Russian microelectronics for space: who produces what
In connection with the well-known events in the news there were reports that the US banned the supply of microelectronics for Russian satellites and military equipment.
Such a development of events may negatively affect the state of the Russian aerospace and defense industries, because the annual import of electronics for the space industry is two billion dollars, and these are chips that are critical for the operation of satellites. Some officials (see the article on the link) have already begun to panic and talk about buying electronics in China, who allegedly started production of everything necessary. I want to talk a little bit about what chips are being developed and manufactured for the space industry in Russia. I’ll say right away that this review in no way claims to be complete and will concern microchips (and silicon), whereas the needs of space are not limited to them - we also need passive components, microwave devices, power discrete elements and much more. partially imported and partially developed and produced in Russia. To describe this all is a completely impossible task, and not very necessary, because the purpose of this review is not to describe everything that is, but to show that everything is not as bad as it seems to alarmists. The specifics of the industry are such that almost all of the really important information is closed, but something interesting could still be dug up.
A small digression, a reminder: chips for working in space should be resistant to radiation. About why this is so and how exactly the radiation affects electronic devices, you can read here and here .
The main problem with the use of imported components in the aerospace and military industries is that these components are usually not designed to work in such conditions (that is, they are ordinary consumer goods originally intended for irons and refrigerators). This situation arose in the nineties, when there was simply nothing else, and what happened was worth more than the space equipment developers could afford. That is why the terms of active functioning of Russian satellites are still very seriously behind American or European ones. For example, the beautiful American on-board computer RAD750 (standing on the Curiosity rover) was never sold to Russia, and its counterparts in Russia did not exist until recently. So the problem with imports did not arise yesterday, and they began to be solved for a long time. In 2007, the federal target program “Development of Electronic Component Base and Radio Electronics” for 2008-2011 was adopted. Further, the final year of the program was 2015, and soon we should see its results, including the replacement of imported components in the space and defense industry with Russian developments.
')
There are lists of microcircuits to be replaced, but they, unfortunately, are classified. In open sources, you can find information (a link to download pdf) that the number of positions in these lists is hundreds or even thousands (and these are only chips, not to mention passive elements such as high-precision resistors, which are even greater). The situation is complicated by the fact that, in the overwhelming majority of cases, hardware developers want them to do “exactly the same, but Russian” (that is, it is copying foreign developments down to pin-compatibility), while the list can certainly be shortened several times due to development of a set of systems on a chip, uniform for all developers and customizable to the needs of specific users. This also includes the unification of data transfer interfaces (which is not available now) and other fairly obvious measures. Work in this direction, as far as I know, is underway, but you understand that it can turn out to be even more expensive to develop onboard equipment from scratch than to develop more chips than the required minimum.
Actually, a review of developers and manufacturers of microcircuits for the space industry (and partly military, since they are quite strongly interconnected) should start with production, because there are significantly fewer of them in Russia than developers, and it is precisely in this area that the lagging behind is abroad. I will not dwell in detail on the factories with design norms of more than half a micron, because they are hopelessly outdated, and some new developments are being carried out on them except of despair. Nevertheless, a lot of things are being done on them, first of all - microcircuits of a small degree of integration, power electronics, microwave, etc. But there are only four factories with more or less modern design standards.
Threat Here , by the way, you can read the distracted gaze of two years ago.
Zelenograd plant "Micron" (here is a post with beautiful pictures) declares the presence of technologies on its website :
1) 180 nm volumetric technology with EEPROM is already a well-known technology for developers, which employs almost everyone who has this capability. On it, for example, chips for tickets of the Moscow metro are made. The presence of an EEPROM is useful for applications such as embedded microcontrollers, which are necessary not only for space, but also for “civilian” applications. Design whales are available to developers.
2) 250 nm silicon on an insulator (SOI) - according to this technology, separate developments are already underway, but I have not heard about finished products yet. The technology, apparently, is a partial port of 180 nm on silicon wafers on an insulator. Design whales are available to developers.
3) 180 nm CED - declared 2012 by the year; in practice, nothing new has been heard about it for quite some time. That is, it is in development, but when it will be ready for design is not very clear. In any case, I have not found such information anywhere.
4) 90 nm volume. Most recently, Micron licensed software for physical verification for this technology from Cadence. No mention of specific developments on it has yet been found; only general words that they are underway.
5) Quite recently, 65nm technology appeared in the news, but everything is still at the level of test crystals. Mass production promise this year .
An important thing worth noting is a quote from the Micron site: “More than 50 companies from 12 countries of the world delivered equipment and materials suppliers, infrastructure partners.” This is something that is often overlooked, speaking about overcoming the import ban on domestic production - a huge amount of spare parts and consumables are also imported, and a ban on their import can be more effective than a ban on the import of ready-made microchips. I hope that this will not happen in the near future, and that there will be money and resources to organize the production of consumables in Russia.
Angstrom (and Angstrom-T), Zelenograd
1) 1.2 micron silicon on sapphire (CNS) - technology for thousands of years, but it is still in demand (apparently, conservative military customers for chips with high radiation resistance, for which reliability and validation of solutions is more important than their timeliness).
2) 0.6 µm, bulk silicon, silicon on sapphire, silicon on insulator, the possibility of manufacturing EEPROM, BiCMOS, IGBT. Apparently, a good process for power electronics.
3) "Created" technologies with design standards of 350-250 nm.
4) “Created” technologies with design standards 130, 110 and 90 nm.
The history of the "creation" of technology at Angstrom is long, difficult and not yet resolved. The process and equipment 130 nm were purchased from AMD, 90 nm - from IBM. Every year for five years they say that everything is about to be, but so far nothing.
Regarding the radiation resistance on the Angstrem website, by the way, the best nonsense was written about the fact that their silicon on sapphire is the only suitable technical process in Russia and that practice has shown the impossibility of making radiation-resistant circuits on technologies less than 250 nm. Let's see what happens when they launch 130 and 90 nm :-) Physics are unlikely to learn, but the presence of production practices can have a positive effect on the understanding that in states, space chips are already being designed at 45 nm, and in Russia - at 65 (true foreign) .
Research Institute of System Studies of the Russian Academy of Sciences, Moscow
The factory of NIISI RAS is located on the territory of the Kurchatov Institute in Moscow and possesses technologies with design standards of 500, 350 and 250 nanometers on bulk silicon and SOI plates. Initially, it is not intended for large-scale production and is positioned as a “research factory of the Academy of Sciences”. Most of the microcircuits produced here are developed by NIISI itself, but the factory also works with external customers, for example, with Voronezh NIIET, which produces its radiation-resistant microcontrollers here.
There are no other details, and the latest open publications about the factory are almost the time of its opening.
Integral, Minsk
The Minsk Integral is considered to be our military and other institutions by a domestic enterprise with all the attendant circumstances. Interesting, isn't it?
The main technologies of Integral are old, with design standards of 0.8 microns and more, but in the last few years, Belarusians have independently designed and launched technologies of 0.5 and 0.35 microns on bulk silicon and SOI. They have only three metals (which is not enough for a microprocessor), but Integral develops SRAM microcircuits with a capacity of 1 Mbit and high radiation resistance, as well as analog microcircuits.
Reports from scientific conferences also feature 0.18 micron technology and 0.5-0.25 micron BiCMOS, flash memory (the only one in the CIS?) And FRAM.
The overwhelming majority (if not all) of the developments on their technologies, Integral conducts independently.
That's all. Looks a little sad, is not it? Yes, radiation-resistant electronics for space do not always require the same design standards as usual, and the lag in several generations is not very scary (the Curiousity rover at 250 nm works fine), but still the new fast-resistant BAE systems are developed using the 45 nm technology, and so far we have not reached the industrial development of 90 nm. On the other hand, five years ago there was practically nothing of this, and today Russian plants have quite real opportunities to provide astronautics with microelectronics of their own production.
Now about the developers. There are many more of them than manufacturers, but until recently it was normal practice to manufacture microchips developed in Russia somewhere abroad, for example, in Taiwan (TSMC), in Germany (XFAB) or in Israel (Tower). At a time when there was nothing better than 0.8 microns in Russia, even the military closed their eyes to it, believing that “developed in Russia” = “Russian”. Now times have changed a little, and Russian developers are increasingly producing their products on Micron (that is, there are more than metro chips doing there).
It is also necessary to note the fact that the overwhelming majority of Russian microelectronics developers are somehow tied to budget money and large orders, especially space or military, and especially commercial orders are few. On the other hand, a significant share of the profits of several enterprises (for example, VZPP-Micron and Angstrem) is export. However, I digress from the main topic of the review, so I’ll confine myself to the observation that the products presented below are not all that are produced, and for many companies space is not the main topic.
Milander, Zelenograd
PKK Milandr, CJSC, based in Zelenograd, is a company with a twenty-year history and, more importantly for us, with the most detailed website among all Russian microelectronic companies. It managed to find this:
1645RU2T - static RAM (SRAM) with a capacity of 64 Kbps. In serial production since 2008.
1645RU5U - static RAM (SRAM) with a capacity of 4 Mbit. OCD ends in 2014.
Judging by the year of production, the first microcircuit is produced on some very old technology, the second - 180 nm (probably on the "Micron").
Under the link (caution, traffic) you can find photos of the radiation-resistant 8-bit microcontroller 1886BE10 (analogue of PIC17), for which reason for some reason there is no information about it.
The technology is Micron 180 nm, which is radically resistant and contains a complete stuffing made from ring transistors and multi-transistor storage elements. There is no accurate data in the public domain, but a chip with such methods of protection must withstand a nuclear explosion, not like a long-term flight in space.
16452 - single programmable ROM (antifuse) with a capacity of 256 kbps. ROC was commissioned in 2013
Here you can see how it looks. The design standards, judging by the opened crystal, are 680 nm.
5576RT1U - single programmable ROM (antifuse) with a capacity of 1 Mbit. The design and development works were commissioned in 2013. The design norms are most likely 180 nm (the Micron technology).
For other Milandra microcircuits, radiation resistance is not stated, however, for example, in the news you can find the following line on the site: “The parameters of resistance to special factors for 1310PN1U microcircuits were updated (significantly improved)”. 1310 is an inductive power converter for which radiation resistance is not claimed. If all the microcircuits supplied with the fifth acceptance have at least some resistance to radiation, then Milandr still has a fairly wide range of interface microcircuits, power management, and ADC / DAC.
The promising development of Milandra is their first radiation-resistant and fail-safe microprocessor. It does not yet have its own designation and is presented at various conferences under the name " Processing-13 ." (link to download pdf) By reference - a presentation about the processor device and its design in terms of ensuring radiation resistance. There are interesting and controversial decisions, but it looks impressive (except for the joint work of the nuclei, perhaps).
The processor is a dual-core ARM Cortex-M4F with separate core operation modes and hardware duplication. Clock frequency - 100 MHz, SRAM 32 KB, 128 KB ROM, a wide range of interfaces and analog peripherals.
Processing-13 will be produced at the German factory XFAB.
SPC "Elvis", Zelenograd
Elvis is currently actively promoting its own products to the space industry, actively cooperating with the Micron plant in terms of technology and the Submicron SRI in terms of the production of space equipment. Elvis also participates in the international working group on the development of the standard for the transfer of data from SpaceWire, to which the European Space Agency and, possibly, Roscosmos will be transferred in the near future.
Test swallow "Elvis" in terms of aerospace applications - memory chip 1657RU1U (SRAM 4 Mbit), manufactured according to foreign technology 250 nm.
I would not like to say something, but on the detailed page with information about the chip (there would be more such) in the parameters of radiation resistance you can find the following: “total accumulated dose 330 krad, KTZ 500 krad”, and in the parameters recorded in the factors according to GOST (below table), the figure is different. What exactly - I will not say, because this GOST is secret, unlike the analogous standards of our American sworn friends. In addition, it is rumored that the tests of the first microcircuits were carried out by some specially trained techniques, so that everything works really well, there are some doubts.
18928 - a dual-core processor with a general-purpose core (compatible with MIPS-32) and a digital signal processing core. The clock frequency is 80 MHz, 480 MFLOPs in floating-point calculations, a wide range of interfaces - in total a rather serious machine is obtained. Technology, like the previous memory circuit, 250 nm CMOS (foreign).
Now "Elvis" is developing several similar processors on micron technologies 180 nm and 250 nm SOI, but the results have not yet gone to the series. The “Multibort” chipset being developed was recently presented at the “New Electronics” exhibition, and on the Elvis website I found this document (download link pdf)
The kit contains more than twenty microcircuits with the start year of production up to 2014: microprocessors, ADCs, external device controllers and switches, which allow you to fully organize the data network on board the spacecraft.
After working out solutions in foreign factories, Elvis makes all promising microchips completely in Russia at Micron (design standards are 180 and 90 nm).
NIISI RAS, Moscow
The Research Institute of System Studies of the Russian Academy of Sciences (NIISI RAS) has the most extensive experience among Russian developers of space processors (since 2001) and produces a series of microprocessors with the COMDIV command system (which has a certain similarity with MIPS32). ( Link to Wikipedia, read the sources at the bottom of the page).
5890BE1T (KOMDIV32-S) - 32-bit microprocessor with built-in interface controller, 33 MHz, 500 nm SOI technology. Judging by open sources, it has long and successfully been flying in spacecraft control systems.
5890BM1T - 32-bit microprocessor with high resistance to single failures. 33 MHz, 500 nm SOI.
5890VG1T - two-channel interface controller interface MIL-STD-1553.
1900VM2T (Reserve-32) is a 32-bit microprocessor with hardware triplication at the level of core components and protection from single failures. Clock frequency 66 MHz, 350 nm technology.
Static RAM (SRAM) 1 Mbit, access time 30 ns. 350 nm SOI technology.
The four processors mentioned above are mass-produced, and for 2014 and 2015, the start of production of four more processors is announced.
1907VM014 - 32 bits, frequency 100 MHz, technology 250 nm. On-chip system controller, SpaceWire, Ethernet and MIL-STD-1553 interface.
1907BM038 - 32 bits, 125 MHz frequency, 250 nm technology. On-chip SpaceWire and Serial RapidIO interfaces.
1907BM044 - 32 bits, 66 MHz, 250 nm, built-in system controller, nuclear tripling and increased resistance to single faults, SpaceWire.
1907BM028 - 64 bits, 150 MHz, 250 nm, built-in system controller, two levels of cache (the rest - one), Serial RapidIO, Ethernet.
All NIISI processors made on SOI technology, resistance to the total absorbed dose, sufficient for space applications, have no thyristor effect, and also apply (for all but 5890EV1T) special measures to increase resistance to single failures (Hamming codes in the cache memory , special cells SRAM, hardware triple-up at the level of composite blocks of the processor core).
Here is another interesting article from the authors from NIISI in the American scientific journal Transactions on Nuclear Science - about some kind of robust 32-bit processor K32R on SOI technology.
In addition, NIISI has the following OCD: “Development of a 128-bit high-performance microprocessor on CNS / SOI 0.25 micron structures compatible with the COMDIV architecture for digital signal processing systems”, Scheme-10 cipher. That is, it is no longer 32 or 64 bits, but as many as 128. The work started in 2012.
STC "Module", Moscow
The module produces DSP processors with its own original architecture and computational modules based on its own and foreign processors, including for space applications.
Main own chip "Module" DSP Neuromatrix (L18791). Clock frequency 40 MHz, 0.5 micron technology (Samsung).
Microassembly 2605VG1T - logic and transceiver interface MIL-STD-1553 with built-in memory.
1895BA1T - the logical part of the controller channel interface MIL-STD-1553
1879BA1T - interface controller for communication of the computing processor with the MIL-STD-1553 interface
Research Institute "Progress", Moscow
NIIMA Progress is one of the leading developers of GLONASS receivers and transmitters.
5512BP2F - system on a chip with a microprocessor core and a basic matrix chip programmed for the needs of the user. Technology 180 nm ("Micron"), the operating frequency of the processor is 150 MHz, the arithmetic coprocessor is 50 MHz. Processor core - "Quark" of the company KM211
"VLSI with MP Somak Almaz-9" - the same set of peripherals with a different core and technology KNI 240 nm ("Micron") to improve radiation resistance. Completion of the ROC in 2014.
Design Center "Union", Zelenograd
DC "Soyuz" is developing analog-digital base matrix crystals based on the "micron" technology SOI 0.24 microns. R & D completion scheduled for 2014 and 2015
5400BK1T, 5400BK2U - general purpose. 110k digital gates, 50k “analog” transistors, 56 op-amps, 56 comparators, 6 ADCs, 6 DACs, a voltage source and other units
5400TP014 - precision. 110k digital gates, 10k "analog" transistors, 3-OU, 2 ADCs, 2 DACs, 2 UVH, voltage source, etc.
PS The basic matrix crystal is a microchip from the basic cells without several top layers of metallization, with the help of which the cells can be connected in the way required by the customer. A sort of antediluvian analogue FPGA. It is still in demand, which is typical.
NPK "Technology Center" MIET, Zelenograd
NPK "Technology Center" MIET works with "Micron" and has its own factory with design standards of 1.5 microns, on which they successfully make radiation-resistant microcircuits of low integration and basic matrix crystals, as well as semi-custom VLSI based on these BMC - interface controllers, external devices, transceivers, etc.
"Multiklet", Yekaterinburg
«» , , 2015- . , , , . — .
«», -, -,
– «» (). , , , , , . , – «»and two incarnations of the Voronezh plant of semiconductor devices - VZPP-S (with is an assembly) and VZPP-Micron .
The main products of all three enterprises are FPGAs and microcircuits of a small degree of integration. With the latter, everything is more or less clear: these are, probably, from the Soviet times (on the relevant design standards) discrete elements of power electronics and logical chips of the 1504, 1505, etc. series. Surprisingly, but a fact: the main source of income for VZPP-Micron, judging by the Micron website, is export, and the website of the VZPP-Micron itself is generally English-language.
It is more interesting with FPGAs, because they are obviously intended for import substitution of Altera products, with which they are software compatible. Develops them, apparently, KTC "Electronics".
50 200 , XFAB. , «».
,
, – « » ( ).
(8- MCS-51, AVR, 16- MCS-96, C166), DSP ( Texas Instruments), / . , – XFAB.
:
183032/183032 – 8 , 12/16 , 256 ( !), , Intel 80C51FA
187405 – 16 , 20 , 488 SRAM, Intel 196
, , , 0,5 .
, – , DSP SPARC ( , LEON3, ). , , SPARC- (). SPARC.
– «»
«» — ( – 1 , SRAM, ), , , , . – . Here are a couple of examples:
16551 – 256 -, 150 , 10 .
16591 – SRAM 256 , 50 . .
1666014 – FRAM 1 .
18814 – 8- (AVR) -, SRAM . 4 .
18801 – 8- (MSC-51) 10- 24 .
14512 – .
5577CX3T – 2000 .
Everything is not as bad as it seems (and how it could be). There is a lot of work on the creation of domestic electronics for military and space applications, and it is quite possible to do without American chips in the foreseeable future.
We are still lagging behind strongly, although not as catastrophically as in the "ordinary" microelectronics. Now, however, the question is not about catching up and overtaking, but about not staying at the bottom of the trough. The question, however, is still not idle, because if someone on the other side of the border guesses to close the deliveries of not ready-made chips, but consumables for the Micron plant, then that's all to be covered with a copper basin.
, ( , , , ), – , , . , , .
This can complete the review, but I have a small postscript about what else awaits us in the coming years. The vast majority of the works presented in the review are financed from the budget, which means that information can be found on the public procurement website. If you google it for a work called “Processing-13”, then there is a very interesting document from May 2012 (link to download the file).
, , , . « - » , 2014- 2015- . , , 2012.
110/11-1-12.04, 111/11-1-12.04, 112/11-1-12.04, 117/11-1-18.04.
« - , 32- RISC- », «-10».
« - SpaceFibre», «-11». SpaceFibre – «»
« 32- SPARC V8 , SpaceFibre, CAN 2.0 B, PCI 2.2, Ethernet USB 2.0», «-12» – LEON .
« 32 RISC ARM , , - 0,25…0,3 », «-13» — «», .
« - DSP- », «-14» — , , .
« - , », «-15».
« 8...16 », «-6».
« 4 », «-8».
« (FRAM) 1 », «-9». . , ? . , .
« 128- / 0,25 , , », «-10». – .
« - 0,25 , 6», «-12».
« - 1867, 1830, 1874 - », «-13». , . «» .
« - 30-50 .», «-5».
Development of a number of radiation-resistant BMK: BMK-400 and BMK-1000, code number Almaz-6. This design center "Union" is almost certainly. Those same two BMK above.
OCD "Development of a number of radiation-resistant BIS digital frequency synthesizer", cipher "Digra-16".
Design development of an analog key chip with a frequency band of at least 1 ... 2 GHz, cipher "Digit-17".
Design and development of special resistant DC-DC voltage converters, Power-7 cipher. And here are two examples, very far from microprocessors. That is, another element base is being developed, which is very good.
R & D "Development of a radiation-resistant quadrature modulator for the operating frequency range of 30-40 GHz", code "Vysotka-13".
« - , , pin- SiC - - », «-14».
« », «-11». «» ?
« , , , », «-4». , . «»? www.svetlanajsc.ru/index.php/ru
« 16- / », «-12».
« « » /GPS », «-13». «».
« », «-14». .
« -2 «» / 16 /», «-8».
«». , .
On this, perhaps, everything.
Such a development of events may negatively affect the state of the Russian aerospace and defense industries, because the annual import of electronics for the space industry is two billion dollars, and these are chips that are critical for the operation of satellites. Some officials (see the article on the link) have already begun to panic and talk about buying electronics in China, who allegedly started production of everything necessary. I want to talk a little bit about what chips are being developed and manufactured for the space industry in Russia. I’ll say right away that this review in no way claims to be complete and will concern microchips (and silicon), whereas the needs of space are not limited to them - we also need passive components, microwave devices, power discrete elements and much more. partially imported and partially developed and produced in Russia. To describe this all is a completely impossible task, and not very necessary, because the purpose of this review is not to describe everything that is, but to show that everything is not as bad as it seems to alarmists. The specifics of the industry are such that almost all of the really important information is closed, but something interesting could still be dug up.
A small digression, a reminder: chips for working in space should be resistant to radiation. About why this is so and how exactly the radiation affects electronic devices, you can read here and here .
The main problem with the use of imported components in the aerospace and military industries is that these components are usually not designed to work in such conditions (that is, they are ordinary consumer goods originally intended for irons and refrigerators). This situation arose in the nineties, when there was simply nothing else, and what happened was worth more than the space equipment developers could afford. That is why the terms of active functioning of Russian satellites are still very seriously behind American or European ones. For example, the beautiful American on-board computer RAD750 (standing on the Curiosity rover) was never sold to Russia, and its counterparts in Russia did not exist until recently. So the problem with imports did not arise yesterday, and they began to be solved for a long time. In 2007, the federal target program “Development of Electronic Component Base and Radio Electronics” for 2008-2011 was adopted. Further, the final year of the program was 2015, and soon we should see its results, including the replacement of imported components in the space and defense industry with Russian developments.
')
There are lists of microcircuits to be replaced, but they, unfortunately, are classified. In open sources, you can find information (a link to download pdf) that the number of positions in these lists is hundreds or even thousands (and these are only chips, not to mention passive elements such as high-precision resistors, which are even greater). The situation is complicated by the fact that, in the overwhelming majority of cases, hardware developers want them to do “exactly the same, but Russian” (that is, it is copying foreign developments down to pin-compatibility), while the list can certainly be shortened several times due to development of a set of systems on a chip, uniform for all developers and customizable to the needs of specific users. This also includes the unification of data transfer interfaces (which is not available now) and other fairly obvious measures. Work in this direction, as far as I know, is underway, but you understand that it can turn out to be even more expensive to develop onboard equipment from scratch than to develop more chips than the required minimum.
Factories
Actually, a review of developers and manufacturers of microcircuits for the space industry (and partly military, since they are quite strongly interconnected) should start with production, because there are significantly fewer of them in Russia than developers, and it is precisely in this area that the lagging behind is abroad. I will not dwell in detail on the factories with design norms of more than half a micron, because they are hopelessly outdated, and some new developments are being carried out on them except of despair. Nevertheless, a lot of things are being done on them, first of all - microcircuits of a small degree of integration, power electronics, microwave, etc. But there are only four factories with more or less modern design standards.
Threat Here , by the way, you can read the distracted gaze of two years ago.
Zelenograd plant "Micron" (here is a post with beautiful pictures) declares the presence of technologies on its website :
1) 180 nm volumetric technology with EEPROM is already a well-known technology for developers, which employs almost everyone who has this capability. On it, for example, chips for tickets of the Moscow metro are made. The presence of an EEPROM is useful for applications such as embedded microcontrollers, which are necessary not only for space, but also for “civilian” applications. Design whales are available to developers.
2) 250 nm silicon on an insulator (SOI) - according to this technology, separate developments are already underway, but I have not heard about finished products yet. The technology, apparently, is a partial port of 180 nm on silicon wafers on an insulator. Design whales are available to developers.
3) 180 nm CED - declared 2012 by the year; in practice, nothing new has been heard about it for quite some time. That is, it is in development, but when it will be ready for design is not very clear. In any case, I have not found such information anywhere.
4) 90 nm volume. Most recently, Micron licensed software for physical verification for this technology from Cadence. No mention of specific developments on it has yet been found; only general words that they are underway.
5) Quite recently, 65nm technology appeared in the news, but everything is still at the level of test crystals. Mass production promise this year .
An important thing worth noting is a quote from the Micron site: “More than 50 companies from 12 countries of the world delivered equipment and materials suppliers, infrastructure partners.” This is something that is often overlooked, speaking about overcoming the import ban on domestic production - a huge amount of spare parts and consumables are also imported, and a ban on their import can be more effective than a ban on the import of ready-made microchips. I hope that this will not happen in the near future, and that there will be money and resources to organize the production of consumables in Russia.
Angstrom (and Angstrom-T), Zelenograd
1) 1.2 micron silicon on sapphire (CNS) - technology for thousands of years, but it is still in demand (apparently, conservative military customers for chips with high radiation resistance, for which reliability and validation of solutions is more important than their timeliness).
2) 0.6 µm, bulk silicon, silicon on sapphire, silicon on insulator, the possibility of manufacturing EEPROM, BiCMOS, IGBT. Apparently, a good process for power electronics.
3) "Created" technologies with design standards of 350-250 nm.
4) “Created” technologies with design standards 130, 110 and 90 nm.
The history of the "creation" of technology at Angstrom is long, difficult and not yet resolved. The process and equipment 130 nm were purchased from AMD, 90 nm - from IBM. Every year for five years they say that everything is about to be, but so far nothing.
Regarding the radiation resistance on the Angstrem website, by the way, the best nonsense was written about the fact that their silicon on sapphire is the only suitable technical process in Russia and that practice has shown the impossibility of making radiation-resistant circuits on technologies less than 250 nm. Let's see what happens when they launch 130 and 90 nm :-) Physics are unlikely to learn, but the presence of production practices can have a positive effect on the understanding that in states, space chips are already being designed at 45 nm, and in Russia - at 65 (true foreign) .
Research Institute of System Studies of the Russian Academy of Sciences, Moscow
The factory of NIISI RAS is located on the territory of the Kurchatov Institute in Moscow and possesses technologies with design standards of 500, 350 and 250 nanometers on bulk silicon and SOI plates. Initially, it is not intended for large-scale production and is positioned as a “research factory of the Academy of Sciences”. Most of the microcircuits produced here are developed by NIISI itself, but the factory also works with external customers, for example, with Voronezh NIIET, which produces its radiation-resistant microcontrollers here.
There are no other details, and the latest open publications about the factory are almost the time of its opening.
Integral, Minsk
The Minsk Integral is considered to be our military and other institutions by a domestic enterprise with all the attendant circumstances. Interesting, isn't it?
The main technologies of Integral are old, with design standards of 0.8 microns and more, but in the last few years, Belarusians have independently designed and launched technologies of 0.5 and 0.35 microns on bulk silicon and SOI. They have only three metals (which is not enough for a microprocessor), but Integral develops SRAM microcircuits with a capacity of 1 Mbit and high radiation resistance, as well as analog microcircuits.
Reports from scientific conferences also feature 0.18 micron technology and 0.5-0.25 micron BiCMOS, flash memory (the only one in the CIS?) And FRAM.
The overwhelming majority (if not all) of the developments on their technologies, Integral conducts independently.
That's all. Looks a little sad, is not it? Yes, radiation-resistant electronics for space do not always require the same design standards as usual, and the lag in several generations is not very scary (the Curiousity rover at 250 nm works fine), but still the new fast-resistant BAE systems are developed using the 45 nm technology, and so far we have not reached the industrial development of 90 nm. On the other hand, five years ago there was practically nothing of this, and today Russian plants have quite real opportunities to provide astronautics with microelectronics of their own production.
Developers
Now about the developers. There are many more of them than manufacturers, but until recently it was normal practice to manufacture microchips developed in Russia somewhere abroad, for example, in Taiwan (TSMC), in Germany (XFAB) or in Israel (Tower). At a time when there was nothing better than 0.8 microns in Russia, even the military closed their eyes to it, believing that “developed in Russia” = “Russian”. Now times have changed a little, and Russian developers are increasingly producing their products on Micron (that is, there are more than metro chips doing there).
It is also necessary to note the fact that the overwhelming majority of Russian microelectronics developers are somehow tied to budget money and large orders, especially space or military, and especially commercial orders are few. On the other hand, a significant share of the profits of several enterprises (for example, VZPP-Micron and Angstrem) is export. However, I digress from the main topic of the review, so I’ll confine myself to the observation that the products presented below are not all that are produced, and for many companies space is not the main topic.
Milander, Zelenograd
PKK Milandr, CJSC, based in Zelenograd, is a company with a twenty-year history and, more importantly for us, with the most detailed website among all Russian microelectronic companies. It managed to find this:
1645RU2T - static RAM (SRAM) with a capacity of 64 Kbps. In serial production since 2008.
1645RU5U - static RAM (SRAM) with a capacity of 4 Mbit. OCD ends in 2014.
Judging by the year of production, the first microcircuit is produced on some very old technology, the second - 180 nm (probably on the "Micron").
Under the link (caution, traffic) you can find photos of the radiation-resistant 8-bit microcontroller 1886BE10 (analogue of PIC17), for which reason for some reason there is no information about it.
The technology is Micron 180 nm, which is radically resistant and contains a complete stuffing made from ring transistors and multi-transistor storage elements. There is no accurate data in the public domain, but a chip with such methods of protection must withstand a nuclear explosion, not like a long-term flight in space.
16452 - single programmable ROM (antifuse) with a capacity of 256 kbps. ROC was commissioned in 2013
Here you can see how it looks. The design standards, judging by the opened crystal, are 680 nm.
5576RT1U - single programmable ROM (antifuse) with a capacity of 1 Mbit. The design and development works were commissioned in 2013. The design norms are most likely 180 nm (the Micron technology).
For other Milandra microcircuits, radiation resistance is not stated, however, for example, in the news you can find the following line on the site: “The parameters of resistance to special factors for 1310PN1U microcircuits were updated (significantly improved)”. 1310 is an inductive power converter for which radiation resistance is not claimed. If all the microcircuits supplied with the fifth acceptance have at least some resistance to radiation, then Milandr still has a fairly wide range of interface microcircuits, power management, and ADC / DAC.
The promising development of Milandra is their first radiation-resistant and fail-safe microprocessor. It does not yet have its own designation and is presented at various conferences under the name " Processing-13 ." (link to download pdf) By reference - a presentation about the processor device and its design in terms of ensuring radiation resistance. There are interesting and controversial decisions, but it looks impressive (except for the joint work of the nuclei, perhaps).
The processor is a dual-core ARM Cortex-M4F with separate core operation modes and hardware duplication. Clock frequency - 100 MHz, SRAM 32 KB, 128 KB ROM, a wide range of interfaces and analog peripherals.
Processing-13 will be produced at the German factory XFAB.
SPC "Elvis", Zelenograd
Elvis is currently actively promoting its own products to the space industry, actively cooperating with the Micron plant in terms of technology and the Submicron SRI in terms of the production of space equipment. Elvis also participates in the international working group on the development of the standard for the transfer of data from SpaceWire, to which the European Space Agency and, possibly, Roscosmos will be transferred in the near future.
Test swallow "Elvis" in terms of aerospace applications - memory chip 1657RU1U (SRAM 4 Mbit), manufactured according to foreign technology 250 nm.
I would not like to say something, but on the detailed page with information about the chip (there would be more such) in the parameters of radiation resistance you can find the following: “total accumulated dose 330 krad, KTZ 500 krad”, and in the parameters recorded in the factors according to GOST (below table), the figure is different. What exactly - I will not say, because this GOST is secret, unlike the analogous standards of our American sworn friends. In addition, it is rumored that the tests of the first microcircuits were carried out by some specially trained techniques, so that everything works really well, there are some doubts.
18928 - a dual-core processor with a general-purpose core (compatible with MIPS-32) and a digital signal processing core. The clock frequency is 80 MHz, 480 MFLOPs in floating-point calculations, a wide range of interfaces - in total a rather serious machine is obtained. Technology, like the previous memory circuit, 250 nm CMOS (foreign).
Now "Elvis" is developing several similar processors on micron technologies 180 nm and 250 nm SOI, but the results have not yet gone to the series. The “Multibort” chipset being developed was recently presented at the “New Electronics” exhibition, and on the Elvis website I found this document (download link pdf)
The kit contains more than twenty microcircuits with the start year of production up to 2014: microprocessors, ADCs, external device controllers and switches, which allow you to fully organize the data network on board the spacecraft.
After working out solutions in foreign factories, Elvis makes all promising microchips completely in Russia at Micron (design standards are 180 and 90 nm).
NIISI RAS, Moscow
The Research Institute of System Studies of the Russian Academy of Sciences (NIISI RAS) has the most extensive experience among Russian developers of space processors (since 2001) and produces a series of microprocessors with the COMDIV command system (which has a certain similarity with MIPS32). ( Link to Wikipedia, read the sources at the bottom of the page).
5890BE1T (KOMDIV32-S) - 32-bit microprocessor with built-in interface controller, 33 MHz, 500 nm SOI technology. Judging by open sources, it has long and successfully been flying in spacecraft control systems.
5890BM1T - 32-bit microprocessor with high resistance to single failures. 33 MHz, 500 nm SOI.
5890VG1T - two-channel interface controller interface MIL-STD-1553.
1900VM2T (Reserve-32) is a 32-bit microprocessor with hardware triplication at the level of core components and protection from single failures. Clock frequency 66 MHz, 350 nm technology.
Static RAM (SRAM) 1 Mbit, access time 30 ns. 350 nm SOI technology.
The four processors mentioned above are mass-produced, and for 2014 and 2015, the start of production of four more processors is announced.
1907VM014 - 32 bits, frequency 100 MHz, technology 250 nm. On-chip system controller, SpaceWire, Ethernet and MIL-STD-1553 interface.
1907BM038 - 32 bits, 125 MHz frequency, 250 nm technology. On-chip SpaceWire and Serial RapidIO interfaces.
1907BM044 - 32 bits, 66 MHz, 250 nm, built-in system controller, nuclear tripling and increased resistance to single faults, SpaceWire.
1907BM028 - 64 bits, 150 MHz, 250 nm, built-in system controller, two levels of cache (the rest - one), Serial RapidIO, Ethernet.
All NIISI processors made on SOI technology, resistance to the total absorbed dose, sufficient for space applications, have no thyristor effect, and also apply (for all but 5890EV1T) special measures to increase resistance to single failures (Hamming codes in the cache memory , special cells SRAM, hardware triple-up at the level of composite blocks of the processor core).
Here is another interesting article from the authors from NIISI in the American scientific journal Transactions on Nuclear Science - about some kind of robust 32-bit processor K32R on SOI technology.
In addition, NIISI has the following OCD: “Development of a 128-bit high-performance microprocessor on CNS / SOI 0.25 micron structures compatible with the COMDIV architecture for digital signal processing systems”, Scheme-10 cipher. That is, it is no longer 32 or 64 bits, but as many as 128. The work started in 2012.
STC "Module", Moscow
The module produces DSP processors with its own original architecture and computational modules based on its own and foreign processors, including for space applications.
Main own chip "Module" DSP Neuromatrix (L18791). Clock frequency 40 MHz, 0.5 micron technology (Samsung).
Microassembly 2605VG1T - logic and transceiver interface MIL-STD-1553 with built-in memory.
1895BA1T - the logical part of the controller channel interface MIL-STD-1553
1879BA1T - interface controller for communication of the computing processor with the MIL-STD-1553 interface
Research Institute "Progress", Moscow
NIIMA Progress is one of the leading developers of GLONASS receivers and transmitters.
5512BP2F - system on a chip with a microprocessor core and a basic matrix chip programmed for the needs of the user. Technology 180 nm ("Micron"), the operating frequency of the processor is 150 MHz, the arithmetic coprocessor is 50 MHz. Processor core - "Quark" of the company KM211
"VLSI with MP Somak Almaz-9" - the same set of peripherals with a different core and technology KNI 240 nm ("Micron") to improve radiation resistance. Completion of the ROC in 2014.
Design Center "Union", Zelenograd
DC "Soyuz" is developing analog-digital base matrix crystals based on the "micron" technology SOI 0.24 microns. R & D completion scheduled for 2014 and 2015
5400BK1T, 5400BK2U - general purpose. 110k digital gates, 50k “analog” transistors, 56 op-amps, 56 comparators, 6 ADCs, 6 DACs, a voltage source and other units
5400TP014 - precision. 110k digital gates, 10k "analog" transistors, 3-OU, 2 ADCs, 2 DACs, 2 UVH, voltage source, etc.
PS The basic matrix crystal is a microchip from the basic cells without several top layers of metallization, with the help of which the cells can be connected in the way required by the customer. A sort of antediluvian analogue FPGA. It is still in demand, which is typical.
NPK "Technology Center" MIET, Zelenograd
NPK "Technology Center" MIET works with "Micron" and has its own factory with design standards of 1.5 microns, on which they successfully make radiation-resistant microcircuits of low integration and basic matrix crystals, as well as semi-custom VLSI based on these BMC - interface controllers, external devices, transceivers, etc.
"Multiklet", Yekaterinburg
«» , , 2015- . , , , . — .
«», -, -,
– «» (). , , , , , . , – «»and two incarnations of the Voronezh plant of semiconductor devices - VZPP-S (with is an assembly) and VZPP-Micron .
The main products of all three enterprises are FPGAs and microcircuits of a small degree of integration. With the latter, everything is more or less clear: these are, probably, from the Soviet times (on the relevant design standards) discrete elements of power electronics and logical chips of the 1504, 1505, etc. series. Surprisingly, but a fact: the main source of income for VZPP-Micron, judging by the Micron website, is export, and the website of the VZPP-Micron itself is generally English-language.
It is more interesting with FPGAs, because they are obviously intended for import substitution of Altera products, with which they are software compatible. Develops them, apparently, KTC "Electronics".
50 200 , XFAB. , «».
,
, – « » ( ).
(8- MCS-51, AVR, 16- MCS-96, C166), DSP ( Texas Instruments), / . , – XFAB.
:
183032/183032 – 8 , 12/16 , 256 ( !), , Intel 80C51FA
187405 – 16 , 20 , 488 SRAM, Intel 196
, , , 0,5 .
, – , DSP SPARC ( , LEON3, ). , , SPARC- (). SPARC.
– «»
«» — ( – 1 , SRAM, ), , , , . – . Here are a couple of examples:
16551 – 256 -, 150 , 10 .
16591 – SRAM 256 , 50 . .
1666014 – FRAM 1 .
18814 – 8- (AVR) -, SRAM . 4 .
18801 – 8- (MSC-51) 10- 24 .
14512 – .
5577CX3T – 2000 .
findings
Everything is not as bad as it seems (and how it could be). There is a lot of work on the creation of domestic electronics for military and space applications, and it is quite possible to do without American chips in the foreseeable future.
We are still lagging behind strongly, although not as catastrophically as in the "ordinary" microelectronics. Now, however, the question is not about catching up and overtaking, but about not staying at the bottom of the trough. The question, however, is still not idle, because if someone on the other side of the border guesses to close the deliveries of not ready-made chips, but consumables for the Micron plant, then that's all to be covered with a copper basin.
, ( , , , ), – , , . , , .
This can complete the review, but I have a small postscript about what else awaits us in the coming years. The vast majority of the works presented in the review are financed from the budget, which means that information can be found on the public procurement website. If you google it for a work called “Processing-13”, then there is a very interesting document from May 2012 (link to download the file).
, , , . « - » , 2014- 2015- . , , 2012.
110/11-1-12.04, 111/11-1-12.04, 112/11-1-12.04, 117/11-1-18.04.
« - , 32- RISC- », «-10».
« - SpaceFibre», «-11». SpaceFibre – «»
« 32- SPARC V8 , SpaceFibre, CAN 2.0 B, PCI 2.2, Ethernet USB 2.0», «-12» – LEON .
« 32 RISC ARM , , - 0,25…0,3 », «-13» — «», .
« - DSP- », «-14» — , , .
« - , », «-15».
« 8...16 », «-6».
« 4 », «-8».
« (FRAM) 1 », «-9». . , ? . , .
« 128- / 0,25 , , », «-10». – .
« - 0,25 , 6», «-12».
« - 1867, 1830, 1874 - », «-13». , . «» .
« - 30-50 .», «-5».
Development of a number of radiation-resistant BMK: BMK-400 and BMK-1000, code number Almaz-6. This design center "Union" is almost certainly. Those same two BMK above.
OCD "Development of a number of radiation-resistant BIS digital frequency synthesizer", cipher "Digra-16".
Design development of an analog key chip with a frequency band of at least 1 ... 2 GHz, cipher "Digit-17".
Design and development of special resistant DC-DC voltage converters, Power-7 cipher. And here are two examples, very far from microprocessors. That is, another element base is being developed, which is very good.
R & D "Development of a radiation-resistant quadrature modulator for the operating frequency range of 30-40 GHz", code "Vysotka-13".
« - , , pin- SiC - - », «-14».
« », «-11». «» ?
« , , , », «-4». , . «»? www.svetlanajsc.ru/index.php/ru
« 16- / », «-12».
« « » /GPS », «-13». «».
« », «-14». .
« -2 «» / 16 /», «-8».
«». , .
On this, perhaps, everything.
Source: https://habr.com/ru/post/217427/
All Articles