I worked at a fab (photolithography specifically, shining angry lights into rocks to make them smart) and knowing what it takes to make chips it blows my mind that this tech is accessible to individuals, thats cool
If anyone happens to see this and can chime in. How does one who is interested in chip design and manufacturing get into the industry? I would be most interested in designing chip architecture (but anything higher or lower than that would be awesome as well), but working in a fab would be insane as well
I am a chip designer and I would say just go pursue EE or CE majors in university and the path will be there. Today chip companies are hiring like crazy and in a couple of years this hiring will continue to grow.@@randomsomeguy156
I worked in the PCB industry 30 years ago and was amazed when it became accessible to the general public, and now it the same with ASICs. MIND BLOWN!!!
When you immediately recognize a 'movement' on 32x32 pixels @ 3 FPS and 4 colors, in 6-8 frames total length, then that movement is really etched into mankind's brain.
I had suspected it was a Rick roll as soon as he said all of the pieces had a description except for that very one. Because people like to have a Rick roll pop up in unexpected places, in unexpected ways …
It’s not really a youtuber being capable of making their own chips, rather it is an open source project being able to produce a chip with lots of peoples designs on it, still at a rather significant cost with no real purpose other than education. The fab has probably cut them a really good deal. It’s not like you imply where a youtuber can just get any custom chip made, it is still not accessible except in this specific open source project.
Most TH-camrs that do tech and science related stuff are pretty gifted individuals, a lot of them work in the industry, a lot of them are self-taught professionals or both.
Doing any actually useful thing just starts in value and satisfaction where frivolous things top out. But obviously this is not a frivolous thing, as it's just a proof of concept and testing of the overall project whil will have plenty of use.
@@Beeti1 This way of thinking is called 'decadence'. A loss of reality caused by not being in touch with actual real needs, laying ground for an completely corrupted set of values. Dacadence leads to whole civilizations declining and dying.
@@conorstewart2214 Which is usually okay as a tool for education. It's a bit like these tiny satellite programs - CubeSats. Not much you can do in a 10x10x10cm satellite. But at least it gets sent to space. Last year they launched 400 satellites, and due to advances in designing micro-satellites, there are now more commercial and amateur projects on board.
@@conorstewart2214Computers got thier start because everyone was sharing thier designs. All the universities, and radio hobbists shared every detail. Thats how we get the idea of file formats and standards that allow the internet itself to work at all. Without that open share model computers would all have thier own file formats, some would use binary other trinary some might even use digitized analog adders but none would be compatable with each other.
I’m currently taking an ASIC class in grad school and I honestly will prolly still take this course. It’s amazing how education at university never seems to match up to classes taught by people with such passion
It's so crazy that we can get custom silicon these days(granted, with a 9 month lead time). Wouldn't be surprised if this gets better and better and in twenty years, we'll have places like PCBway giving you custom chips to go with your custom boards.
This is a great project from many points of view. Just Great. !!! I'm Senior electronic Design engineer myself - and love your inspiring projects , openness with misshapes and the story telling. Just, Thanks.
No way ... WAY! FABulous! What a smart and fun project! Brilliant and very inspiring! THANK YOU for keeping to create such fun stuff! Keep right on creating!
@@ArneChristianRosenfeldtWouldn't the 0ohm resisters act, if a bit flawed, like a fuse? Granted, the tolerance for it would have to be lower than what you're powering with it...
@@rya3190 I was more thinking if on boot up you could set the power limits for each tile ( most of the to zero ). Why else would they care if a single tile can pull down the whole chip? Isn’t there really a power management?
Count the pins! I wonder how the surge protection works. Maybe we want to do intense compute and need a lot of current in one tile, while the other is just sram?
@@GodzillaGoesGaga the fab process is open source and all projects so far seem to be typical homework assignments. I don’t even know what the cutting edge is right now, but it seems to deal with scaling to many transistors. Far more than you can buy here.
I really wanted to do tiny tapeout after hearing about it a bit ago, but I've been a bit too busy. Hopefully it'll still be going when I can get to writing something in verilog.
One step closer to open source hardware and I LOVE YOU FOR IT! One step closer to open source touchscreens, e-ink displays, and maybe eventually even cpus/gpus! 😃
hoping someday we can manufacture ASIC's from home. would be incredible to have even the more basic of minimum-required tooling to make something without relying on a few major companies that are politically involved. even if not at home - like some others have said - PCBWay is a great example of advances in electronics/embedded engineering for the masses at low expense!
Just came across a video on Twitter of "Touchable Flames". The "flames" are made from water vapor and red led lights. It looks like flames, even while putting your hands through the misting.
This is really really really cool, I am just sad you cant order quantities. $300 for a custom (really small) chip is awesome, just wish I could order 1000 more for $1 each.
Maybe through technical advances, economies of scale and multi-project wafer services, one day silicon production can get similar turn-around like what happened in the PCB industry.
It would take a lot of changes. Making stencils for all the layers is really expensive. You can’t “print” them out like with PCB where tracks are very wide (certainly compared to features in a chip). The tools used to dope silicon and deposit SiO2 are also not cheap. It would definitely be cool to get to that point. But it’s pretty far away for now
Nice I was just about to get a small design made at Aisler, thanks for the five bucks. Tinytapeout is such a cool initiative too, if I ever have an idea for an ASIC now I can actually consider getting it made, if I can remember any of the VHDL from uni >.>
I'm more interested in SIP and modules. You have dived much deeper. Actually, it would be enough if I gathered the ready-made products under one roof. (Especially in SIP form and Esp32 s3 and sx1262 chips)
6:42 - I might have chosen a 91K with a 100K for the divider to get the 1.8V from the 3.3V, but maybe that would have been an extra component value that you didn’t care to have to “handle”
Kiloohms?? That would be way off, if the chip draws any current on the 1.8V net. Even the 100 Ohm voltage divider is far from ideal, as the current draw on a logic core supply is not static. So you modulate a voltage ripple on the core supply voltage thanks to the resistors. Depending on the chip, not meeting the core voltage specs and not using a stable supply voltage can give you all kinds of errors and issues that are super hard to track down. Worst case, it could even kill the core, and for a chip that you can't just order replacements, i'd think twice about the risk of destruction because of lazy circuit designs.
As this behaves like a rom, wouldn't it had been interesting to create a "rom", which returns a different image every time you read it? Then you could still animate something (by just reading/displaying the picture in a loop from the ESP32), but really make use of the ASIC
I have no clue at all except I am a bit aware of asic developement for FPV flying or goggles and the hd stream in a project that did not take off cause the investment back then was finally far too big - a million € figure ,not even a 2 or 3 but rather a 5 million figure to get that done. But funding did not happen. But that was rather a companion role and no technical one. But I like this video really a lot and all the struggles with the iterations.
Nice one, bitluni! Great to see, and thanks for sharing this level of detail :) I hope you do more Tiny Tapeout submissions in future! Maybe time to try an analog layout...? :)
Usually ~50'000$+, but this company is starting at just 10k$ and the guys are splitting the cost between 160 people and allocating them even smaller part of the chip so it's just 300$ per design.
@@ArneChristianRosenfeldtI can't possibly imagine that they didn't produce at LEAST double the amount they needed - the expensive part is creating the masks and setting everything up - running the scanners 1,2 or 10 times costs basically the same.
Did you use some kind of special fluid to safely remove soldered parts? If so, I have never heard of such a product and would be VERY interested to know what it is!
He just uses flux paste that helps the solder to flow better. Then you have to heat the solder to melting temperature again to remove the chip. I guess you could try to find some very specific acid that only attacks the tin in the solder, but i don't think this exists and it would just eat away all the metals, including the legs of the chip or other components.
For TT02 the analog part (filters) would have not been possible. I've heard they want to include analog capabilities in the future revisions though! Someone is surely going to short the SID hoarders ;)
@@eitantal726 The SID schematics have been reproduced from the die shots. I can see that the filter uses opamps and FETs with gates driven by DACs to control cutoff. This is unfortunately a no-no for direct implementation, but if one were instead to implement a right biquad filter and a lookup table for its parameters (to perform the filtering like DSPs do) I guess it could work. Sadly the digital filter part would probably exceed the other (oscillator + ADSR) in complexity.
The best choice of use for a custom ASIC XD I mean, you could have made a VGA controler, or anything, and no, it's a Rick roll (without sound), glorious!
Um... 9 months wait for the ASIC? Wow. You really need to plan ahead. Can't be in a rush to get the part. BTW, you probably realized by now that 3.3V divided by two is 1.65V and not 1.8V.
It sounds like having an FPGA manufactured with factory programming. An ASIC that would be interesting would be like the T5L (used in DWIN display), but it would be more interesting if the IC supports being updated via OTA instead of SD card or JTAG.
is tinyTapeout only RTL, or can you do custom layout? also what tools are used for layout, i use cadence for work but that is anything but free and open source lol
I worked at a fab (photolithography specifically, shining angry lights into rocks to make them smart) and knowing what it takes to make chips it blows my mind that this tech is accessible to individuals, thats cool
"Shining angry lights into rocks to make them smart" needs to be on a t-shirt 🤣
@@mllarson We then trap lightning in the rocks to make them think.
I worked for AMAT. Someone said, "8000 people looking for a new job."
If anyone happens to see this and can chime in. How does one who is interested in chip design and manufacturing get into the industry? I would be most interested in designing chip architecture (but anything higher or lower than that would be awesome as well), but working in a fab would be insane as well
I am a chip designer and I would say just go pursue EE or CE majors in university and the path will be there. Today chip companies are hiring like crazy and in a couple of years this hiring will continue to grow.@@randomsomeguy156
I worked in the PCB industry 30 years ago and was amazed when it became accessible to the general public, and now it the same with ASICs. MIND BLOWN!!!
What changed?
Gotta love that the QC sticker reads "Accetpable" at 6:35!
Good spot! It's one of our standard mispelt qc stickers!
Someone needs to add QC to the QC stickers.
QCeption
well yeah, you have to control the Quilaty!
When you immediately recognize a 'movement' on 32x32 pixels @ 3 FPS and 4 colors, in 6-8 frames total length, then that movement is really etched into mankind's brain.
Lmao! The moment I saw it I went "fuck you!"🤣 it's crazy that we can immediately recognize what it is haha
@@BirkinIdkit truly is!
That’s crazy but how
I an instantly going to guess that it’s either a rickroll or bad apple
I had suspected it was a Rick roll as soon as he said all of the pieces had a description except for that very one. Because people like to have a Rick roll pop up in unexpected places, in unexpected ways …
The fact that photolithography is now accessible enough that a youtuber was capable of making their own chips... blows my mind!
It’s not really a youtuber being capable of making their own chips, rather it is an open source project being able to produce a chip with lots of peoples designs on it, still at a rather significant cost with no real purpose other than education. The fab has probably cut them a really good deal.
It’s not like you imply where a youtuber can just get any custom chip made, it is still not accessible except in this specific open source project.
I think you underestimate yotubers power.
Such times, much progression, right..
Most TH-camrs that do tech and science related stuff are pretty gifted individuals, a lot of them work in the industry, a lot of them are self-taught professionals or both.
You can do it in your garage using older fab like nmos and pmos with micron feature sizes
Like photo etching a pcb
It would be much more mindblowing to do something else, useful, than a tiny ROM chip.
There's no better purpose in life than putting so much effort into something so frivolous.
What’s life without whimsy?
Doing any actually useful thing just starts in value and satisfaction where frivolous things top out. But obviously this is not a frivolous thing, as it's just a proof of concept and testing of the overall project whil will have plenty of use.
Actually is massive Karma
@@elmariachi5133 - No, it's frivolous.
@@Beeti1 This way of thinking is called 'decadence'. A loss of reality caused by not being in touch with actual real needs, laying ground for an completely corrupted set of values. Dacadence leads to whole civilizations declining and dying.
That’s such a smart idea! Doing a group buy for silicon is insanely clever! 🤯
The only problem is the small size of the design and that everyone gets access to your design.
@@conorstewart2214 Which is usually okay as a tool for education. It's a bit like these tiny satellite programs - CubeSats. Not much you can do in a 10x10x10cm satellite. But at least it gets sent to space. Last year they launched 400 satellites, and due to advances in designing micro-satellites, there are now more commercial and amateur projects on board.
No it's just a niche at this point
@@conorstewart2214Computers got thier start because everyone was sharing thier designs. All the universities, and radio hobbists shared every detail.
Thats how we get the idea of file formats and standards that allow the internet itself to work at all.
Without that open share model computers would all have thier own file formats, some would use binary other trinary some might even use digitized analog adders but none would be compatable with each other.
@@conorstewart2214 that's not a problem, it's a feature
Rideshare for ASIC, this is fascinating and hilarious and wholesome all at the same time, thanks so much for broadening our horizons.
I’m currently taking an ASIC class in grad school and I honestly will prolly still take this course. It’s amazing how education at university never seems to match up to classes taught by people with such passion
We hope tiny tapeout gets used by lots of universities!
It's so crazy that we can get custom silicon these days(granted, with a 9 month lead time). Wouldn't be surprised if this gets better and better and in twenty years, we'll have places like PCBway giving you custom chips to go with your custom boards.
This is a great project from many points of view.
Just Great. !!!
I'm Senior electronic Design engineer myself - and love your inspiring projects , openness with misshapes and the story telling.
Just, Thanks.
Rick can be proud to the first publication on the new ASIC media.
No way ... WAY! FABulous! What a smart and fun project! Brilliant and very inspiring! THANK YOU for keeping to create such fun stuff! Keep right on creating!
Always check for shorts with an ohm meter before powering on a board. I'm sure you've hear that a million times by now.
A fuse would even protect if you drop a metal wire .
@@ArneChristianRosenfeldtWouldn't the 0ohm resisters act, if a bit flawed, like a fuse? Granted, the tolerance for it would have to be lower than what you're powering with it...
@@rya3190 I was more thinking if on boot up you could set the power limits for each tile ( most of the to zero ). Why else would they care if a single tile can pull down the whole chip? Isn’t there really a power management?
This is how advertising should be done. In the context. With interested audience. By example.
Hmmm sounds like an idea, but how about another decade of focusing on gender, race politics, etc? In every ad? No?
Man, I would love to be able to make SNES and NES ASIC chips for a custom clone console.
Count the pins! I wonder how the surge protection works. Maybe we want to do intense compute and need a lot of current in one tile, while the other is just sram?
A similar project for a handful of consoles and retro pcs we have working on fpgas would be a novel idea.
Can't you simulate those old things on a 328?
@@Ironclad17 you mean to re-use the design as written in VDHL ? Mister Cores get updates all the time. So better stick with pre 1990
It’s not a waste of resources.. you are building up your skills and knowledge. Fantastic job.. so exciting..
MOSIS has been doing this since the 1980s which favored US student engineers taking VLSI design courses, I think Europe had a similar project too.
But isn’t it great that now you don’t have to sign an NDA?
@@ArneChristianRosenfeldtand you can use free and accessible open source tools!
@@ArneChristianRosenfeldt You have to open source your project though. Not particularly good for development of new technology.
@@GodzillaGoesGaga the fab process is open source and all projects so far seem to be typical homework assignments. I don’t even know what the cutting edge is right now, but it seems to deal with scaling to many transistors. Far more than you can buy here.
I really wanted to do tiny tapeout after hearing about it a bit ago, but I've been a bit too busy. Hopefully it'll still be going when I can get to writing something in verilog.
Look at the “Tiny Tapeout Chips” section in the sidebar. They’ve got chips planned every few months at least until the end of 2024.
This is a COOL idea! And open source... WOW!
One step closer to open source hardware and I LOVE YOU FOR IT! One step closer to open source touchscreens, e-ink displays, and maybe eventually even cpus/gpus! 😃
0 to asic and tiny tapeout is amazing. Just think of what's to come in the future. 😁
Never had leading edge tech been put to a better use. Well done, Sir.
Well, that's not really the leading edge technology 😂
What a huge accomplishment! Well done!
hoping someday we can manufacture ASIC's from home.
would be incredible to have even the more basic of minimum-required tooling to make something without relying on a few major companies that are politically involved.
even if not at home - like some others have said - PCBWay is a great example of advances in electronics/embedded engineering for the masses at low expense!
A wild communist furry?
Amazing that this is so accessible.
Just came across a video on Twitter of "Touchable Flames". The "flames" are made from water vapor and red led lights. It looks like flames, even while putting your hands through the misting.
Fun idea and very informative! whimsy for the win.
You could've... But you... TinyTapeout really is amazing. Love it!
wow, Großartig, Glückwunsch, da geht mir's Herz auf als Techniker, bin begeistert
For a very, very, very, very specific definition of "first" and "kind".
While the definition of “kind” may need to be somewhat specific, I don’t think “first” needs an unusual definition?
Thank you for your chary spirit, you give me motivation every time I watch your videos
Is that an ESD safe fingerprint on the chip when you are assembling it? 😆
I'm really impressed we can do this in our modern age, I thought they only let the biggest and brightest minds to have a crack!
I heard that multiplying numbers is hard for processors. So what if someone made a multiplication table chip? Would that speed up computers?
That was also probably the world's nerdiest rick roll!
Loved it!
I was like "if it isn't a rickroll I'll write a mad comment" the moment he mentioned it being something graphical
I was not disappointed :D
Just imagine if you uploaded the chip design only to discover that you forgot to place one junction
Imagine that simulation went a long way and this for top students who don’t make mistakes anyway.
that's why we encourage everyone to verify!
Where'd you get the solder paste stencil? Does Aisler provide those with their orders?
This is really really really cool, I am just sad you cant order quantities. $300 for a custom (really small) chip is awesome, just wish I could order 1000 more for $1 each.
Maybe through technical advances, economies of scale and multi-project wafer services, one day silicon production can get similar turn-around like what happened in the PCB industry.
It would take a lot of changes. Making stencils for all the layers is really expensive. You can’t “print” them out like with PCB where tracks are very wide (certainly compared to features in a chip). The tools used to dope silicon and deposit SiO2 are also not cheap. It would definitely be cool to get to that point. But it’s pretty far away for now
Fun fact: the human brain is able to recognize certain things instantly, such as faces, your own name, and the rickroll meme.
DIY photolithography and custom chips are a fucking insane concept.
Perfect sample choice.
Wow, fun and cool project bitluni!
Dang it, I never expected to be Rickr-olled in a video about a custom ASIC
a very interesting project :)
also a lot of funny moments :) Those revisions are the bane of every PCB designer :) Keep dong this, you are great !!!!
Nice video! This makes me want to learn more about ASIC
You pass butter!
Oh my GOD!
such an awesome project!
Great work! That Timelapse music sounds straight out of Sim City
This is such a cool concept.
So even as a group buy - what did one of those chips cost you :)? (and how much cheaper would it have been if you ordered more?)
Nice I was just about to get a small design made at Aisler, thanks for the five bucks. Tinytapeout is such a cool initiative too, if I ever have an idea for an ASIC now I can actually consider getting it made, if I can remember any of the VHDL from uni >.>
Thank you for using our service :)
the selection idea is genious!!!
I'm more interested in SIP and modules. You have dived much deeper. Actually, it would be enough if I gathered the ready-made products under one roof. (Especially in SIP form and Esp32 s3 and sx1262 chips)
When I'm lazy with a VR, i use a few diodes and resistor (series)... you know this tho....
Great project : )
10:10 the most billiant idea I've seen since I woke up today.
Clearly you are never going to give that ASIC up.
6:42 - I might have chosen a 91K with a 100K for the divider to get the 1.8V from the 3.3V, but maybe that would have been an extra component value that you didn’t care to have to “handle”
Kiloohms?? That would be way off, if the chip draws any current on the 1.8V net. Even the 100 Ohm voltage divider is far from ideal, as the current draw on a logic core supply is not static. So you modulate a voltage ripple on the core supply voltage thanks to the resistors. Depending on the chip, not meeting the core voltage specs and not using a stable supply voltage can give you all kinds of errors and issues that are super hard to track down. Worst case, it could even kill the core, and for a chip that you can't just order replacements, i'd think twice about the risk of destruction because of lazy circuit designs.
As this behaves like a rom, wouldn't it had been interesting to create a "rom", which returns a different image every time you read it? Then you could still animate something (by just reading/displaying the picture in a loop from the ESP32), but really make use of the ASIC
loved the video! Thanks Luni!
I have no clue at all except I am a bit aware of asic developement for FPV flying or goggles and the hd stream in a project that did not take off cause the investment back then was finally far too big - a million € figure ,not even a 2 or 3 but rather a 5 million figure to get that done. But funding did not happen.
But that was rather a companion role and no technical one. But I like this video really a lot and all the struggles with the iterations.
Nice one, bitluni! Great to see, and thanks for sharing this level of detail :) I hope you do more Tiny Tapeout submissions in future! Maybe time to try an analog layout...? :)
This is sooooo nerdy! But I love it!!!
How much was one chip? And the whole project?
I wish i knew about this, this is so cool!!
Fine solution to the microchip crisis! Easy to do some really simple and primitive CPUs!
Anyone knows how many ASICs you get for participating?
i was happy to be rickrolled :)
You are awesome. and thank you for this video
You used my text message ringtone! Mario 1-up
So how much would the cheapest fab charge for my own ASIC?
They were part of a marketing - research -google - stunt.
Im sure this is supposed to be super expensive
Its a $300 buy-in
Usually ~50'000$+, but this company is starting at just 10k$ and the guys are splitting the cost between 160 people and allocating them even smaller part of the chip so it's just 300$ per design.
In the future I hope there will be a discount for additional chips. Now you only get 1
@@ArneChristianRosenfeldtI can't possibly imagine that they didn't produce at LEAST double the amount they needed - the expensive part is creating the masks and setting everything up - running the scanners 1,2 or 10 times costs basically the same.
In the history of humanity, This will be remembered as a day when we all got rick-rolled by the first custom home-made ASIC.
hehe i made this kind of pinout mistake too on my board;p botched a fix before the re-design;p
Did you use some kind of special fluid to safely remove soldered parts?
If so, I have never heard of such a product and would be VERY interested to know what it is!
He just uses flux paste that helps the solder to flow better. Then you have to heat the solder to melting temperature again to remove the chip.
I guess you could try to find some very specific acid that only attacks the tin in the solder, but i don't think this exists and it would just eat away all the metals, including the legs of the chip or other components.
Congrats, no small achievement
Could one take a Mister FPGA core and build an ASIC form it? Can it do analog circuits? Would a C64 SID replacement be possible?
What model is your SMD reflow heater? Would you recommend it for purchase?
Awesome video as always
10:06 your quiet f**** made me laugh. That’s why I use a power cutter and shape it off the board before soldering the legs on.
now we need to make an asic that only runs doom
Oh you can compile c on toaster? I rewrote doom in silicon
Wait hasn't someone already done this with Redstone? Free template
I understood nothing but it was very nice to watch lol
Did anyone think of implementing a SID in one of the desings?
For TT02 the analog part (filters) would have not been possible. I've heard they want to include analog capabilities in the future revisions though! Someone is surely going to short the SID hoarders ;)
The TH-cam video from TT shoes an R2R DAC. Put those in 3 of the tiles. How does analog go over the bus?
@@MichalKobuszewski How well do you know the existing SID?
@@eitantal726 The SID schematics have been reproduced from the die shots. I can see that the filter uses opamps and FETs with gates driven by DACs to control cutoff. This is unfortunately a no-no for direct implementation, but if one were instead to implement a right biquad filter and a lookup table for its parameters (to perform the filtering like DSPs do) I guess it could work. Sadly the digital filter part would probably exceed the other (oscillator + ADSR) in complexity.
@@MichalKobuszewski I'm interested in that. Will you be willing to spend time explaining those to me?
great job !
Nicely done!
dime how did u get ur hand on a lithography machine that actually work
Did you need to reveal the content of the "secret file", or had been accepted without it?
The best choice of use for a custom ASIC XD
I mean, you could have made a VGA controler, or anything, and no, it's a Rick roll (without sound), glorious!
Send this on the next Voyager probe. 😂
How much money required for asic?
Bhai maja aa gaya
Epic
Bhai tm BABA h
Um... 9 months wait for the ASIC? Wow. You really need to plan ahead. Can't be in a rush to get the part. BTW, you probably realized by now that 3.3V divided by two is 1.65V and not 1.8V.
I'm not smart enough to do it but I would implement Ben Eater's breadboard computer.
It sounds like having an FPGA manufactured with factory programming. An ASIC that would be interesting would be like the T5L (used in DWIN display), but it would be more interesting if the IC supports being updated via OTA instead of SD card or JTAG.
Great, now i have to listen to "never gonna give you up" again 😂
is tinyTapeout only RTL, or can you do custom layout? also what tools are used for layout, i use cadence for work but that is anything but free and open source lol
You can do analog on the latest tinytapeouts. The most supported toolchain is xschem + magic + ngspice
You know the rules and so do I.
You, Sir, are a genious!
Thanks for the interesting video.
What PCB software do you use?