Soviet production of electronic components

[janoc]

The only annoying thing about the Russian ICs was that the pin pitch was (intentionally) 2mm instead of 2.54mm (0.1inch), so it was kind of funny to insert a 2x20 DIL Russian chip into a socket or a PCB made for 0.1inch pin spacing.   

That's interesting, which chips are those? I have some salvaged Soviet chips (74xx series logic) from some sort of computer and they are all in standard 0.1" pitch DIPs.

I believe chips made for export were on 0.1" or 0.05" pitch and ICs meant for internal Russian/Soviet market used the metric pitch (so the same IC could have existed in both, the only difference was a letter prefix in the marking). Which could explain why I haven't seen those metric ones.

I'm not aware of any imperial variant for export. Probably you simply did not notice the difference since it's small for low pin count ICs. It only becomes an interchangeability problem for 28+ pin ICs.

That's not true according to this:
https://en.wikipedia.org/wiki/Soviet_integrated_circuit_designation

The E prefixed chips were the export variants.

I wish I could find the old Soviet chips - they are somewhere in my junk bin but can't find them right now I have found only the Czech Tesla ICs and those are clearly on standard 0.1" pitch.

[free_electron]

In the photo I noticed that one lamp broke

they survive nuclear EMP , but they will fail when sitting idle on a bench ...

[wraper]

That's not true according to this:
https://en.wikipedia.org/wiki/Soviet_integrated_circuit_designation

The E prefixed chips were the export variants.

It's Э, not E. Though I'm not convinced they existed before Soviet Union collapsed. I've seen those only made by Integral (Belarus factory). On pictures I can find, the oldest ICs were made in mid 90's

[vk6zgo]

It seems that the clunky old "Postmaster General's Dept" was more forward looking than the Air Force!

More likely than not that "Postmaster General's Dept"'s gear didn't have to survive many Gs, handle large temperature swings and be reliable in the field (or at least easily repairable).

Solid state stuff in the 60s was nowhere near this level (that's where the "milspec" parts come from). Also, the military procurement has some particularities, like certification and the ability to obtain replacements even 20-30 years down the road. Tubes had all this established, those newfangled semiconductor things not yet.

When it comes to mission-critical, humans-will-die-if-it-fails stuff, conservative and proven will always beat the new shiny.

No high "Gs", but a lot of it  did have to sit out in the bush, unattended for years, & stay in operation. 24/7/365 for, in some cases, 30+ years, & certainly, was easily repairable, if you possessed the skills.

In most cases, the equipment was in airconditioned huts, but not always, & temperature extremes were quite severe in those cases.

The PMG certainly used truckloads of tubes over many, many, years, but the solid state stuff  of the 1960s was reliable.
Certainly, mid to late 1950s was a different story with so many germanium devices, but silicon was a quantum leap in predictability & reliability.

[vk6zgo]

Yeah, the idea that transistors weren't available until the late 60's or early seventies is flat wrong.  By the early 60's the leaders in the semiconductor industry were beginning to integrate several transistors into modules -- IBM had what the called MST back then if my memory serves me.  By the late 60's early 70's we had the first large scale integrated circuits like microprocessors. 

Please, do read again what I have posted. Then you will see that you have completely missed the point.

The MIG25 was developed in the 50s, first flew in 1964!  Please, don't tell me than in 1960s there were power transistors available on the market capable of handling 20-40W of power, in wide temperature ranges, as required for this application. The issue is not whether transistors as such were available but whether trapnsistors suitable for the task were. Which is not quite the same thing.

But that is not what you posted--- you baldly stated the following, with no qualifications as to applications:-

Transistor has been invented only in 1947 and reliable mass-produced transistors did not arrive until late 60s/early 70s - and that was in the West,

We can only answer what you posted, not what you intended.

Also, we are talking about Soviet Union, so maybe 10 years R&D lag in electronics, perhaps more, with a country still recovering from the war, with Stalin in power and planned economy. What IBM was doing in the late 60s, early 70s is thus totally irrelevant.

[coppice]

Transistor has been invented only in 1947 and reliable mass-produced transistors did not arrive until late 60s/early 70s - and that was in the West, not in Soviet Union, hampered by embargoes!

In the west very cheap high volume transistor radios were available at the beginning of the 60s. By the mid 60s most stereo systems were 100% semiconductor, mostly using discrete transistors. By the late 60s transistors and ICs had reached the stage where the only vacuum tube left in most TVs was the CRT. Your timing is way off.

[Cerebus]

By the late 60s transistors and ICs had reached the stage where the only vacuum tube left in most TVs was the CRT. Your timing is way off.

I'd push the transition from mostly vacuum tube to mostly solid state for new TV sets to a bit later than that. I knew three of the local TV sales/repair shops well so used to see the inside of more TVs than most youngsters (one was closest, where I used to go to buy components, one was run by one my father's only army mates and one was run by the father of a close friend at school). There were hybrid vacuum tube TVs being still sold at the start of the 70s (solid state tuner, vacuum tube HV section) and I'd put it closer to the mid 70s before all new sets reached the point where one could say with any confidence that the only vacuum tube in a new set was the CRT.

[coppice]

By the late 60s transistors and ICs had reached the stage where the only vacuum tube left in most TVs was the CRT. Your timing is way off.

I'd push the transition from mostly vacuum tube to mostly solid state for new TV sets to a bit later than that. I knew three of the local TV sales/repair shops well so used to see the inside of more TVs than most youngsters (one was closest, where I used to go to buy components, one was run by one my father's only army mates and one was run by the father of a close friend at school). There were hybrid vacuum tube TVs being still sold at the start of the 70s (solid state tuner, vacuum tube HV section) and I'd put it closer to the mid 70s before all new sets reached the point where one could say with any confidence that the only vacuum tube in a new set was the CRT.

The two major TV suppliers in the UK in the 60s were Philips and Thorn. The Philips G8 and Thorn 2000 were all semiconductor (except the tube), and both were shipping in volume in 1969 (although some historical sites say 1970 for the G8, but you had to be there, man  ). By about 1973 B&O were famous for the being the last maker with valves/vacuum tubes in their TVs.

The TV models before the G8 and 2000 had a high semiconductor content. These were just the last steps in eliminating the valves. For comparison, the TV my parents bought in 1963 had a few semiconductor diodes, but no transistors at all.

[janoc]

That's not true according to this:
https://en.wikipedia.org/wiki/Soviet_integrated_circuit_designation

The E prefixed chips were the export variants.

It's Э, not E. Though I'm not convinced they existed before Soviet Union collapsed. I've seen those only made by Integral (Belarus factory). On pictures I can find, the oldest ICs were made in mid 90's

  Both Э and E are transliterated as E in latin ... And yes, they did exist, that marking system was used between 1973-1980 and the Э and K prefixes were likely abandoned later.

[janoc]

By the late 60s transistors and ICs had reached the stage where the only vacuum tube left in most TVs was the CRT. Your timing is way off.

I'd push the transition from mostly vacuum tube to mostly solid state for new TV sets to a bit later than that. I knew three of the local TV sales/repair shops well so used to see the inside of more TVs than most youngsters (one was closest, where I used to go to buy components, one was run by one my father's only army mates and one was run by the father of a close friend at school). There were hybrid vacuum tube TVs being still sold at the start of the 70s (solid state tuner, vacuum tube HV section) and I'd put it closer to the mid 70s before all new sets reached the point where one could say with any confidence that the only vacuum tube in a new set was the CRT.

Yeah, Coppice has no idea that in the Eastern bloc that we are talking about, B&W TVs full of tubes were a common sight in households way into the 80s ... I remember taking apart at least two or three around 1988 or so.

[janoc]

But that is not what you posted--- you baldly stated the following, with no qualifications as to applications:-

Transistor has been invented only in 1947 and reliable mass-produced transistors did not arrive until late 60s/early 70s - and that was in the West,

We can only answer what you posted, not what you intended.

Please, at least don't selectively quote one sentence out of context. This is what I wrote, in full:

Realize that MIG25 was designed in the late 50s, with the first flight in 1964. So the reason the tubes were used in the MIG25 is simply because nothing else was available. Transistor has been invented only in 1947 and reliable mass-produced transistors did not arrive until late 60s/early 70s - and that was in the West, not in Soviet Union, hampered by embargoes! Also, the largest piece of electronics in MIG25 was the very powerful radar - - there you simply didn't have a choice because high frequency/high power transistors suitable for such uses didn't arrive until the plane has been retired (if ever!).

I wonder how did you manage to miss all of that in order to pick out one single sentence out of the middle of that paragraph to reply to? Does that text give an impression I was talking about pocket radios or something like that, with the MIG25 plane being mentioned no less than 3 times?

[coppice]

By the late 60s transistors and ICs had reached the stage where the only vacuum tube left in most TVs was the CRT. Your timing is way off.

I'd push the transition from mostly vacuum tube to mostly solid state for new TV sets to a bit later than that. I knew three of the local TV sales/repair shops well so used to see the inside of more TVs than most youngsters (one was closest, where I used to go to buy components, one was run by one my father's only army mates and one was run by the father of a close friend at school). There were hybrid vacuum tube TVs being still sold at the start of the 70s (solid state tuner, vacuum tube HV section) and I'd put it closer to the mid 70s before all new sets reached the point where one could say with any confidence that the only vacuum tube in a new set was the CRT.

Yeah, Coppice has no idea that in the Eastern bloc that we are talking about, B&W TVs full of tubes were a common sight in households way into the 80s ... I remember taking apart at least two or three around 1988 or so.

Can you please try to make sense. I know perfectly well what Eastern bloc equipment was like. I replied to a comment from you about the adoption of semiconductors in the west.

[wraper]

Both Э and E are transliterated as E in latin ... And yes, they did exist, that marking system was used between 1973-1980 and the Э and K prefixes were likely abandoned later.

Russian alphabet has it's own E letter. Neither Э or K was ever abandoned, you can buy ICs with such marking with 2020 date code. And I never seen Э prefix with date code earlier than 1994, and made by any factory other than Integral in Belarus. K letter is present on almost every Soviet IC which is not military grade (even on ICs with different numbering scheme made earlier than mid 70's).

[free_electron]

AHA ! Now i found who makes the 'Viltage' converter !  https://integral.by/en/products/power-supply-ics

[vk6zgo]

But that is not what you posted--- you baldly stated the following, with no qualifications as to applications:-

Transistor has been invented only in 1947 and reliable mass-produced transistors did not arrive until late 60s/early 70s - and that was in the West,

We can only answer what you posted, not what you intended.

Please, at least don't selectively quote one sentence out of context. This is what I wrote, in full:

Realize that MIG25 was designed in the late 50s, with the first flight in 1964. So the reason the tubes were used in the MIG25 is simply because nothing else was available. Transistor has been invented only in 1947 and reliable mass-produced transistors did not arrive until late 60s/early 70s - and that was in the West, not in Soviet Union, hampered by embargoes! Also, the largest piece of electronics in MIG25 was the very powerful radar - - there you simply didn't have a choice because high frequency/high power transistors suitable for such uses didn't arrive until the plane has been retired (if ever!).

I wonder how did you manage to miss all of that in order to pick out one single sentence out of the middle of that paragraph to reply to? Does that text give an impression I was talking about pocket radios or something like that, ewith the MIG25 plane being mentioned no less than 3 times?

Sorry, it doesn't work that way!

You evidently included "and that was in the west" to add emphasis to what you were saying.
If you had just said "in the Soviet Union", nobody would have commented further on what was happening in those years outside the then "Eastern Bloc".

Because you used those words, it is quite fair for others to refute them.

On your most recent comment:-
HF, VHF & UHF Communications transmitters & receivers are hardly "pocket radios".

[jmelson]

Yeah, the idea that transistors weren't available until the late 60's or early seventies is flat wrong.  By the early 60's the leaders in the semiconductor industry were beginning to integrate several transistors into modules -- IBM had what the called MST back then if my memory serves me.  By the late 60's early 70's we had the first large scale integrated circuits like microprocessors. 

IBM started producing transistorized computers in about 1958-59.  The 7094 was produced in about 1962-64, and had 55,000 transistors on 11,000 small circuit boards.  This pretty much boggles my mind.

Then, for the System/360, they produced SLT (Solid Logic Technology) which were 1/2" square ceramic hybrid modules with generally 2 transistors and 4 diodes, plus 4 thin-film resistors.  They made many millions of these SLT modules.  The System/360 was introduced in summer 1965, and started shipping late that year.

MST was Monolithic Solid Technology, and was used in the 360/85 in 1968 and the model 195 in 1969, before coming out
in all the 370 models in 1970.  It was VERY similar to Motorola MECL.

Jon

[raptor1956]

Yeah, the idea that transistors weren't available until the late 60's or early seventies is flat wrong.  By the early 60's the leaders in the semiconductor industry were beginning to integrate several transistors into modules -- IBM had what the called MST back then if my memory serves me.  By the late 60's early 70's we had the first large scale integrated circuits like microprocessors. 

IBM started producing transistorized computers in about 1958-59.  The 7094 was produced in about 1962-64, and had 55,000 transistors on 11,000 small circuit boards.  This pretty much boggles my mind.

Then, for the System/360, they produced SLT (Solid Logic Technology) which were 1/2" square ceramic hybrid modules with generally 2 transistors and 4 diodes, plus 4 thin-film resistors.  They made many millions of these SLT modules.  The System/360 was introduced in summer 1965, and started shipping late that year.

MST was Monolithic Solid Technology, and was used in the 360/85 in 1968 and the model 195 in 1969, before coming out
in all the 370 models in 1970.  It was VERY similar to Motorola MECL.

Jon

When I worked at IBM East Fishkill, 1982-1993, they may still have been making some of the devices as they had quite a few old and small manufacturing lines in operation.  When I started they were still making chips using 57mm, 82mm, and 100mm wafers.  For a long time IBM went there own way and when the industry was at 50mm IBM squeezed 57mm out of it -- similarly when the industry was at 75mm IBM was doing 82mm.  By the time 100mm rolled around IBM realized they were paying too much for the marginal gains.

When I started in 1982 the feature sizes were still above 1um so they could still tolerate the kind of particulate issues that a decade later would have destroyed 100% of all chips so we began to investigate methods to reduce the particulate levels in our 4-stack horizontal Tempress furnaces used for oxidation and diffusion processes.  At the time the wafers were put into quartz boats about 25-50 per boat and then those boats were put onto a quartz sled and the sled, with about 200 wafers, was then pushed into the hot zone of the furnace with a quartz push rod.  We used a half-step stepper motor system to push/pull the sled at about 5 inches per minute to avoid warpage.  One of the engineers calibrated the control in units of 'furlongs per fortnight'.  The solution to our problem with particulates was to eliminate the quartz-on-quartz sliding of the sled in the quartz process tube by using a cantilever paddle.  Not long after we began working on this project the mechanical engineer that was designing the test fixtures was accepted into the IBM advanced degree program (forgot the name of it) and he left -- I was put in charge of completing the design and testing and in the end the solution we came up with was the same that just about everyone else in the industry came up with -- the Norton Crystar SiC paddle.  By the late 80's or early 90's the first of the vertical furnaces came out with built in wafer handling automation and the days of the horizontal furnace were numbered.  I left IBM about that time and joined SVG (Silicon Valley Group) which made, among other things, vertical furnaces.  Before SVG, which by then had been bought by ASML, went out of business, they were producing a new class of tools called ALD for atomic layer deposition -- basically you puff small quantities of reactants into an evacuated process tube and can produce films a single atomic layer thick.


Brian

[Canis Dirus Leidy]

Necromancy time again. In the spring (April issue) of 1981 in the journal "Science and Life" there was an article about microcalculators, to which was attached a photo essay on the production of LSICs. Scans of pages with this essay, plus my short translation of photo captions:

1. Silicon ingot before cutting into wafers.
2. Wafers are ground and polished on such machines.
3. Photolithography: exposure of photoresist on wafer surface.
4. Wafers (in quartz tubes) are loaded into a diffusion furnace.
5-6. Loading wafers into a vacuum chamber for metal spraying.
7. Cutting the wafer into individual ICs.
8. Checking and rejection of ICs.
9. Control probes close-up.
10. Welding the terminals to the crystal.
11. One of the control computers.

[GlennSprigg]

Sticking to the 'Topic' description... (people that know me know how important that is to me! hahaha    )...
As an 'old' bastard, (or old 'bastard'!) and into old valve/tube equipment, I've always loved certain old Russian equipment/components!!
Yes, they were often built cruder, but always (seemingly to me), more robust! Meaning that they lasted, especially with physical abuse.
In regard to todays modern electronic components, I've always wondered about the following...

Inside I.C's, (even to a point with SMDS), they make the ACTUAL chip say typically about 2mm square. Obviously, this has to be EXTENDED
internally to wire it to tiny little wires to the normal size (and solder-able/accessible) typical IC casing/pins. I understand that either today, or
in the future, there are/maybe numerous quantities of individual IC's all built into one, but why so small?  Why would the cost of a silicon wafer
be so high with it's associated component design/integration, when in the end it needs to expand to physically usable dimensions? 

[helius]

They are not "extended". They are wire bonded to a lead frame which contains the external leads, before encapsulation. In more modern technology the die is soldered face-down onto an interposer using small solder bumps.
The reason a die must be as small as possible is cost. Fabricating each wafer is a fixed cost, no matter how big or small the chips are. Smaller dice means more dice per wafer, and higher yield of working devices, since the randomly located defects on the wafer affect a smaller fraction of the total devices. If chip A is half the area of chip B, it will be about a quarter the cost. This is also the reason that the most cost-competitive CPUs use multiple "chiplets" per interposer in MCM (multi-chip module) arrays. The economics is vastly better than large monolithic chips with the same functions.

[wraper]

If chip A is half the area of chip B, it will be about a quarter the cost.

Not really. If both dies are small enough (for the process maturity) they will cost almost the same for the area used because the percentage of defective dies will be small in both cases. Also it depends on process maturity and if defective parts of the chip can be disabled (cores, cache, etc). Many chips are already produced with spare functional blocks. For example all PS3 SOC had 8 SPE but only 7 were used for better yield. All modern DRAM has spare rows and columns. 2x size for 4x the cost will only be true if silicon die is very large, defective parts cannot be disabled and manufacturing process is not mature enough (low yield). Nor this makes any sense mathematically. For example you make 2 chips on the same process. One chip has 16 times larger die area (2^4) that other. Then larger chip should cost 256 times (4^4) more according to what you said. Which would mean 16 times higher cost for the same die area. It could only be true only if for the larger chip the yield was only a few %.

This is also the reason that the most cost-competitive CPUs use multiple "chiplets" per interposer in MCM (multi-chip module) arrays

Also not entirely accurate. When chiplets are used, the modern, immature and more expensive process can be used only for what it's actually needed and the rest of the chip can be made on an older and cheaper process. For example Ryzen CPUs use 7nm process for smaller CPU core chiplets (often there is only one such chiplet) and 12nm or 14nm for larger IO die. Not only they save the money because of better yields, they don't need to produce different masks (extremely expensive) for different CPU models. But when it comes to laptop CPUs, the same AMD does not use chiplets and makes a much larger die on a latest process because it lowers power consumption compared to chiplet design.

[Canis Dirus Leidy]

"Technology for the production of special microassemblies[1]"

1. Hybrid ICs with thick-film passive components on ceramic substrate.