Interesting concept. But I'd guess they will have serious yield problems. Also it is probably pretty difficult to extend this operation principle to flexible paper like display, as it relies on a mechanical effect.
It's fairly simple to adjust the amount of light reflected to make greyscale images. What is very hard to do is adjust the wavelengths reflected to make different colors.
The operating principle of a full color e-ink display is the same as that of a grey scale one. You just need colored pigments/filters. Its merely an issue of manufacturing, but not a limit of the technology. This is being worked on...
The original poster was referring to silicon LEDs. These can be produced on a silicon wafer using bulk processes - eg. you can create an arbitrary amount of LEDs.
ST is in the process of commercialising their silicon-LEDs. Their first product will be a fully integrated opto-coupler.
Unfortunately other applications are still way off. A display would require far more brightness than the device is able to deliver, not to speak of different colors. Their efficiency is low, so heat may be a question. Also yield and uniformity is something to take account for.
Another thing that cannot be ignored is that these devices can not be stacked with transistors. This means that the driving transistors would have to be between the pixels, adding ugly bad space. (bad fill factor)
My bet is rather on organic LEDs for microdisplay.
These guys keep ignoring some questions about the reliability of devices. Its a long way from academic reserach to industrial research to a practical process to a product.
LM is not a semiconductor giant.. they probably just supporting it because it is less sensitive to ionising radiation. A characteristic it shares with MRAM.
Intel has already demonstrated a cycle life of 10^12 and expects data retention of 10 years at 120dC
1e12 is MUCH too little for practial matters (you mentioned dram replacement), competing technologies are magnitudes ahead. And afaik the tested structures were low density in 0.25 or 0.35 um technology. Smaller spacing will increase the temperature gradient and thus decrease reliability. Never trust these announcements.
They are currently running OUM alongside the Pentium 3.0Ghz line (not commercially viable / internal R&D use only).
I doubt any waver that has deposited chalcogenide on it does ever see the production line of a Pentium4 FAB.
Ovonyx non-volatile memory technology offers significantly faster write and erase speeds and higher cycling endurance than conventional Flash memory. It also has the advantage of a simple fabrication process, which allows the design of semiconductor chips with embedded non-volatile memory using only a few additional mask steps.
True, but competing technologies claim exactly the same.
I would not be here if I was not a whore for the technology. I fully expect to do one of two things with my investment:
Well, you are a "fan". This is the worst thing that may happen when investing - it means you wont get out in time.
Even if OUM becomes the NVMEM technology of choice, intel and others will not be stupid enough to let ovonyx grow too much.. Start ups with good technologies are bought up by giants. This is not a new market.
Still, it is made out of a cheap'n'cheerful plastic. You could have a "sugar-cube" like holder for the device, and just pop a new one in when it has filled up.
Unfortunately, "plastic" aka polymers does not equal to "cheap" in this case. This is a common misconception. These substances have to extremely pure, which makes them expensive.
In addition - In my opinion the whole story is a publicity stunt, I fail to see the real advantages.
-There are many other fuse materials avialable, I do not see the advantage of using PEDOT -You still need a silicon backplane. -High memory density requires high resolution photolithography which does also contribute significantly to the cost. (There may be ways around though..)
Well, you should not forget that there are many many competing technologies for a flash successor. Thinks to consider when discussing OUM:
-Intel (and ST to a minor factor) are the only companies investing into this, all competitors have different technologies. -A demonstration of reliable high density operation has still to follow. Handling local temperature differences of several hundrend kelvins within a sold state device poses many problems. Thermal stress will contribute to wear out, characteristics drift etc etc.. -How about process compatibility.. no mention about this, yet -Intel is also investing in competing technologies
Some companies (for example Infineon) are also investigating sub 45nm flash-alike memory utilizing FinFets.
None of the new technologies for nonvolatile memories come even close to the density of advanced flash concepts like NROM, Mirror Bit flash, multilevel flash.
Hmm... I'm having trouble visualising 0.57 microns square. Lets see - even with these reduced cell sizes, you'd need 3600 square meters (half the size of a football pitch) of SRAM to have one bit per person on the world.
Yes, I know its the fault of the metric system, everything would have been easier with mils, Angstrom and squarefeet.
But the correct result is 0.0036 m^2. Does a Gigabyte of Dram (=8 Billion Bits), which is obtainable in todays technology, take up a football pitch? no!
Well, the difference between the 6502 and the Z80 is that the 6502 is a very efficient design, that already uses a bit of pipelining. The Z80 is based on a very conservate and simple state machine and takes roughly 4 cycles to do what the 6502 does in ones.
It was actually possible to know everything there is to know about it.
Of course, it is possible to know everything that is stated in the official documentation. Unfortunately this does only reproduce the specifications of the C64 chipset, not the actual implementation.
One amazing thing about the C64 is that it offers and endless amount of ways to abuse the hardware to do even more amazing things. (Does the documentation state anything about 320x200x16 graphic modes,8 bit samples, fullscreen graphics?) Still today people find new coding tricks now and then. (Check out this site, the top ten list)
The C64 CPU is well known, all of its illegal opcodes have been decoded and documented. You can even find the reverse engineered circuit on the web. (click - sorry its hungarian, use your imagination to navigate the site).
Also the C64 video chip, the VIC, is thouroughly documented based on empirical knowledge (here). However, there are still pieces and bits left..
The sounds chip, the SID, has been reverse engineered to a state where fairly accurate emulation is possible. Unfortunately this is only true for the digital part. The analog part of the sid (mainly programmable filters) does hardly follow the specification and is subject to variations between different SID-chip revisions etc. This problem can only be solved by reverse engineering the actual circuit of the filter.
Besides, I do not believe that 41% efficiency was accomplished just with a-Si. Was it a cis solar cell? If this was possible they would already be in production.
Ok, just checked it. As anticipated the Solarex solar cells are multijunction cells, just using a-Si for one of the junctions. a-SiGe IS another matter..
Thin-film technology held a lot of promise, but that's all it was. File it away with global warming and cold fusion under "BS".
Good thing I already bought Thin-Film-Transistor Display, before people found out that the technology does not work at all.
Besides, I do not believe that 41% efficiency was accomplished just with a-Si. Was it a cis solar cell? If this was possible they would already be in production.
a) Chips are already "stacked". Layer over layer of silicon
False, there is just one active layer of single crystalline silicon that contains the devices. The remaining layers are interconnects.
b) If you are talking about stacking dice (That is, literally stacking chips inside the package) then the distance the information would have to travel when going trough the "vertical interconnects" would be thousands or tens of thousands bigger than the distance of any on-chip interconnection.
How, why? the lateral extend of any die is usually bigger than its height. In fact the distance would be much shorter. Active layers would be seperated by less than 100micrometers.
The inventor of the motor seems to be a simple mechanican - he has no formal training on theory and engine design. Since he obviously lacks background I am extremeley skeptical about the engine. It may work - but I am sure there are many additional things that have to be taken care of.
etching for silicon is more likely to done with a chemical etch (pirhanna sol'n, HF, etc.) than a physical one.
Boy, thats 70ies stuff. Today, wet processing is avoided as much as it is possible. But you are right in one point, purely physical etching is not used frequently. However there are combined physical/chemical methods. Do a websearch on Reactive Ion Etching, Plasma Etching etc.
The frequent usage of the word "familiar" hints very much at what lycoris became: A cheap Windows XP clone. All the description and advertising is trying to explain that I almost get the functionality of Windows XP. However why dont buy the real thing then ? I want extended functionality and improvements.
Yeah, but I guess it will just end like: "Oh, you are using lycoris/linux. Can't you afford Windows?"
Carbon, on the other hand, isn't so obliging - It doesn't melt, it sublimes directly from a solid state into a gaseous one, so this way's out..
You can grow diamond from the vapor phase. (CVD-Diamond). This does work and is state of the art. There are also some people out there who try to grow diamond from a fluid phase using a precursor/solvent, but results have yet to be shown..
High electrical conductivity and high thermal conductivity tend to run together. For instance copper has an electrical conductivity of 5.8x10^7 S/m and a thermal conductivity of 200 W/mK.
This is known as Wiedeman-Franz Law in Physics. It describes the relationship between eletron heat transfer and conductivity. However it is only valid for Metals. Heat transfer in semiconductors is dominated by lattice vibration transport. Due to the bandgap there is little phonon/electron interaction.
A notable exception is diamond with a low electrical conductivity on the order of 1 S/m and a high thermal conductivity of 700 W/mK.
Its not an exception, its a semiconductor with a large bandgap and behaves exactly as expected.
Diamonds are a very very useful natural resource, but instead of being able to take advantage of that, we're forced to pay huge prices only to have them end up as decorations on some floozy. Well, why ? Industrial Diamonds are cheap. And aritificial Diamonds have matured somewhat. Some russians labs are already producing diamonds of considerable size. There is a also promising research into polycrystalline Diamonds, just serach for "CVD Diamond"
Slashdot Mirror
Interesting concept. But I'd guess they will have serious yield problems. Also it is probably pretty difficult to extend this operation principle to flexible paper like display, as it relies on a mechanical effect.
It's fairly simple to adjust the amount of light reflected to make greyscale images. What is very hard to do is adjust the wavelengths reflected to make different colors.
The operating principle of a full color e-ink display is the same as that of a grey scale one. You just need colored pigments/filters. Its merely an issue of manufacturing, but not a limit of the technology. This is being worked on...
The original poster was referring to silicon LEDs. These can be produced on a silicon wafer using bulk processes - eg. you can create an arbitrary amount of LEDs.
ST is in the process of commercialising their silicon-LEDs. Their first product will be a fully integrated opto-coupler.
Unfortunately other applications are still way off. A display would require far more brightness than the device is able to deliver, not to speak of different colors. Their efficiency is low, so heat may be a question. Also yield and uniformity is something to take account for.
Another thing that cannot be ignored is that these devices can not be stacked with transistors. This means that the driving transistors would have to be between the pixels, adding ugly bad space. (bad fill factor)
My bet is rather on organic LEDs for microdisplay.
These guys keep ignoring some questions about the reliability of devices. Its a long way from academic reserach to industrial research to a practical process to a product.
Not true - Lockheed Martin is also a developer
LM is not a semiconductor giant.. they probably just supporting it because it is less sensitive to ionising radiation. A characteristic it shares with MRAM.
Intel has already demonstrated a cycle life of 10^12 and expects data retention of 10 years at 120dC
1e12 is MUCH too little for practial matters (you mentioned dram replacement), competing technologies are magnitudes ahead. And afaik the tested structures were low density in 0.25 or 0.35 um technology. Smaller spacing will increase the temperature gradient and thus decrease reliability. Never trust these announcements.
They are currently running OUM alongside the Pentium 3.0Ghz line (not commercially viable / internal R&D use only).
I doubt any waver that has deposited chalcogenide on it does ever see the production line of a Pentium4 FAB.
Ovonyx non-volatile memory technology offers significantly faster write and erase speeds and higher cycling endurance than conventional Flash memory. It also has the advantage of a simple fabrication process, which allows the design of semiconductor chips with embedded non-volatile memory using only a few additional mask steps.
True, but competing technologies claim exactly the same.
I would not be here if I was not a whore for the technology. I fully expect to do one of two things with my investment:
Well, you are a "fan". This is the worst thing that may happen when investing - it means you wont get out in time.
Even if OUM becomes the NVMEM technology of choice, intel and others will not be stupid enough to let ovonyx grow too much.. Start ups
with good technologies are bought up by giants. This is not a new market.
(And the EU, at that!)
Not yet...
Still, it is made out of a cheap'n'cheerful plastic. You could have a "sugar-cube" like holder for the device, and just pop a new one in when it has filled up.
Unfortunately, "plastic" aka polymers does not equal to "cheap" in this case. This is a common misconception. These substances have to extremely pure, which makes them expensive.
In addition - In my opinion the whole story is a publicity stunt, I fail to see the real advantages.
-There are many other fuse materials avialable, I do not see the advantage of using PEDOT
-You still need a silicon backplane.
-High memory density requires high resolution photolithography which does also contribute significantly to the cost. (There may be ways around though..)
Well, you should not forget that there are many many competing technologies for a flash successor. Thinks to consider when discussing OUM:
-Intel (and ST to a minor factor) are the only companies investing into this, all competitors have different technologies.
-A demonstration of reliable high density operation has still to follow. Handling local temperature differences of several hundrend kelvins within a sold state device poses many problems. Thermal stress will contribute to wear out, characteristics drift etc etc..
-How about process compatibility.. no mention about this, yet
-Intel is also investing in competing technologies
Some companies (for example Infineon) are also investigating sub 45nm flash-alike memory utilizing FinFets.
None of the new technologies for nonvolatile memories come even close to the density of advanced flash concepts like NROM, Mirror Bit flash, multilevel flash.
Hmm... I'm having trouble visualising 0.57 microns square. Lets see - even with these reduced cell sizes, you'd need 3600 square meters (half the size of a football pitch) of SRAM to have one bit per person on the world.
Yes, I know its the fault of the metric system, everything would have been easier with mils, Angstrom and squarefeet.
But the correct result is 0.0036 m^2. Does a Gigabyte of Dram (=8 Billion Bits), which is obtainable in todays technology, take up a football pitch? no!
RISC for the masses... what else can I say?
No, that credit must go to the Acorn which was out way earlier and featured a much RISCier CPU, the ARM.
3. Commodore 64
Cheep! Cheep! Computing for the masses, however, I didn't like the emulators, and there's a shitload of bias here, m'kay?
Not to forget its superior graphics capabilties (Sprites!) and the unparalleled sound at its time.
Well, the difference between the 6502 and the Z80 is that the 6502 is a very efficient design, that already uses a bit of pipelining. The Z80 is based on a very conservate and simple state machine and takes roughly 4 cycles to do what the 6502 does in ones.
Yes, but they are pretty good nowadays.
This is the best emulator for the Plus/4: YAPE
Try this page for software: Plus/4 World
... Just let the madman talk ...
Of course, it is possible to know everything that is stated in the official documentation. Unfortunately this does only reproduce the specifications of the C64 chipset, not the actual implementation.
One amazing thing about the C64 is that it offers and endless amount of ways to abuse the hardware to do even more amazing things. (Does the documentation state anything about 320x200x16 graphic modes,8 bit samples, fullscreen graphics?) Still today people find new coding tricks now and then. (Check out this site, the top ten list)
The C64 CPU is well known, all of its illegal opcodes have been decoded and documented. You can even find the reverse engineered circuit on the web. (click - sorry its hungarian, use your imagination to navigate the site).
Also the C64 video chip, the VIC, is thouroughly documented based on empirical knowledge (here). However, there are still pieces and bits left..
The sounds chip, the SID, has been reverse engineered to a state where fairly accurate emulation is possible. Unfortunately this is only true for the digital part. The analog part of the sid (mainly programmable filters) does hardly follow the specification and is subject to variations between different SID-chip revisions etc. This problem can only be solved by reverse engineering the actual circuit of the filter.
Ok, just checked it. As anticipated the Solarex solar cells are multijunction cells, just using a-Si for one of the junctions. a-SiGe IS another matter..
Good thing I already bought Thin-Film-Transistor Display, before people found out that the technology does not work at all.
Besides, I do not believe that 41% efficiency was accomplished just with a-Si. Was it a cis solar cell? If this was possible they would already be in production.
What was your function at that company? Admin?
False, there is just one active layer of single crystalline silicon that contains the devices. The remaining layers are interconnects.
b) If you are talking about stacking dice (That is, literally stacking chips inside the package) then the distance the information would have to travel when going trough the "vertical interconnects" would be thousands or tens of thousands bigger than the distance of any on-chip interconnection.
How, why? the lateral extend of any die is usually bigger than its height. In fact the distance would be much shorter. Active layers would be seperated by less than 100micrometers.
The inventor of the motor seems to be a simple mechanican - he has no formal training on theory and engine design. Since he obviously lacks background I am extremeley skeptical about the engine. It may work - but I am sure there are many additional things that have to be taken care of.
Boy, thats 70ies stuff. Today, wet processing is avoided as much as it is possible. But you are right in one point, purely physical etching is not used frequently. However there are combined physical/chemical methods. Do a websearch on Reactive Ion Etching, Plasma Etching etc.
The frequent usage of the word "familiar" hints very much at what lycoris became: A cheap Windows XP clone. All the description and advertising is trying to explain that I almost get the functionality of Windows XP. However why dont buy the real thing then ? I want extended functionality and improvements.
Yeah, but I guess it will just end like: "Oh, you are using lycoris/linux. Can't you afford Windows?"
They have information on CVD-Diamond. Also do not forget to check out pictures. Of actual diamond products.
You can grow diamond from the vapor phase. (CVD-Diamond). This does work and is state of the art. There are also some people out there who try to grow diamond from a fluid phase using a precursor/solvent, but results have yet to be shown..
This is known as Wiedeman-Franz Law in Physics. It describes the relationship between eletron heat transfer and conductivity. However it is only valid for Metals. Heat transfer in semiconductors is dominated by lattice vibration transport. Due to the bandgap there is little phonon/electron interaction.
A notable exception is diamond with a low electrical conductivity on the order of 1 S/m and a high thermal conductivity of 700 W/mK.
Its not an exception, its a semiconductor with a large bandgap and behaves exactly as expected.
Diamonds are a very very useful natural resource, but instead of being able to take advantage of that, we're forced to pay huge prices only to have them end up as decorations on some floozy.
Well, why ? Industrial Diamonds are cheap. And aritificial Diamonds have matured somewhat. Some russians labs are already producing diamonds of considerable size. There is a also promising research into polycrystalline Diamonds, just serach for "CVD Diamond"
It all may depend on your resumee and the interviews. I have had a lot of luck (very little rejections).
But anyways, I think an internship is a great opportunity - even if it is unpaid.