Took me three reads to realize it can process 86 different instructions, and not x86 instructions. I kept thinking, that’s weird, why does this super small processor support x86? Lol, I wonder if the engineers did this purposely.
I'm not a RF expert, but 21 mW seems like quite a lot, even for a large antenna.
If anyone has relevant knowledge about harvesting ambient power, what are the best prospects and how much can they realistically provide? Could a thermocouple integrated into a wristband provide enough power from human body heat? How much power to RFID chips usually take?
RFID chips typically run 1 to 4uW (microWatts) so about 10,000 times less power than the plastic ARM (21mW). Harvesting RF energy yields very little power and the antennas are large (10's of mm in size) versus the RFID chip being powered which is typically less than 1mmsq. You *might* get a few mW if the RFID antenna was on top of or within a few cm of the broadcast antenna but at reasonable distances of a few meters a few uW is the best you can hope for. Max ranges are about 10meters on a good day. You also loose a lot in the conversion process of rectifying the RF signal down into a DC voltage for the chip to run off.
I think it's most likely to be integrated into existing products that already use such materials. And to the extent that it can reduce things like food spoilage, it has the potential to actually *reduce* waste.
If you want to reduce plastic waste, invest in alternatives and scaling up of monomer-based recycling technologies.
No, the suggestion of integrating it into plastic food packaging seemed very odd to me, as that's where a vast amount of unrecyclable packaging waste is that needs to be cut out of the system for consumers (despite the pushback from manufacturers). Even if you charitably assume that all those currently unrecyclable materials are changed through enlightened self-interest to widely recyclable ones in any meaningful timescale, adding in highly complex chips made of different plastics bonded with metals and any other chemicals in those TFTs will make that just as unrecyclable in BULK, since this is being discussed as something to be produced at mass scale. Current recycling processing can't cope with mixed materials well, so this application will be an ecological disaster even if it saves a few food tons in some niche sector that can afford to put this sort of processing in the packaging.
> the suggestion of integrating it into plastic food packaging seemed very odd to me
Why do you presume it's for single-serving containers? There could be lots of bulk or industrial uses where you'd like to measure storage or transportation conditions.
> adding in highly complex chips made of different plastics bonded with metals > and any other chemicals in those TFTs will make that just as unrecyclable
It's 7 mm on a side and probably paper-thin. I'd be surprised if there weren't more impurities in the typical recyclable container, already.
"The second factor is cost, with flexible processing at scale being orders of magnitude cheaper than equivalent silicon designs."
Also, when integrating electronics into flexible products, I think a likely source of electro-mechanical failures is at the interface of rigid and soft materials. You can mitigate this by adding additional layers of soft material to create a hardness gradient, but that adds further material, bulk, and manufacturing steps.
Interesting research - however the extremely high power consumption (21mw is huge compared to silicon M0) and very low performance (29kHz vs 50MHz) relegates this to an unusable curiosity.
To be usable an idle power consumption of 1mW or less and a 1MHz speed are minimum requirements. (Silicon M0 can manage with 11.2 uW/MHz - this plastic M0 needs 724mW/MHz - over 60,000 times less efficient.)
it's still 12x better than the previous leading design, and still much cheaper. How much compute do you think you need for these very far edge applications?
Perhaps Duncan is thinking of running a full network stack, but there's no way you can even think about that with so little RAM and ROM.
What I go back to is all the electronic appliances in decades past. They didn't have much more compute than this, and yet it was fine for simple automation tasks.
The way to think about this is perhaps some sensor control you can embed in smart clothes, that acts as glue for something like a smartwatch.
These things are typically going to be used in conjunction with more powerful processors. Maybe they won't stay connected full-time, but there are usually going to be more conventional processors in the picture, somewhere in the operational cycle.
Depends a lot on what you want to use it for. There are plenty of applications for which the speed is adequate. The bigger issues are probably power & its minuscule memory (RAM and ROM).
It's already good enough to control a sensor and an antenna.
Portable computers were also an unusable curiosity at first. I'd argue Intel's Atom laptops still are. Regardless, this will be iterated upon, and improved, until it's everywhere. I'm just not sure that's a good thing.
Like a herd of lemmings, the march of technology pushes us inexorably onward. The best we can do is try to steer it in ways that mitigate the downsides.
I get that this is a momentous achievement and just the first, full step into the realm of plastic microcontrollers, but I can't help but be disappointed by the amount of "RAM" (which is really just its register file) and ROM. And looking at the amount of area they occupy gives me little hope this will be significantly expanded, without significant further technology improvements.
I'm actually more disappointed by the lack of RAM/ROM than the low clockspeed.
Also, I wonder how much of its power budget is consumed by refreshing the 128 bytes of DRAM.
Was the main reason not to use SRAM simply due to density? I know that's the usual argument, but this is such an exotic technology, there could be other reasons.
That looks like a mistake to me. Tristate latch RAM is conventionally static - at least all the ones I use in chip design are. That's a common gate type for a regfile but there are also more efficient/higher performance types too. You also don't want your reg file as dynamic ram because reads are usually destructive (you have to read & write back to keep the value) so I think that's a mistake in the paper or the article.
Would it be correct to say that it is more about « non silicon transistors » (because my understanding is that it is IGZO Thin Film Transistors (TFT)) deposited on a « flexible plastic substrate » (my understanding is that the substrate is polyimide)
So it seems more « non silicon transistors on a flexible plastic substrate » than genuine plastic electronic using organic transistors (like the technology developped by the company Plastic Logic) where the transistors themselves are plastic.
Also I recall to have seen some Youtube video where the PragmatIC CEO was clearly stating that they have a pragmatic approach by combining different materials to achieve their goal of creating « ultra low cost flexible electronics »
The Atari 2600 had 128 bytes of RAM and a much higher-clock but also simpler CPU (fewer transistors).
The Fairchild Video Entertainment System had 64 bytes of scratchpad RAM but also a faster clock than this, despite the fact that the CPU wasn’t on a single die.
The Fairchild, to my knowledge, didn’t have a graphics accelerator chip — relying instead on the CPU to work with its whopping 4 colors and low resolution.
The Atari, which was hardware-upgraded after the release of the VES (in order to be significantly better) had a sound chip and some sort of odd graphics chip that drew everything line by line from a point as I hazily recall. It enabled a large color palette and rainbow effects but also had big drawbacks.
So, would this be fast enough to run a Fairchild VES game, one that didn’t come with on-cart RAM (as Maze and Chess did)? Or, is the 29 KHz speed just too slow, even with a much more powerful CPU design? I assume it can handle VES games but wonder more about 2600 games, let’s say if this chip were given the TIA sound chip and whatever the GPU was.
(Interestingly, the Intellivision is odd because it shipping with a 16-bit CPU — internally, the bus was cut to 8 bits — and ran in the KHz, not MHz, range. I think the RAM was also very slow due to how it was implemented but it’s hard to remember the details.)
The Maze game shipped with 1K of SRAM and Chess had 6K, which was massive at the time.
Fairchild would have been better off to have given the base unit more RAM rather than ‘waste’ chips in Maze cartridges. That 1K was needed for the AI of the ‘cat’ in the ‘Cat and Mouse’ version of the Maze game.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
34 Comments
Back to Article
LeftSide - Thursday, July 22, 2021 - link
Took me three reads to realize it can process 86 different instructions, and not x86 instructions.I kept thinking, that’s weird, why does this super small processor support x86? Lol, I wonder if the engineers did this purposely.
mode_13h - Thursday, July 22, 2021 - link
> I wonder if the engineers did this purposely.No. This thing is far too resource-constrained for anything like that to be out of cuteness.
Wereweeb - Thursday, July 22, 2021 - link
I wonder if they could power this off an antenna that captures the energy from broadband or radio frequencies.mode_13h - Thursday, July 22, 2021 - link
I'm not a RF expert, but 21 mW seems like quite a lot, even for a large antenna.If anyone has relevant knowledge about harvesting ambient power, what are the best prospects and how much can they realistically provide? Could a thermocouple integrated into a wristband provide enough power from human body heat? How much power to RFID chips usually take?
TimSyd - Thursday, July 22, 2021 - link
RFID chips typically run 1 to 4uW (microWatts) so about 10,000 times less power than the plastic ARM (21mW).Harvesting RF energy yields very little power and the antennas are large (10's of mm in size) versus the RFID chip being powered which is typically less than 1mmsq.
You *might* get a few mW if the RFID antenna was on top of or within a few cm of the broadcast antenna but at reasonable distances of a few meters a few uW is the best you can hope for. Max ranges are about 10meters on a good day.
You also loose a lot in the conversion process of rectifying the RF signal down into a DC voltage for the chip to run off.
Wereweeb - Friday, July 23, 2021 - link
Thank you.CaptainChaos - Thursday, July 22, 2021 - link
Great! Just what the world needs: More plastic to foul the ecosystem!mode_13h - Thursday, July 22, 2021 - link
I think it's most likely to be integrated into existing products that already use such materials. And to the extent that it can reduce things like food spoilage, it has the potential to actually *reduce* waste.If you want to reduce plastic waste, invest in alternatives and scaling up of monomer-based recycling technologies.
asmian - Thursday, July 22, 2021 - link
No, the suggestion of integrating it into plastic food packaging seemed very odd to me, as that's where a vast amount of unrecyclable packaging waste is that needs to be cut out of the system for consumers (despite the pushback from manufacturers). Even if you charitably assume that all those currently unrecyclable materials are changed through enlightened self-interest to widely recyclable ones in any meaningful timescale, adding in highly complex chips made of different plastics bonded with metals and any other chemicals in those TFTs will make that just as unrecyclable in BULK, since this is being discussed as something to be produced at mass scale. Current recycling processing can't cope with mixed materials well, so this application will be an ecological disaster even if it saves a few food tons in some niche sector that can afford to put this sort of processing in the packaging.mode_13h - Friday, July 23, 2021 - link
> the suggestion of integrating it into plastic food packaging seemed very odd to meWhy do you presume it's for single-serving containers? There could be lots of bulk or industrial uses where you'd like to measure storage or transportation conditions.
> adding in highly complex chips made of different plastics bonded with metals
> and any other chemicals in those TFTs will make that just as unrecyclable
It's 7 mm on a side and probably paper-thin. I'd be surprised if there weren't more impurities in the typical recyclable container, already.
Wereweeb - Friday, July 23, 2021 - link
Bio-polyimide is being developed and is biodegradable. Literally just google "Bio polyimides"drexnx - Thursday, July 22, 2021 - link
this is so cool for wearables and smart clothes, did they give a temperature spec?drexnx - Thursday, July 22, 2021 - link
combined with flexible TEGs this could be perfect for wearable body heat powered sensors or devicesskavi - Thursday, July 22, 2021 - link
isn’t the standard silicon M0 tiny enough that it doesn’t really need to be flexible?mode_13h - Thursday, July 22, 2021 - link
The article says:"The second factor is cost, with flexible processing at scale being orders of magnitude cheaper than equivalent silicon designs."
Also, when integrating electronics into flexible products, I think a likely source of electro-mechanical failures is at the interface of rigid and soft materials. You can mitigate this by adding additional layers of soft material to create a hardness gradient, but that adds further material, bulk, and manufacturing steps.
Duncan Macdonald - Thursday, July 22, 2021 - link
Interesting research - however the extremely high power consumption (21mw is huge compared to silicon M0) and very low performance (29kHz vs 50MHz) relegates this to an unusable curiosity.To be usable an idle power consumption of 1mW or less and a 1MHz speed are minimum requirements. (Silicon M0 can manage with 11.2 uW/MHz - this plastic M0 needs 724mW/MHz - over 60,000 times less efficient.)
drexnx - Thursday, July 22, 2021 - link
it's still 12x better than the previous leading design, and still much cheaper. How much compute do you think you need for these very far edge applications?mode_13h - Thursday, July 22, 2021 - link
Perhaps Duncan is thinking of running a full network stack, but there's no way you can even think about that with so little RAM and ROM.What I go back to is all the electronic appliances in decades past. They didn't have much more compute than this, and yet it was fine for simple automation tasks.
mode_13h - Thursday, July 22, 2021 - link
The way to think about this is perhaps some sensor control you can embed in smart clothes, that acts as glue for something like a smartwatch.These things are typically going to be used in conjunction with more powerful processors. Maybe they won't stay connected full-time, but there are usually going to be more conventional processors in the picture, somewhere in the operational cycle.
mode_13h - Thursday, July 22, 2021 - link
Depends a lot on what you want to use it for. There are plenty of applications for which the speed is adequate. The bigger issues are probably power & its minuscule memory (RAM and ROM).Wereweeb - Friday, July 23, 2021 - link
It's already good enough to control a sensor and an antenna.Portable computers were also an unusable curiosity at first. I'd argue Intel's Atom laptops still are. Regardless, this will be iterated upon, and improved, until it's everywhere. I'm just not sure that's a good thing.
Spunjji - Monday, July 26, 2021 - link
Seconding that entire sentiment!mode_13h - Monday, July 26, 2021 - link
> I'm just not sure that's a good thing.Like a herd of lemmings, the march of technology pushes us inexorably onward. The best we can do is try to steer it in ways that mitigate the downsides.
mode_13h - Thursday, July 22, 2021 - link
I get that this is a momentous achievement and just the first, full step into the realm of plastic microcontrollers, but I can't help but be disappointed by the amount of "RAM" (which is really just its register file) and ROM. And looking at the amount of area they occupy gives me little hope this will be significantly expanded, without significant further technology improvements.I'm actually more disappointed by the lack of RAM/ROM than the low clockspeed.
mode_13h - Thursday, July 22, 2021 - link
Also, I wonder how much of its power budget is consumed by refreshing the 128 bytes of DRAM.Was the main reason not to use SRAM simply due to density? I know that's the usual argument, but this is such an exotic technology, there could be other reasons.
TimSyd - Thursday, July 22, 2021 - link
That looks like a mistake to me. Tristate latch RAM is conventionally static - at least all the ones I use in chip design are. That's a common gate type for a regfile but there are also more efficient/higher performance types too.You also don't want your reg file as dynamic ram because reads are usually destructive (you have to read & write back to keep the value) so I think that's a mistake in the paper or the article.
mode_13h - Thursday, July 22, 2021 - link
Very informative. Thanks for this & the other reply!ghostbit - Thursday, July 22, 2021 - link
I wonder what the "mopfet" cross-section looks like. They can dope plastics with p and n?Diogene7 - Friday, July 23, 2021 - link
Not sure to fully understand.Would it be correct to say that it is more about « non silicon transistors » (because my understanding is that it is IGZO Thin Film Transistors (TFT)) deposited on a « flexible plastic substrate » (my understanding is that the substrate is polyimide)
So it seems more « non silicon transistors on a flexible plastic substrate » than genuine plastic electronic using organic transistors (like the technology developped by the company Plastic Logic) where the transistors themselves are plastic.
Also I recall to have seen some Youtube video where the PragmatIC CEO was clearly stating that they have a pragmatic approach by combining different materials to achieve their goal of creating « ultra low cost flexible electronics »
Oxford Guy - Sunday, July 25, 2021 - link
This wasn’t what I had in mind when I asked for consumer-level wafer-scale.Oxford Guy - Sunday, July 25, 2021 - link
The Atari 2600 had 128 bytes of RAM and a much higher-clock but also simpler CPU (fewer transistors).The Fairchild Video Entertainment System had 64 bytes of scratchpad RAM but also a faster clock than this, despite the fact that the CPU wasn’t on a single die.
The Fairchild, to my knowledge, didn’t have a graphics accelerator chip — relying instead on the CPU to work with its whopping 4 colors and low resolution.
The Atari, which was hardware-upgraded after the release of the VES (in order to be significantly better) had a sound chip and some sort of odd graphics chip that drew everything line by line from a point as I hazily recall. It enabled a large color palette and rainbow effects but also had big drawbacks.
So, would this be fast enough to run a Fairchild VES game, one that didn’t come with on-cart RAM (as Maze and Chess did)? Or, is the 29 KHz speed just too slow, even with a much more powerful CPU design? I assume it can handle VES games but wonder more about 2600 games, let’s say if this chip were given the TIA sound chip and whatever the GPU was.
(Interestingly, the Intellivision is odd because it shipping with a 16-bit CPU — internally, the bus was cut to 8 bits — and ran in the KHz, not MHz, range. I think the RAM was also very slow due to how it was implemented but it’s hard to remember the details.)
mode_13h - Sunday, July 25, 2021 - link
Thanks for the specs. I didn't know about anything quite that far back.Oxford Guy - Monday, July 26, 2021 - link
The Maze game shipped with 1K of SRAM and Chess had 6K, which was massive at the time.Fairchild would have been better off to have given the base unit more RAM rather than ‘waste’ chips in Maze cartridges. That 1K was needed for the AI of the ‘cat’ in the ‘Cat and Mouse’ version of the Maze game.
Oxford Guy - Monday, July 26, 2021 - link
It’s hilarious to think that this CPU has less RAM (much much less) than the Maze cartridge for the 1976 Fairchild console.