Russian Electronics: Semiconductor and Processors

kvs wrote:

I was involved with occultation instrument development in Canada. Please don't prance around making it like Russia cannot
design CCDs for space based systems and yapping about "old" stuff. It is physically impossible to rad harden dense IC components.
All that one can hope for is noise post processing. At least with RAM one can build in some bit error control. With a CCD you cannot
remove the noise in hardware.

The radhard controllers / computers in the links shown a no bigger than 16megs memory CCD controller.

It is not fit to drive anything bigger than say 2 megapixel circuit ,and that is definitively not enough for the job.


The manufacturing of the radhard component COMPLETELY different from the normal semiconductors.

The normal stuff manufactured /grown on silicon wafer, that is conductive.Means if a gamma photon dislocating a charge then it will propagate to the nearest active element,and make a false signal there. The silicon is the starting of the production.


In the case of radhard elements the process starting with an insulator, like sapphire ( the RF is the main supplier of the single crystal sapphire wafers)

Actually the sapphire is the starting point for the GaN circuits as well, as it seems from the articles on the net.

Anyway, example a CCD for non-radhard environment start from silicon, for radhard environment from sapphire.

http://ieeexplore.ieee.org/document/1477639?reload=true
This conference happened in 1973, about sapphire based CCD .

Just as a coincidence , the first US satellite with CCD launched in 1976.

http://www.epj-conferences.org/articles/epjconf/pdf/2017/01/epjconf_spectro2017_01002.pdf
This paper above has been published this year, means it is quite recent.

Showing how to grown check the MgO thickness/cover on sapphire substrate.

MgO is insulator as well , maybe they did it to deposit GaN ? Seems like the sapphire not matching the exact latice spacing of GaN.

A Russian company here :
http://technospark.ru/en/proekty/

Advanced laser microsystems

Founded: 2014

Product/service: Laser lift-off removal of the sapphire wafer

Seems like it is related to the radhard parts manufacturing.

http://indico.cfr.mephi.ru/event/4/session/29/contribution/19/material/slides/0.pdf

Above presentation contain the high speed radhard boards usable for modern satellites.
There is one circuit schematic, showing one central core with 26 bit address bus, capable to adders 256 MB of memory - enough for a spy satellite.


Everywhere mention SRAMs , that can be used only in space, due to the problems with capacitors.

Example the SRAM requirement means that the radhard electronic doesn't share even the memory technology with the non-radhard components.

https://en.wikipedia.org/wiki/KOMDIV-32


I checked , the IRIDIUM satellites using 7 pcs of 200 MHz powerPC CPUs, and each of them has sum 2.5 Mbit/sec bandwidth, on 1100 channel.

The RF has to make high frequency radhard RF modules as well.

Again, it has to start with sapphire - GaN.

Then check Rostec website. They say just that. They been working on gan quite a bit.

miketheterrible wrote:Then check Rostec website. They say just that. They been working on gan quite a bit.

C'mon, mate, the name of the game is "get as much specs as possible from the internet" . : )


example:
Crocus NanoElectronics, a portfolio company of Russia’s nanotech giant, Rusnano, and the Moscow Institute of Physics and Technology (MIPT), a leading Russian technology university, continue their joint research program in an effort to develop a next gen STT-MRAM magnetic memory and test production technology for that.

The partners expect to jointly develop new materials, design devices, and come up with new modeling and control methods. If they pull it off, the groundwork will be laid for the production of STT-MRAM-based items on the premises of Crocus NanoElectronics.

The Spin-Transfer Torque Magnetic Random Access Memory (STT-MRAM) technology is built around the idea of transferring spins for re-saving memory cells. Using this effect in making conventional magneto-resistive memory helps reduce electrical current required for storing data in a cell, and enables production to 90-to-22 nanometer design rule and lower.

The international majors manufacturing DRAM dynamic memory all push their own STT-MRAM development programs, as they believe the technology is most likely to replace DRAM in a near future.

Crocus NanoElectronics is Europe’s only company, and one of the world’s few, to provide a commercial hub to manufacture magnetic tunnel structure based memory and sensors to 90/65 nanometer design rule on wafers 300mm in diameter.
http://marchmontnews.com/Technology-Innovation/Central-regions/21597-Russia-pushes-its-STT-MRAM-next-gen-memory-program-.html

I'm not holding your hand. It is a few clicks away after hitting load more on Rostec website. If kVS wants to do it for you, fine. I'm not.

miketheterrible wrote:I'm not holding your hand. It is a few clicks away after hitting load more on Rostec website. If kVS wants to do it for you, fine. I'm not.

That website doesn't contain anything.

Couple of catalogue.

http://rostec.ru/news/4519353
http://rostec.ru/news/4519307

You can find news on Russian GaN since 2013

miketheterrible wrote:http://rostec.ru/news/4519353
http://rostec.ru/news/4519307

You can find news on Russian GaN since 2013

The semiconductor industry is a quite interesting one.

The machines are extremely expensive, each of them cost 8 million $ or more.

The building/utility extremely expensive as well, start around 1 billion $ and can be 10 billion $ or more.


The process of a new IC start with the design,afterwards they have to translate the design to the manufacturing process ( this step can cost 10 or 100 millions of $ as well if the technology is less than 90nm).

Then,you start the sample manufacturing process.

The challenge is that you have a lot of machine in the shop,and you have to do say 80 passes on the machines to make the final finished wafer.

It can take weeks or months.

If you make radhard ICs then the number of passes increase.

It has two different result:
1. Radhard component needs more time to make and fine tune - say it takes 3 weeks for a non radhard,and 6 weeks for a radhard component to make. Means if you need 6 iteration then the development time will be 18 weeks for the normal , and 36 weeks for the radhard. But the radhard needs more step, means the time can be way more than this calculation.
2. The 3-4 billion $ plant can make say x amount of non-radhard ICs, and say x/2 radhard ICs. But you have to use the normal manufacturing line for sampling, means due to the above the radhard IC development can burn like the hell the available capacity of the fab.

The above is the main reason why everyone using 150 -90nm technology to make radhard ICs. Simply even the military can't afford the high cost of the new fabs, they using the old cheap ones instead.

So, the interesting can be to see the result of the serial manufacturing, and to see satellites flying up there : )

that is true, equipment is too expensive. And the output is low so production is low thus costs are high as well.

The other concept too is the need in performance vs not. In a lot of cases, you do not need the latest and greatest to do the job.

http://rostec.ru/news/4519778

Ruselectronics has developed its own varactors
The holding company has mastered the import substituting production of semiconductor diodes

Ruselectronics began to produce varactors Russian production to replace imported articles, used in the Russian electronic equipment.

In Tomsk Institute of semiconductor devices (NIIPP) of the holding company Ruselectronics developed silicon and GaAs varactors. According to their electrical parameters (capacitance, the overlap factor), they are not inferior to foreign analogues, and in some ways surpass them. In particular, the Russian GaAs devices have better performance on quality factor: its value reaches 1000 or more at a frequency of 50 MHz.

NIIPP developed the varactors in two versions – for the attachment and surface mounting. Institute engineers have changed the approach to the design and production of epitaxial structures, thereby reducing the manufacturing process of the varactors to increase the reproducibility of the parameters and the yield rate of products.

New design GaAs varactors are protected by Russian patent.

For the developments of NIIPP has received the National award in the field of import substitution "Priority 2016" in the category "electronics".

A varicap is a semiconductor diode that changes its capacitance proportional to the reverse applied voltage from a few to hundreds of picofarads. The varactors are convenient because, applying a constant bias voltage, can be remotely and almost bezynertsionnoj to change their capacity. The varactors are used for frequency tuning resonant circuits and filters, amplification and generation of microwave signal and locked loop.

NIIPP is one of the leading Russian developers and manufacturers of devices on the gallium arsenide has technological lines for silicon, III-V compounds, in-house production of photomasks and epitaxial structures, a wide range of measuring equipment.

While googling something about laptop accessories, Russian computer brands came to my mind, so while doing my search I found this report from RT from 2009

Asian laptop brands price out Russian producers
Russia's laptop computer producers may cease to exist, due to tough competition with Asian brands. Only Rover Computers hopes that new import regulations will make its life easier.
Large-scale laptop production in Russia began in the last ten years, as local products were a cheaper alternative to western brands.

The market grew 100 percent year on year brining huge profits to domestic producers.
The situation changed dramatically in 2004 after Asian brands, for example Acer and Asus, entered the growing Russian market with aggressive prices intended to build market share.

Rover Computers was the pioneer on the home market, assembling entry-level laptops.

Merlion followed Rover’s example and launched its own production line under iRU brand. Now iRU has already stopped producing laptops. In order to compete it needed huge investment and its shareholders decided that the project was not profitable enough. The company re-directed its activity to the desktop sector.

The sole survivor – Rover Computers – now has 10% of the market and assembles 250,000 mobile PCs per year.

The company’s management, however, is optimistic about the future.

“Nowadays the share of so-called grey imports is significant. However, new customs regulations will soon come into force. All devices will have an individual tracking number. Thus the conditions will become tougher for importers and more favourable for domestic producers. So the market structure will change,” said Konstantin Kupchik, Rover Computers Director General.

The company plans to increase its assembly capacity and almost double production over the next year.

https://www.rt.com/business/asian-laptop-brands-price-out-russian-producers/

its been 8 years, has this issue of Russian/locally manufactured computer brands losing to foreign brands been addressed by the government? with the current import substitution drive I think it is high time to look into this area as well

No.

Kraftway still makes them, so do others brands. Kraftway is quite huge in Russia.

Not much else to say. Russia has healthy competition.

Baikal Electronics started large scale production of microprocessors

https://sdelanounas.ru/blogs/91899/

https://vpk.name/news/180104_razvitie_opk_v_poiskah_putei_povyisheniya_nadezhnosti_elektroniki.html

In other words, a lot of talk about how Russian electronics are behind, quality, blah blah blah. So the idea is to create a center to test it all and where the government can order from.

http://rostec.ru/news/4520237

"Elbrus" will protect the plant from cyber attacks
United Ruselectronics is involved in the development of a secure system for power facilities

The United Ruselectronics holding participates in the development of secure microprocessor systems of relay protection and automation (RDA) for the Russian power substations. The system became a domestic computing systems microprocessor-based "Elbrus".

The system is designed for installation on new and reconstructed substations, has a high level of fault tolerance that guarantees the absence of "bookmarks" and protection from unauthorized interference in the work of power facilities. In the "Roselektronika" the project is the Institute of electronic control machines (INEUM). I. S. Bruk has developed a range of microprocessors "Elbrus" and computers based on them.

As the Deputy Director General of the United Ruselectronics holding Arseniy Brykin, in the framework of the program of import substitution today the company solves the problem of cyber defence, critical infrastructure, responsible for the livelihood of human settlements and the state as a whole.

"It's no secret that many foreign models of computers contain hidden channels, allowing to silently remove information remotely to affect systems, disabling them, and even create an emergency situation. In contrast, they created a system where electronic components, design, software – everything Russian. This guarantees a high level of cyber security software and hardware and allows the use of this technique for the solution of critical problems," he commented.

Create a secure system integrated into the automated control systems of technological processes of substations of the unified national electric network. It allows you to carry out the tasks of protection and control equipment in accordance with the IEC 61850 standard, equipped with self-diagnosis function and can operate without human intervention during the entire period of operation.

The system microprocessor relay protection and automation will be presented during the International exhibition-conference "relay protection and automation of power systems-2017," which opens April 25 in St. Petersburg.

Actualy this is the motibation behind all russia CPU development.


It is cheaper to buy an Intel CPU, but very few person know actualy how many backdor implemented into each cpu.

And that is not so safe for say an air deffense system.

Article about 28nm semiconductor production, can anyone elaborate?

http://bmpd.livejournal.com/2593209.html

Guarantee you it is Mikron who will get it. They have been pushing the 28nm and lower for some time now.

https://i.imgur.com/eezbRGE.jpg

http://www.planetanalog.com/author.asp?section_id=385&doc_id=561004#msgs

The 28nm is quite expensive, they need huge volume to justify that investment.

It is not for low volume military/aerospace ICs .

Is there still such a thing as low volume any more?

Everything has electronics these days including rifle scopes and most vehicles.

If they are setting up networks for their military every node needs to have Russian made chips and components to ensure security...

From what I am able to piece together it looks like entire output of Elbrus CPUs is being scooped up by military so it prevents them from hitting consumer market properly even though they could.

But it makes no difference. Consumer market already has access to various imported products, military gets locally made CPUs that they need and Elbrus gets sales. So far win-win.

Here is another example: Belorussian company that makes military displays is about to launch products based on Elbrus processors:

https://sdelanounas.ru/blogs/93378/



So one more Elbrus batch that goes into arms industry.

GarryB wrote:Is there still such a thing as low volume any more?

Everything has electronics these days including rifle scopes and most vehicles.

If they are setting up networks for their military every node needs to have Russian made chips and components to ensure security...

Small production lot with 28nm = 100 000 pcs.

AFAR Roselektronika able to repeat the shape of the object carrier
https://vpk.name/news/185079_afar_roselektroniki_sposobnyi_povtoryat_formu_obektanositelya.html

The specialists of the holding company Ruselectronics of state Corporation rostec, is developing transmit-receive modules of active phased antenna lattices (AFAR) X-band, capable of repeating the shape of the object carrier.


Development carried out by Tomsk JSC "research Institute of semiconductor devices", based on technology jointly low-temperature fired ceramics (LTCC, Low Temperature Co-Fired Ceramic) and provides high durability of the product to external influencing factors at a record low thickness – 30 mm. the modules can be combined without violating the radiator step in the canvas AFAR unlimited space, including a repetition of the shape of the object carrier.


This solution of the planar, multi-channel AESA modules provides application of small radar, in particular, unmanned aerial vehicles, and the need for development of advanced aerospace systems, for example, of the sixth generation fighter.


Modules under development include all of the elements of AESA - active elements, the antenna radiators, the distribution of microwave signals and control a secondary power source, the control digital controller with an interface circuit, a liquid cooling system. Have high properties of mutual integration, that reduces requirements for the supporting structure. This circumstance in aggregate with a minimum mass modules much easier for consumers the task of creating a AFAR to meet specific needs.


The modules have a significantly low cost.


Products will be presented at the VIII International military-sea salon which will pass in St.-Petersburg from 28 June to 2 July. The combined exhibition Roselektronika – pavilion 7, stand 701.


JSC "research Institute of semiconductor devices" develops and produces planar receiving, transmitter and transceiver modules AFAR ranges S, C, X, Ku, Ka for specific tasks of customers.


In accordance with the international classification under the X-band refers to the part of the electromagnetic spectrum with wavelengths from 3.75 to 2.5 cm and a frequency of 8 to 12 GHz. S-range – 15-7. 5 cm, 2-4 GHz; C-band and 7.5-3.75 cm, 4-8 GHz; Ku-band – 2,5-1,67 cm, 12-18 GHz; Ka-the range of 1.13-0.75 cm, 26,5-40 GHz.