Inside the digital clock from a Soyuz spacecraft
righto.com
righto.com
Because soviet spacecraft, as with all spacecraft at that time, were designed to survive all sorts of failures. The clock is an absolutely essential part of the spacecraft. It needs to keep running even when everything else is failing, especially when everything else is failing. A space clock needs to be both integrated into a dozen other systems, and able to cut itself from those systems when necessary. Cosmonauts in a failing capsule, waiting to fire the return burn necessary to get home, won't be looking out the window. They will be looking at that clock.
It's worth pointing out that flatpacks (both ceramic and EP) were and probably still are a mainstay of military and aero electronics. These were never used in consumer electronics. Apart from microprocessors and EEPROMs, consumer electronics never really used ceramic or hermetic metal packages for cost reasons. Perhaps the most common components would be TO-3 power transistors and small metal can transistors before TO-92 and similar packages obsoleted all of those.
> Many of the components in the power supply look different from American components. While American resistors are usually labeled with colored bands, the Russian resistors are green cylinders with their values printed on them.
High grade or high precision resistors usually had their value printed on them, though. Meanwhile, only larger SMD resistors have markings today (I think they stop applying them below 0604 imperial).
> The Russian diodes have orange rectangular packages (below), unlike the usual cylindrical American diodes.
Semiconductor packages were all over the place in the past; I've seen cube-ish moulded diodes, resistors and capacitors in European stuff.
Overall this thing looks a lot like something from the mid 60s, not so much mid 80s. In that case, American stuff from the same period looks pretty similar, quite possibly due to copious copying by the Russians.
> One nice thing about Russian ICs is that the part numbers are assigned according to a rational system, unlike the essentially random numbering of American integrated circuits.
On a related note, I really like IEC/ISO schematics for this reason, because we have a graphical language to describe logic and this means logic devices appear as a composition of symbols which explain the function of the gate to anyone who knows this language. On American schematics only the most basic gates (AND, OR, NOT, ...) have symbols, everything more complicated than that is generally drawn as a box with the part number ('193) in it and the pins just labelled with their abbreviations.
One nice thing about Russian ICs is that the part numbers are assigned according to a rational system, unlike the essentially random numbering of American integrated circuits.
This makes so much sense, it's a shame this style of labeling didn't catch on. It would have made my EE labs in college so much easier.
I actually sorted that pile once for fasteners, but resistors are so cheap...
The fact that these consumer electronics are built according to the American way as opposed to the Russian way - is that a reflection of selling to a U.S. market, U.S. companies selectively working with manufacturers that do things in a familiar way, general wide-adoption of the U.S. ways, or something else?
If I were to be in Russia or a former Soviet state, would I see consumer electronics looking more like this Soyuz clock?
Edit to add: there were various ribbon cables used in eastern block electronics, but mostly with hand soldered connectors, not IDC. Two exceptions I can think of are Shugart-compatible floppy drives (although some Czech computers actually use ribbon cables with hand soldered board-to-card connectors) and some Metra-branded measurement equipment which uses MicroD-like IDCs internally.
Precision or otherwise higher grade (e.g. ceramic wirewounds) components were always usually marked with text.
https://www.cia.gov/library/readingroom/docs/CIA-RDP80-00809...
The document is from 1950 though, so it greatly predates the standards used the 1983 Soyuz here.
Edit: If anyone can find labeled resistors for a reasonable price, I would be grateful.
My lab productivity soared.
The one important thing is to not pinch on the pincers - buy a good pair... :)
Additionally, regional distributor ELFA had bins of 1000 1% 0805 resistors for ~$10/ea, so I just purchased a couple of bins at a time as I needed them until I had an E12-ish set.
If there is one nearby, I'd strongly suggest you visit a maker space - chances are they have SMT kits already.
However, I still use wired parts for breadboarding.
Googling, I see one for Android, which I don't use:
https://play.google.com/store/apps/details?id=com.mhdev.resi...
or
https://play.google.com/store/apps/details?id=com.mhdev.resi...
there is an app for identifying resistor color combinations, could that help? (didn't use that myself)
A typical US IC part is 7490. 74 indicates the TTL family. But the 90 is arbitrary. The 7490 is a counter but 7491 is a shift register. The part number doesn't give you any idea what the IC does.
See Wikipedia for details on the Soviet IC system: https://en.wikipedia.org/wiki/Soviet_integrated_circuit_desi...
I dislike color bands for two reasons:
1) I can never remember which color is which digit (it follows the rainbow, but not quite).
2) The colors can be hard to identify due to varying color shades, and quality ("is that read or brown?"). It also doesn't work well for people who are color blind.
I believe this is to maintain a single point ground, usually the chassis on a spacecraft. The chassis of the clock is probably connected to the isolated side's ground, which avoids ground loops when the clock is integrated into the S/C.
More detail on grounding and isolation in S/C is covered here: https://standards.nasa.gov/standard/nasa/nasa-hdbk-4001
What a really cool output device.
BTW, do you have any measurements of that clock you could share? I'm seriously considering mocking one up just for the cool factor... It'd be nice to get it at least approximately the right size. (I'm also considering building a working model Globus indicator too. Maybe I'll just size everything based on the most suitable globe I can find...)
The US GPS constellation wasn't "complete" till '95, and GLONASS till '96.
The '96-02 vintage of this digital clock means it's _possible_ the Soyez were listening GPS satellites - but given it's '64 heritage I doubt it...
You're talking about NAVSTAR GPS, which is what most people refer to as simply "GPS". However other radio based US military nav sats date back to the 50s with the Navy's TRANSIT system, and Timation 1 was launched in '67, which was probably the first sat with a signal helpful for real-time, fast moving calculations, but was almost certainly too big to fit in a space ship at the time. Usable LORAN dates back to ~1940, if you don't limit it to sat based systems.
I'm not familiar with what the Soviets were capable of at the time in terms of either their independent nav systems, or ability to take advantage of the US ones, but it's reasonable to think they were putting as much military effort into it as the US was. Hell, a lot of info on the US history has only been ~recently declassified.
Point being, the space race happened specifically to develop this type of nuke delivery technology. At the time, this thinking definitely included manned space capsules (think nuclear bombers, but flying in space), as well as ICBMs, both of which would require precise navigation.
That said, the IMP merely calculated and displayed updated position based on manually input orbit parameters, which were calculated on the ground and radioed up.
Those, in turn, were based on location/velocity observations using both radio and radar range-finding, but all ground-based.
They have color coded Russian resistors, probably just for smaller sizes. Diodes have a variety of shapes, but I don't remember seeing rectangular packages like that one either. Some of those component might have also been specially sourced high tolerance components that might be different than what you'd find in consumer electronics?
> The logos on the integrated circuits reveal that they were manufactured by a variety of companies.
Oh very interesting. I have been wondering about what those logos meant when I used to play with electronics back in the day. They even have a link to the full list http://madelectronics.ru/book/prominfo/2009-04-16-08-29-39-3....
Why does the Soyuz clock contain over 100 chips instead of being implemented with a single clock chip? Soviet integrated circuit technology was about 8 years behind American technology, so TTL chips were a reasonable choice at the time.
Because in the paragraph above:
I expected the Shuttle computer to use 1980s microprocessors and be a generation ahead of the Soyuz clock, but instead the two systems both use TTL technology, and in many cases almost identical chips.
This. The author points out the 54F00, seemingly distracted by the 54'00 part and similarities in TTL glue logic layout that the F part is completely dismissed without acknowledging that these chips had sub-4ns edge rates and were indeed fast while remaining compatible with older TTL families. Throw in high SMD density on a multi-layer controlled-impedance PCB designed to survive brutal operating environments when PCs and CAD were still in their infancy...even today, it's humbling to contemplate just how much work would have been required to qualify such a design.
> But it's interesting that the Shuttle was still using TTL...
These systems had super long lifecycles and were required to be extremely reliable. I'd be surprised if the contractor that was responsible for the design would have been allowed to integrate any IC that wasn't listed in a QML.
So I'm guessing the Shenzhou program doesn't count?
https://en.wikipedia.org/wiki/Shenzhou_(spacecraft)
(Yes, I know the OP goes on to say "and used for flights to the ISS". Still irritatingly inaccurate.)
Wikipedia says "Soyuz has served as the only means for crewed space flights in the world since the retirement of the US Space Shuttle in 2011", so I guess that's wrong?
Edit: And does it still work?
You could say it makes the original statement unintentionally true.
Source: http://www.astronautix.com/s/shenzhou.html
As I understand it, China is excluded from treaties governing cooperation in crewed spaceflight that go back to the Apollo-Soyuz link-up and include other third parties such as ESA and Japan: this locked them out of the ISS program. So they had to develop their own space station, too.
The clock was probably developed much earlier, and then perhaps slightly modernized with LEDs and such. Developing a new clock for spacecraft would probably take a year or two, for all the testing and certifications, so nobody bothered.
Same functionality more or less, probably a fraction of the weight and cost.
I'd attribute the difference to how military/government projects pan out and not necessarily to the less advanced soviet IC abilities. There are many similar examples in western military equipment.
It never worked right :-)
In more detail, first there were bipolar transistors (NPN and PNP) and later MOS transistor (NMOS and PMOS). Bipolar transistors are your basic semiconductor transistors with three layers of semiconductor. MOS (metal oxide semiconductor) transistors have an insulating oxide layer between the silicon and the metal (or polysilicon) on top. MOS transistors were developed later than bipolar, and started to become popular in the 1970s.
One type of logic that you can make from bipolar transistors is TTL (transistor-transistor logic). TTL is better than earlier logic families such as resistor-transistor logic (RTL) or diode-transistor logic (DTL). TTL was cheap, reliable, fast, easy to use, and popular with minicomputer manufacturers. TTL has two main problems, though. It uses a fair bit of power, and you can't make it very dense. I.e. you can't put a lot on a chip.
MOS, on the other hand, has the advantage that you can make very dense circuits from it. (I.e. Moore's law applies.) NMOS was used for early microprocessors such as the Z-80 and 6502. The problem with NMOS is that it uses resistors (sort of) in the logic gates, and these waste power.
The solution was CMOS, complementary MOS. You use both NMOS and PMOS transistors (the complementary part), and you can get rid of the resistors, and your chip uses very low power. The problem with CMOS is it's more complicated because you need two types of transistors, and twice as many (sort of). However, the need for low power won out in the mid-1980s and microprocessors such as the 80386 started using CMOS. Use of CMOS has continued to the present.
I'm oversimplifying the history somewhat. The books "To the digital age" and "History of semiconductor engineering" go into much more detail.
Now I'm dating myself, but there was a time period when CMOS gates were slower than TTL or even faster logic families like RTL. So a fast machine like the Cray was built with RTL, if I recall correctly. But eventually CMOS won this race.
RTL (resistor-transistor logic) is entirely different. It was used in the Apollo Guidance Computer because that's all integrated circuits gave you at the time, but it's not very good. (The inputs affect each other because they are only separated by resistors. Even diode-transistor logic is much better.)
CMOS is more like the electronic equivalent of a stiff mechanical lever system. Logic voltages change the charge distribution inside the MOS transistors. There's almost no current flow outside of them.
This is great for super-low power draw, but it's not particularly snappy, it can't drive LEDs directly, and it's not happy driving long wires without buffering. It's also very easily damaged by static electricity - which is why wrist grounding straps are a thing. Because the transistors are so tiny it's easy to scale it to VLSI.
TTL is mid-way between the two. The transistors saturate cleanly, so switching is faster than CMOS. The source/sink currents are much smaller than with ECL, so power requirements are manageable, although switching glitches need good decoupling. It's a robust and reliable system, with low-power options. But it's not as dense as CMOS, which is why 74xx series TTL never got any more complex than a simple 4-bit ALU, while MOS soon expanded into full 8-bit VLSI microprocessors.
Fast enough for practically all processors made today
> it can't drive LEDs directly
That depends on the design, but CMOS chips generally have much better symmetric output characteristics than TTL ever had. Most MCUs can drive quite some current on the output pins, and 74HC and similar logic families can certainly drive LEDs.
This kind of stuff is fascinating.
Smart! The color band system has +ve and -ve, but overall I hate it. But better than caps with their 104 = 100,000 pf/nF?
I've known the US color band system since I was a kid, but damn the base color of the resistor epoxy has a huge impact, as does the low wattage high precision: you need a magnifier to read all 5 bands on a 4mm resistor.