I'm not the one who remembers that trick! Wow, that's going back a very long time.
I remember somewhat later cutting the top off a salvaged BC108 with a razor saw to make an NPN phototransistor, because proper phototransistors were expensive and not generally found on scrap / surplus boards. A drop of clear resin or similar was recommended for protection.
Back in the CK-722 era I made a code practice oscillator. With a socket, so I could check various transistors. I discovered that a transistor with its case removed was light sensitive. It would oscillate or not, depending if the light was hitting it.
"Hop to it," on a channel called Usagi (rabbit in Japanese) electric. I see what you did there, and i love it!
It's only a matter of time before you get a Cray. Calling it.
X-1 or X-MP? Or both?
Hello @UsagiElectric, i caught a minor error in the video, which has gotten me several times with chemical elements, at 4:42 you have the total ore production of Thallium TI which was used for rat poison and insecticides etc and Titanium (Ti) which for the longest time the major producer was the USSR, which if your bendix is using a Ti drum would make it not only pricey, but VERY hard to source without going through intermediary countries, who were friendly with the Soviet Union. It was also a MAJOR issue with trying to build the various versions of the SR-71 family of planes, whose titanium requirements used at times apparently almost all of the USSR supply.
Loved the video, thanks for keeping these bits of history alive, and available for all to see. I will likely never be able to travel to see these mainframes due to immune disorders, but your videos scratch my itch for this era of computing.
Good catch!
The reason I showed Thalium is because Titanium comes after Thallium alphabetically and I accidentally pulled the wrong page from the PDF in. In my sleep deprived fog of editing, I totally didn't even realize until it was pointed out to me. Whoops!
However, I think everyone is slightly off the mark with their timeline of Titanium prices. Everything I've seen shows that by the late 1950s, US production of Titanium was ramping up dramatically. The US produced more Titanium in 1955 alone than Bendix would have used in the entire 8 year run of G15 construction. One thing to remember is that each G15 only used around 45 pounds of Titanium total. Sure, the SR-71 had to source Titanium from overseas, but it would have used several orders of magnitude more Titanium in one plane than Bendix used for the entire G15 construction.
Here are the sponge prices for Titanium, and what's most important to note here is that they drop precipitously as the 1950s come to a close.
(Source: pubs.usgs.gov/sir/2012/5188/sir2012-5188.pdf )
1956: 2.75
1957: 2.25
1958: 1.82
1959: 1.60
Our own density measurements line up almost perfectly with using Titanium, and when you consider stability, clearances, thermal properties, etc. etc., Titanium makes perfect sense. The only counter-argument I've heard against Titanium is that it was too rare, but if anyone could have swung it, Bendix absolutely could have!
@UsagiElectric OK good to know, I know bendix was involved with aerospace stuff, but I was under the impression the USA was not a major producer of Ti til later in 60's , either way Bendix would have absolutely been in line to get Ti as their mainframes likey ran calls for military research etc.
Thanks again it is great to see such technology lovingly restored by you. Can't wait for the DC switch to be flipped
""it needs to be cleaned, and all the spinning bits oiled" - quote of the week!!
10:12 those are certainly not electrolytics. They're likely just paper capacitors in a metal case. The right way to do it, is to check if electrical leakage will result in overbiased tubes (leading to destroyed tubes and transformers), and/or if one shorting out will result in burnt out solenoids etc.
A little bit of leakage may be tolerable in some applications, but in others it will result in damage. So verify that they're not in some critical DC blocking configuration to the punch solenoids or something.
I literally had one short out on me just a week ago, taking an easily replaced resistor with it. A similar one in the same configuration shorted out upon HV testing it. Most often, they don't short - just leak electrically.
As a snubber, like i think i see in the circuit, they won't really do damage - just cause a fault as if the contacts are stuck closed. Sometimes smoke, if it's a HV switch.
It appears those are the capacitors used for filtering the power. (Frankly, I think its less about protecting the reader from noise, then about protecting the rest of the computer from the reader.) I would advise replacing them. If they become leaky, they're going to go BANG! or possibly start a fire. Being across the power line, they have virtually unlimited current to do bad things. (or create shock hazards.) They should be replaced with modern X and Y rated caps. This is very similar to replacing "death caps" in vintage guitar amps.
@@russellhltn1396 Yes, any capacitors that are across the line or between line and ground should be replaced with proper X and Y rated capacitors regardless of whether they are working or not. That is a major safety issue.
Just to be clear, I'm fine with leaving caps like in tone circuits where there's no real damage if they fail. But this isn't one of those cases.
It did not have the first optical tape reader, but was probably the first to use photo diodes for it.
Colossus had an optical tape reader in 1943-1945. Colossus was a set of computers developed by British codebreakers in the years 1943-1945 to assist in the cryptanalysis of the Lorenz cipher.
Good catch! The Colossus was pretty epic in what it was capable of at the time!
@@UsagiElectric Charles Forbin never should have built it.
When I was working as an engineer, our model shop had an old Agie wire EDM that ran off punch tape that was loaded on what resembled film reels. The way it worked was the NC program was ran through a punch machine and the resulting paper punch tape was then loaded onto the reel and the wire EDM made the part. It was all analog but very accurate. They eventually upgraded it to run off I think a Greco box so they could disperse with the punch tape which was prone to damage.
Colossus's paper tape reader was optical using photocells... December 1943. (The "bed-stead" was the nickname for the tape loop reader.)
But was the Bendix the first using photodiodes?
Love your channel. Love this project.
1958 computing never looked so possible.
In the mid 70s I used an HP "mini" that had an optical tape reader. It was designed to "NOT' transfer the tape to a 2nd reel but to just dump it on he floor. Amazingly, it wound up with no problems every time!
I have almost no knowledge of any sort of computing tech before the 90's and I find this so immensely fascinating learning about these old monsters from before my time.
Love this content!
i love the fact it needs a hopper to collect all the punched holes
The 2D21 was also used in the first model (V200) of Seeburg jukeboxes with the 'Tormat' selector system. Tormat was actually a magnetic core memory, in 1955 (!). They went to the 2050 with the next machines. The 2050 tube was used to trip the machine from 'scan' mode into loading a record to play. They used Tormat from 1955 to the late 70's It was a neat system, when it worked, and a bear to fix when it didn't.
I still remember the smell of this kind of old electronics, the thick layers of dust inside and the oil used to smooth the rolling of axes and motors. The non-metric, kind of stainless steel, screws that could only be used there, at that specific spot and nowhere else. Massive plates of metal for the casing within another metal case. Huge connectors that were one of a kind, sometimes I think what got into their minds to produce stuff like that, so heavy, so large, so unliftable.
I can't help but wonder just how much of that price-tag is *labour hours* of fabrication and wiring all those backplanes. There's a lot of work in that big heavy teal beast!
I would say a considerable bit! My guess is that it's about half and half labor/R&D recoup and materials cost. But even then, $250k equivalent materials cost is pretty impressive!
#PUSHTHEGREENBUTTON! The suspense is killing me! I don't know how but I have been sucked into your videos and the way you explain everything is very good. I like videos that don't hide all the troubles they had getting to the point they are at.
Such a neat project but I have to admit: I'm super happy the bunnies are doing OK!
I am amazed that you can find schematic diagrams that are from so long ago. That is some awesome research skills.
It's amazingly small and compact considering most computers filled a room in those days, the innovation is amazing with no expense spared, this was absolute cutting edge technology even down to the optical tape reader and it was certainly built to last, it's incredible how this can be in such good condition after over 65 years, most of which it's been redundant and left to deteriorate with time. I can't wait until the day you get this processing code, still a long way to go, almost certainly with some unexpected challenges ahead, but you'll get there.
I've had the dubious pleasure of smelling a burned-up Se rectifier. It was pretty unpleasant but not enough to drive me from the room.
This fancy gas-filled tubes actually are called thyratrons - similar to semiconductor brothers - thyristors, acts same way. Also was used as switching devices for large currents.
Selenium rectifiers are durable when are sealed, moisture can destroy it within hours. Check coating on these modules, if the paint is intact and whole computer was stored in dry place everything should be fine.
Awesome work, looking forward to engaging DC!
2024 is the year of the Bendix!
I'm loving this project so far and I can't wait for the next episode!
Man, what a journey this is, can't wait for the moment when that beast of a machine comes fully alive! Cheers from Finland!
Wrote my first program on a G15D Pomona College 1958
there's a really easy way to see when a machine is made: when the wires are knotted together you can be pretty sure it's pre space age tech, since zip-ties were invented during this era so everything before was beautifully tied together with special wax-wire
Wow, you have a collection of round-tuit's at 8:19. Now you have these, there is no reason to say "when I get a round toit" and many of those jobs that have been waiting can now finally be done :-)
Those are some beautiful schematics. Someone really cared.
Looking forward to the big switch on. It's a great series to watch.
Ferranti computers from the early 1950s Mark I, and Mercury had also optical readers, they were very fast and pretty common in UK. In fact you can even see here in YT a few videos of those days, and see the light from the reader.
I might investigate replacing the vinyl tubing on those rectifiers. Such flexible PVC has plasticizers in it, and as it leaches out some of them can be rather bad for other stuff in the area.
Your movies are like a long and everlasting journey and you let us be part of it. Thank you :-)
You should nickname the reader "Judy", so you could have your own Punch and Judy show.
When my friends daughter Judy got the first bicycle that she didn't have to share with her brother it got an utter thrashing. It was 100% Judy cycle.
Hehe, I'll forever think of Punch and Judy every time I look at the paper tape punch.
🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣🤣
I found your channel a week or 2 ago and sadly i watched the most recent video of the Bendix 15 first so I've been going back and watching all the other ones now!
Honestly, ever since learning about Engineering Research Associates in St Paul last year, I've been fascinated with old drum computers!
I would check the ESR and leakage on the capacitors for one. If they are the paper-wax type I would replace them for sure. I would also make sure they are not shorted.
08:45 Ha! I totally misidentified those components. I looked and said to myself, "Those aren't capacitors. They're clearly 93A6 diodes in 66Y2-029 yellow vinyl tubing." But of course I was wrong; they're in _66C2-029 clear_ vinyl tubing!
Why not just test those rectifiers before replacing?
You get the bad smelling smoke if the test fails, there is no pretest for that.
Incredible mechanical work you do, absolutely fascinating.
Man, this Bendix machine is so cool. Thank you for introducing me to the world of early computing. It's all so interesting.
Selenium rectifiers are just about the coolest looking components ever.
theyre still made for certain applications ,
@@andygozzo72 wow like what applications?
@@supercompooper battery charging mainly, as their higher internal resistance helps to limit and partially regulate current
Nice work! I suggest putting heat-shrink on those free-hanging rectifier wires, in case they end up touching anything nearby.
i love the G15 series how after so many months you slowly revive this beast
17:21 The reason the forward voltage drop on the selenium rectifiers is so large is because they're stacked in series - as you can see just by looking at them - to give them the high voltage rating they need in that circuit. If you stacked 6 modern silicon diodes the same way, the voltage drop would be about 3.6V. Of course now you can get modern silicon diodes that can easily handle an Amp, & much higher voltages for just a few cents each, such as a 1N4007. They're tiny enough that you could just solder them next to the old rectifiers & move the top wire, leaving the antiques on place for nostalgia & authenticity's sake. :)
awesome series!, can't wait for the DC bus power up!
Inspiring as ever. All your babies get the best of care including the electro-mechanical ones. (I know the Bendix isn’t “your” baby but. 😊)
Absolutely gotta see System Source when I’m next in USA. No visits planned but … this decade!
Colossus in WWII use for cracking the Lorense code use by the top brass of the Narsi party. Used an optical reader for the punch tape. I think it used CDS cells as diodes had not been invented at the time.
I'd love to see a breakdown of that paper tape, like how many holes, speed, how it loads that onto the ram disk, you know clocking and all that. I'm really enjoying this series BTW.
Interesting to realize that time, the tape reader was state-of-the-art futuristic technology. I can imagine the engineer bragging about it.
Can't wait to see some program running on this beast! That would be so epic!
Back in the day, there was a hack you could do to use an OC71 germanium transistor as a phototransistor, which involved scraping the black paint from the right part of the glass encapsulation to expose the junction to light.
Love the cotton tails. Ohh really like circuit sim use too. But then cotton tails. That's epic !
its very cool just to see this stuff, and how it was constructed!
Totally agree!! Only change a component when it’s failed otherwise your are just creating waste
They sure used a lot of slot screws in those days!
Great presentation as usual.
@usagi Electric The code cracking COLOSSUS computer that was built in WWII used an optical tape reader too. From the wiki "A tape transport with an 8-photocell reading mechanism."
I love the channel - but I’m so impatient. I hope the Bendix becomes operational in my lifetime.
When I was exposed to this in high school, they ran a program that played Tic-Tac-Toe on the typewriter. I hope you can find that somewhere. It would be fun to see it happen again.
Did you solder the diodes the right way round?
Where the plus sign is on the Selenium diode you have soldered the negative pole of the diode.
Well, you are correct in that the banded end of a diode is "more negative" when the diode is conducting. But in both the selenium and silicon diodes the markings are on the "cathode" end. In quotes as they have no actual cathode.
I double/triple checked the diodes were installed correctly according to the Federal datasheet on those specific selenium rectifiers. For whatever reason, the "+" mark on the selenium rectifiers is equivalent to the bar on the diode.
Check out the Federal datasheets here: www.worldradiohistory.com/BOOKSHELF-ARH/Technology/Federal-Selenium-Rectifier-2nd-1953.pdf
@@UsagiElectric yep, the + sign on old rectifiers usually indicates the output polarity
Not just old rectifiers! Modern bridge rectifier packages are often marked with "AC" or "~" for the legs that connect to the transformer, and "+" and "-" for the DC outputs.
I was thinking the same. Good thing I read through the comments before posting 😅
My dogs tail goes bonkers when he sees the Chewy box. Sorry about the vase. Cool viddy so far!
Found very similar looking actual capacitors in an (ultimately deemed beyond economical repair for different reasons) 1960s lab power supply - *they weren't leaking and measured fine* (capacitance, ESR, leakage current at nominal voltage), it was just the insulating plastic that had degraded in an incredibly icky way. Can be slit with a utility razor and removed - though the glue residue on the capacitor body is hard to get off, stoddard solvent and paper towels will get most of it off (likely including the printing :( ) but not quickly, then the parts can get a new heatshrink...
I don't know, it's that "wire twisted safely out of the way" safe enough? It looks so near the leg of the new diode...
Particularly on a part of the machine that is mechanical so will have some inherent vibration. Plus the machine needs to get transported back when it's working again. Definitely should be insulated IMO.
One thing regarding the RF decoupling filter, those old style capacitors in that unit can fail short and make the chassis live. Might be worth a look.
As always a great video.
If anything, I would probably just give that grungy looking sleeving on those diodes and chokes a misting of isopropyl alcohol, just to take care of whatever fungi has taken root in the sleeving. The only small concern with the selenium diode replacement is the difference in the voltage drop (7V on the selenium rectifier vs 1V on the 1N4007). Hopefully other portions of the circuit can compensate; otherwise, R = E/I and P = EI should be what you need to figure the value and size of the series resistance needed to soak up the remaining 6 volts. But otherwise, looks good and good to see things are slowly and steadily moving along on the Bendix! :)
As he said, it is already in series with a big resistor, and since it is used as a flyback diode, clamping the flyback pulse to 1v instead of 7v would be better anyway
If it is bendix that drive belt is probably stock from a washing machine
I'm seriously dying to see this beast run! 👍
So you didn't add a dropping resistor to the 1N4007? Remember, that old rectifier has a more dropping voltage compare to the 1N4007/.
As the selenium rectifiers have a voltage drop of several volts you could also have let them in circuit, they wouldn't get any current with the 1N4007 in parallel.
Those dangling cables give me the creeps xD
Stunning project!
22:07 Nice job! Though you might want to slip some heat shrink over the dangly wires remaining exposed.
The typewriter looks like a Friden Flexowriter. In the 1960's I used one which was the main control device (apart from the control panel) on an Elliott 803.
The voltage drop on the selenium rectifier is about 7 volts, quite a bit more than the 1.1V max drop (.6 typical) of the 1N4007. I don't know if that is going to matter or not, but it might. Thyratrons are vacuum tube versions of a SCR. Once they switch on, they continue to conduct until the voltage drops below a threshold voltage. The lower voltage drop may alter when they switch off.
Can't wait for DC!
Hmm I would put a resistor as well as a diode as in this case it would increase you plate voltage so those 2D21s might not turn off/on as expected. Looking at the Datasheet at 160v plate the operation curves are all quite non-linear all depends on the voltage on the grid (1) if it Negative then you have good linear curve a positive value get you into funky town.
The old guys at the radio club swear that the high voltage heavily stressed seleniums rarely blow it is the lower voltage (30) are the ones that go.
I would check the voltages first but you are most likely right there is no need for a resistor.
I bought a reel to reel recorder last year that had a selenium rectifier dated 1957, and it still works perfectly. They are pretty robust.
yep, although a lot say they should be replaced on sight as can fail disastrously with a horrid stink, as long as not overloaded and run reasonably cool, too much heat is a big killer for them, they'll likely be ok , but i'd advise fitting a fuse in its feed, if it doesnt have one or a mains input fuse
@@andygozzo72 Over here in the UK the mains plugs have fuses in them, and the motor has a fan fitted with vents right under it. So It's prety well catered for. But yeah, I've heard you can never get rid of the smell. LOL
@@frankowalker4662 i'm in the uk, you cant rely on say a 3 amp plug fuse popping if you get a direct ht short, it wont, it'll explode the rectifier, i very strongly advise fitting a fuse in the feed the rectifier and/or a more suitable current rated fuse on the mains input, 0.315 to 0.5amp, antisurge should be about right, what make/model tape recorder?
@@andygozzo72 It's a Grundig, not sure of the model number. I gave it a recap and replaced some resistors that were out of spec. I kept checking it with my thermal imaging camera and it barely rose above room temp.
@@frankowalker4662 grundigs usually had internal fuses anyway i have a TK20, used to have a couple of others but sold them some years back
19:30 Consider 3D printing it out of flexible filament like Ninjaflex. I have repaired a vintage mechanical calculator with a 3D printed belt and it worked great!
I recently bought a kingdom kt-70PC terminal, and it just came in today. Bought manual separate, and both came in today
The PDP-12 might just be the most beautiful computer ever. Perhaps tied with the PDP-15.
Will be cool to see those capacitors releasing the magic smoke 😁😁
David, the next time you come to NJ for VCF East you should try to get into the Bell Labs facility in Murray Hill. Nokia Bell Labs is building a new home in New Brunswick, NJ and there is no telling what will happen in the original place where transistors and Unix were created.
Aside from the collosus the only other thing I can think of that was using photodiodes at this time frame was the Soviet Luna 3 which used photodiodes to scan the photos it took of the far side of the moon and send them back to earth via radio. That was 1959
"Thyrotron".... tube version of an SCR. :)
$49500 for the basic computer! I think, this was the price for a luxury house in the US in the 1950s.
Indeed: My parents built the three bedroom split level house that I and my siblings grew up in and paid $26,000 for it in 1960. ~$50k would have bought a lot of house in those days.
@11:55, I'm pretty familiar with motor control circuits and I see a 'DC Braking' supply right there in the middle. So when both F and R relays are off (via series normally-closed contacts), it applies DC power to one of the windings in each motor. This is commonly used to make the induction motor quickly stop. But I'm confused a bit since this DC power is only supplied with incoming AC when one of the nomrally-open F or R contacts are closed.
So I'm wondering if those first F/ R contacts have some time-delay dropout or such?? Also that switch in the upper left, only controls the contacts on the tubes along the left-hand side? Something about those relays and that switch just aren't very clear.
personally i'd say those selenuim diodes wouldnt need replacing, going by the rating of those series resistors, theres not that much current passing in them, theyre more likely to fail under ac power rectification use during the reverse part of the cycle..
if you really must replace them , try to track down some early 'top hat' silicon diodes from the early 60s like BY100, BY105, etc. or whatever equivalents you had over there! as they wouldve likely been used as service replacements back in the day ...
That price list was for Thallium, not Titanium. A very different metal.
That's because Titanium comes after Thallium alphabetically and I accidentally pulled the wrong page from the PDF in. In my sleep deprived fog of editing, I totally didn't even realize until it was pointed out to me. Whoops!
@@UsagiElectric Great work , please keep it up. And you sure would not want to be handling Thallium. Its very poisonous and its use featured in some high profile murders in the UK in the early 1970 - the Bovingdon Teacup Poisoner Etc. My Chemistry teacher lived in Bovingdon so we were always careful around him!
With all of this relay madness, the EMI filter may protect the logic of the computer from the paper tape reader, not the other way around...
That machine will run some code before it's even 70 years old.
I'm worried that you've put the new diodes around the wrong way. Hopefully everything is all good there. All the best, and I love your work.
Was about to mention that!
Edit: someone mentioned that and Usagi answered, the selenium diodes have the + markings on the cathode end of the component, he linked a datasheet too.