when you're running arch linux on a risc-v architecture in a fpga and still feel like you are getting robbed by tsmc and asml so you start your own foundry using dirt behind your garage...
I'm extremely surprised how not only he knows how to make the 2nd gen chip at home by himself, but what's even more impressive is that he explained it all basically in under 6 minutes only.
If you've done a degree in electronics at universities, some universities have professional grade equipment for making integrated circuits, so that is one way the knowledge can be obtained.
Eh, I’m fine with it. It is more self explanatory to layman like me than something like “home lab made silicon chip”. Actually, that’s more explanatory, but I’m still fine with homemade.
Thanks!! It’s a microscope so I look thru the eyepiece and move the XY around, the projected image is visible on the chip. First I project it using red light for alignment, then switch to UV for exposure once everything’s right
@@SamZeloof Your project is really cool! I can only imagine your homemade silicone chips improving! Do you think if making a mold(the mold blocking the unwanted rays) for the 2:06 part where instead of a projector, X-rays!(with a x-ray machine of sorts) Just an idea because the wavelength of x-rays is much smaller allowing for much smaller transistors(hopefully). And even possibly making it front and back sided on a single chip(theoretically for more complicated tech)? I am looking forward to your next homemade chip videos even if its next year... or the next! I am now subscribed!
@@blacdiamond2334 we are getting into dangerous territory here. theoretically, yes, but he's not limited by his wavelength yet. Remember, he's not in a cleanroom. There's only so much you can do in a garage. Also, working with xray is very very dangerous and not worth it at all. Also, there's a reason the industry skipped xray and moved into EUV (cost, risk, DUV is good enough, etc...)
As someone working in a large semiconductor fab it is surreal to see the work that we do carried out (albeit on a much smaller scale and much larger scale, if you catch my meaning) by one person in his garage. This is impressive as hell.
You mean you arent? Gee I always thought people working in those big semiconductor fabrication jobs would have mini manufacturing plants in their home like Breaking Bad but instead of Meth ya'll dealing silicon wafers.
@@Yeeren I guess it’s different in the silicon world. In the machinists world they all got lathes and 5 axis milling machines in their garages. Figured the more nerdier the industry the more diehard they are. Perhaps they overwork everyone in the silicon industry? Well we know it’s no secret the entire tech industry is overworked when they hang nets outside Apple buildings, and have yoga studios and massage parlors at work next to the Smoothie Bar.
@@ChristopherJones16 I suspect it’s partly a matter of scale. This video was amazing, but the tools used in this video are so different from what someone would use on a job, that the process is actually pretty different. It also means that what you can get out of it is really different. Unlike with machining, in the integrated circuit world you are much better off just buying whatever you want, because general purpose micro controllers, PLDs, and FPGAs are so cheap and flexible. What I saw in this video is absolutely awesome, but it’s awesome because it’s being done as a single man operation. Not because it beats a cheap FPGA in most uses.
Dude. You just brought the computer industry back into the garage... ;) Seriously very cool. With the latest 2/3nm processes being so out of reach of the average amateur geek, it is brilliant to be able to go back to basics again. And I love that you are showing everything, it is so much more interesting to actually see the process being done physically, seeing the etching process, washing, layering, etc.
I worked in a small semiconductor fab in the summers of 1972/73, your setup brought back some good memories of what it was like back then. We grew our own silicon from a melt. All seems a very long time ago now.
Just a question, he's doping a P-type wafer with an N-type solution, Does the mixture of these two impurities doesn't ruin the efficiency of these gates?
Interesting... Being semiconductor professional, I think that this could be a good example for college labs, where students can craft their own transistors and do manual layout. With some extra addition of steppers to have good alignment. After that, one can transition to shuttle wafers for higher integration. This gives so valuable hands-on experience! Great job!
Yes in our class all we got to do was a single layer "antenna"... and it was almost entirely hands off... I think someone setting a leaking canister of HF on thier foot a few years back spooked them out of letting anyone do anything serious in the lab as undergrads... which is sad.
That would mean actually teaching something, too dangerous. Let's just read some useless materials instead and teach some nonsense math that we will never go in depth anyways because computers already do that for us or something. And also let's charge a ridiculous amount of money to basically teach people to push a button.
@@ethanspaziani1070 it started like a medium for sharing knowledge, where people had to work hard to learn and get knowledge. Now it's just a Ford's style production line where the only purpose is to get a piece of paper for validation, no point in actually sharing knowledge. An empty shell of a glorious past. But who I am to question the social machine.
This is without a doubt one of the most impressive uses of the term "homemade" that I have ever seen! Can't wait to see how your process improves over time!
This is fantastic because so much of technological advancement is about building machines that then build smaller machines recursively - so being able to build a chip from scratch is a real talent and excellent for humanity too. You never know, you may discover something that was missed in the early development of chips that proves invaluable to modern chips. Great work!
Yup the lithography machine is a projector bolted to a microscope, so I can look thru the eyepiece to align chip layers. Use red light to align, UV to expose. but the image shifts and distorts when changing between wavelengths due to chromatic aberration so you need to do some calibration
You really know your stuff. I work on silicon diode radiation detectors which are quite easily ruined by any contamination or slightly wrong process conditions. Getting these mosfets to work in a garage is really impressive, especially with only one guy doing it! Btw, we also have the 4145A semiconductor analyzer in everyday use, old but gold :D
Someone : " Can you build a computer for me?" Sam :" Sure I can" Someone :" wait why don't we go into the computer store?" Sam : "You said you want me to build a computer"
I absolutely love how youre helping reduce the complexity for the masses to have a general enough understanding of the equipment and processes to make the building blocks of computers. You have successfully torn down hours of frustrating and difficult research on my end to achieve my goals. You have massive respect from me
lithography is but one part of designing useful chips. One of the really hard par is also how to connect those transistors into useful circuits. By no means is his work trivial, it's so freaking impressive to see how he does it, but having to hook all those transistors into patterns that can execute jobs is also another field in it self and can take years to release a newer and better version of your old processor. I think there's even more great videos coming from this guy I just found like 10 minutes ago :D
@@trentcoble3793 Yeah I'm so sorry that computer electronics is not available in Tennessee. I should have picked advanced buck fucking? that way you could have picked a college in Alabama or South Carolina since you seem drawn to educated people. Maybe that way you'd be closer to a real university over seas where I got my education. Idk that might be better then just relaying on other people's education, no? because idk about the whole "relay on people's UA-cam comments" to learn think, don't you think? ah right no you don't think for yourself! But hey (don't confuse it those words with buck or Hay) I don't want to be too redundant as to write too many words for big 'old Trent :)
@@trentcoble3793 you commented as if lithography was a mystical, alien technology that you had just been granted special access to, but then start crying in response to a comment pointing you to the further manufacturing marvels of processor design. "You have successfully torn down hours of frustrating and difficult research on my end to achieve my goals. You have massive respect from me" transformed into... "I honestly really don't think i ever want to figure out what is wrong with your brain" this change in your replies in response to a simple comment makes it clear which one of you could be suffering from brain damage. what is wrong with you Trent?
This is super impressive! I've worked in the semiconductor industry for quite some time now and would not have thought that this is possible. Thank you for proving me wrong!
This really makes me wonder what drives the cost of commercial IC manufacturing. I know the machines are expensive, but if he can scrap together some older stuff, it seems like there must be more to it for modern processes.
@@thealienrobotanthropologist Excellent question. As usual it is complicated. Overall cost per chip has actually fallen steadily, but the cost per factory has risen sharply. Bigger wafers need bigger machines and better alignment, which drives up the cost of the machines, but bigger wafers can contain more chips and thus increase production throughput (process time is dependent on number of wafers, not on number of chips). The trend towards smaller structures allows to put more chips on the same area, but also increases production cost, since the machines have to have better alignment and finer control over the process. Occassionally you need completely new processes because you hit a physical limit. So, overall the amount of chips per wafer increases quadratically with finer structure, but the machine cost also increases roughly quadratically with finer structures. Negatives and positives of both trends largely cancel out. In the end it comes out as cheaper chips because of automation and economy of scale. Automation allows higher production speed, more uniform quality and better control. Economy of scale is the reason that only a few super big companies are left at the leading edge of semi-conductor manufacturing. If you build a new factory you have to buy machines within the last few generations, because they just don't build the cheaper old ones anymore. (Why would they, if they can sell an expensive one?) And now for the big super secret sauce that no one talks about: there are still hundreds of older factories whose machines have paid for themselves twice over and which just crank out cheap, boring, routine chips for which nobody asks what feature size they have. Those factories also increase their automation and efficiency, but nobody talks about them... because micro-meters are boring - we want nanos!
@@KonradTheWizzard But I can't think of any piece of equipment that needs to be expensive enough to explain the cost of a new fab. Nothing in the video looks particularly expensive to build. HEPA filters, bunny suits, ESD protection, UV free lighting, etc. aren't that big of a deal financially. Optics of the caliber needed for the state of the art are expensive, but not enough to explain the fab cost. What makes a 5nm fab more expensive to build than a 50nm, 500nm, or 5000nm fab? If they kept the production volume the same, I'd expect the capital costs to be similar.
@@thealienrobotanthropologist Nothing in the video is particular expensive (on an industrial scale), because it works on relatively manageable scales. Let's work with your example scales and use etching as an example: 50µm: no problem, dunk it into some acid that is suitable for the job and wait till enough material is gone. Maybe stir a bit. 5000nm (5µm): not a big problem, you have to stir evenly, control the temperature a bit and be cognizant of the timing. 500nm: liquid etching is becoming really difficult, you have to precisely control the flow (speed and direction) of the acid, temperature has to be precise to at least 1K, timing must be precise to a fraction of a second. Using a plasma instead starts to look like a really good idea now. Not all liquids will even work at that feature size (see below). 50nm: forget liquids. Liquids will not enter such small features and if they are inside they will not leave. So you have to use plasma etching. Doing plasma etching on this level is much more expensive than a simple acid bath, it also requires timing down to milliseconds, high vacuum, flow control and a very controllable microwave. 5nm: the feature size is rapidly approaching a statistical basketball game of whether the ions (proverbial ball) will even find the feature wells they are supposed to go into (basket on a court in a different city). Steering the flow becomes more like black magic than science. That's just one example. Every process has hundreds of problems that need to be solved at every smaller feature size. Occassionally I spend a few hours with people who researched for decades on how to overcome one particularly tricky problem that I didn't even know existed at a specific feature size. It is fascinating for me to listen to... and was expensive for the companies spending the research dollars... I hope that gives you an idea.
I'm a computer sciecne student and people think I'm nerdy because I assemble and fly FPV drones and stuff, but you make me feel like an imbecile lol. You're AMAZING! Please do more videos like this, it's fascinating.
You're a student still my guy. You start somewhere. You end up wherever the journey takes you. Flying FPV drones is a good pursuit. It will teach you all the mechanics of flight plus you learn to troubleshoot and fix stuff every time you crash. Maybe even CAD and 3D Printing. Just follow whatever interests you.
watching people smarter or more capable than yourself keeps yourself groundet avoiding Dunning Kruger effect. So thats nice. More people should do that.
@@tami6867 fun fact: the dunning-kruger study (and ones replicating) uses a flawed methodology that introduces auto-correlation into it's conclusion; without correlating non-independently measured data, the effect disappears. alas, avoiding hubris has not such a simple remedy.
Absolutely incredible! I've been involved with IC's since the 4004 and even recognise some of the HP gear too, but you do seem to have used $500,000 of equipment to make a $1 chip ! I would never have thought it possible to do this type of fab work AT HOME. Just brilliant. _(Edited 2 months later to say :- The $500k I loosely mentioned refers to the prices when some of the historical gear featured was 'new'. I deal in this stuff today, and the same equipment sells for 1-5% of its original cost and generally it all works like new, and some have bang up to date calibration certs. But its bargain equipment now. Sometimes the gold inside is worth more than the used price of the whole machine !)_
@@techalyzer - some ppl buy a $80K car when a used $5K would get the job done. How we spend our money reflects both needs and wants. ( and he may be getting equipment donations from supporters, labs etc. )
not to argue here but first "so does every chip manufacturer everytime they made a chip." n second "kudos to the guy who makes single handedly a f*** microchip on his garage" this videos are amazing men wow
I just wanted to say that your work is awesome. I'm a software engineer and I've always wanted to look at fab'ing, but always felt like it was passed me. Stunning to see what you've accomplished in your garage. Keep it up!
Don't feel that! You can really fabricate your own PCB at home (not like this that you need a lab), and while the scale is larger, it's a very similar process. Have a look at it!
the question is, what do you wanna fab? as an SW-Eng I guess you want to start going one step down to HW -> so PCB edging and soldering your own board is much much easier and doesn't require his fairly expensive equipment (at least by what I can tell in the video). If you want to make it even easier on yourself: start by designing your own boards and circuitries and simply order the PCB online (very cheap these days) cause to be honest: there is not much to gain from making your own transistors ;)
@@diego032912 that comes down to how you define the word "useful"... having some self-made blinky-lights circuitry (which is about the integration level he can get to) is not much use, since you can buy a chip exponentially more powerful for much less of the money it costs him to run his lab for say, an hour. Point being: its very impressive, what he does, but useful? come on... If you want to make your own computer (and one, that deserve the name) start by writing your own programs, compilers, design your own boards, edge your own boards and solder them in your garage.... all of that is much less costly and way more useful than making your own semi-conductors.
This makes me so happy because it blows up this fear I have that if we have some kind of natural or man-made disaster (a big war) our society won't rebuild. With knowledge and skill like this, even stuff like this is possible in your own garage at home. Of course you have some pretty sweet gear. Awesome video.
Even if we went back to the middle ages from a disaster.. I speculate we wouldnt be in the "middle ages" for long.. It would probably take civilization a year to go from middle age to the 1800s in terms of technology and production(pre-industrial revolution). Look at what the Professor accomplished on Gilligan's Island : D . This would make for a great experiment/reality tv show. Take a 100 engineers, doctors, chemists, farmers, and miners, drop them on an island with nothing and see what they can accomplish in one year. I imagine it would be most impressive.. Look at what one man accomplishes in his youtube channel "Primitive Technology" (his channel is pure Gold). So have no fear. Society will bounce back quickly.
yes! this! the fear that the common man couldn't continue. the truth is, we would and seeing this makes me happy. even if WW3 happens, and modernity ends, we will probably leave enough information behind to grow again
I don't want to be that person but... The manual process used in this video was discovered in the 70s, almost 55 years ago. The structures modern devices use today are 1/10000 of the size which is shown here, which absolutely did take decades of equipment tuning to achieve. 90% of high-end wafers today are exported from TSMC fabs in the Taiwan, whose independence as you might know is in disagreement with the country with the big red flag. The reason why the red-flag-country won't ever touch Taiwan is because there is a big red button in the TSMC's Ceo office that when pressed, will set the whole world at least 15 years back in the silicon manufacturing progress. That's why the US and NATO keep very fresh tabs on what these two countries are doing.
Exactly my thoughts. There are exactly 3 companies producing chips for the entire planet. None of those are in the USA. And China is being a cunt. The world needs independence from them.
@@janospeh9503 Intel, TSMC and Texas Instruments I believe. TSMC is by far the largest by production and I think Texas Instruments only makes simple chips for calculators and other low power devices. There’s a few other smaller players, and a few that only operate in mainland China
@@jerry3790 Don't forget ones like Samsung, GlobalFoundries, Micron, and SK Hynix. Pretty much all the biggest companies do have fabs in the US as well. With the exception of China, most of these companies operate multinationally.
That’s pretty much a pro lab in my opinion , and what you’re doing is awesome and is opening doors for this to become common knowledge , thank you for sharing. I would really appreciate listing all the equipments to replicate such a great lab.
That's exactly what the chip manufacturers want you (and competitors) to think. Obviously they are light-years ahead of the general public unfortunately.
@@felizr_comSame, open source hardware is clearly feasible just not at a meaningful scale or jigh performance. I can see this being a possibility for as close to guaranteed secure communication as you can.
@@Jell0zz well, but a lot of that computing goes to bloatware. One can imagine garage made chips being used for regular business applications, and not hyper-realistic open-world war sims.
Was laughing at this one in the video 🤣 Don't know why - but it is funny to beat the industry by doing this at home. TechTechPotato channel should feature this - great work regardless 👍💪
Unreal! I was taught these concepts in school 15 years ago but could never practically visualize what it might mean to make such features (doping, wells, etc.). Also it would be amazing to de-centralize some of the semiconductor industries and do some grassroots innovation on things like custom sensors and FETs. Keep up the great work!
Honestly, it's got more value for instructional and maybe development purposes than for commercial purposes. "Boutique" ic manufacturing is too easily replaced by programmable logic for everything except analogue, and the analogue can be handler by D/A and A/D converters, or the (admittedly rare) "programmable analogue" chips, which just about always will price out this sort of thing. Education & development, in contrast, finds more value in turn-around time, direct experience, and total fab cost than in per-chip cost, and thus has great use for this sort of thing.
An industrialized version of his low spec process could be used to mass produce chip types that are too simple for the extreme factories that make extreme density modern chips. Things like basic logic and analog chips could be made from redesign or licensing from the ones that were introduced in 1965 to 1975. Also power electronics could benefit from the large feature size, such as input circuits for power supplies or DC to AC inverters for solar power etc.
@@johndododoe1411 Well I've got good news for you U.S. patents only last 20 years unlike copyright so anyone can actualy do whatever they want with those. A neat usecase if this goes ahead would be replacement chips for retro devices like computers and games consoles.
@@alexa.davronov1537 not really. Visual6502 has reverse engineered the original masks. He could copy the masksdirectly without even understanding how they work in detail. But bonding so many pins could be difficult.
@@Gameboygenius Seriously? Well, he would need not only a single mask, but many. Layers that connects individual transistors require separate steps in production.... There are many metallic layers on top of silicon die which require their own resist/etch/PVD cycle. He also would need to debug it ...
Bro this is fascinating, I think about you and your work once about every two weeks, I haven't seen any of your videos for over a year, yet you still come to my mind. I love what you do, it is very very much appreciated. Thank you!
Wow, that's awesome. I can't describe well enough how much of a paradigm shift this is for me to see that someone can actually build these from scratch in their own home. Huge kudos to you for putting the effort into learning all this stuff. I wish you all the best for future endeavours. This is so cool. Cheers!
@@Hisu0 By that logic, he also didn't mine the silicon himself, or create the tools to mine the silicon himself. He really didn't do anything at all, I guess.
@@Cackl_ It's not about credit, it's about validity of a claim. If you say (or imply) you made something from scratch, it means you made it without using the same [level of] technology. Otherwise, I can 3d print a hammer and claim I "made it at home" or "built it from scratch" when it's obviously not what I was doing. Similarly, it's not expected that you'd mine silica, because chip-grade silicon is available on the level of technology preceding chipmaking.
It's really great that you are providing this information to us. Chip manufacturing is such an enigma and I think an open source chip is something that people deserve to have. Thank you for your service!
No. People do not need open source mask programmable chips. That's a fallacy. What people need are field programmable gate arrays (FPGA). And these exist and you can use them without having to manufacture them yourself. A considerably more efficient way to implement a digital circuit. Home made mask programmable chips have only two uses, and that is for the purposes of fun and education only.
@@NuggetInAJar Then you obviously don't know anything about mask programmable integrated circuits and the equipment and highly dangerous chemicals involved. I do know. Because I have made them in a laboratory setting. So you are agreeing to something without understanding what's involved. Let me tell you now. There is no way on earth any government is going to allow the use of Hydrofluoric acid in a residential home setting. Never in a million years. So stop spewing out nonsense. And go learn more about it, much more about it before making such completely idiotic posts.
I had been following your progress for the last 2 years, and I have to say that I never expected you to get this far, you are an inspiration for many entire communities. Great work!
Finally, custom ASICs directly from your garage! Seriously though, I do believe that we could have small affordable machines to manufacture custom semiconductors at home in the next 20 years or so, just like 3d printers.
Bravo, that is absolutely amazing! I am 58 years old, and from NJ (Python developer) and opposite of standard, you are my inspiration young man. Great job!
Everyone here is amazed watching the process and I am here thinking that putting this lab together is a much more impressive feat that this (relatively) simple process. Knowledge about chip manufacturing and semiconductor devices is very accessible in this day and age. Installing (and potentially fixing) these ebay bought (and donated) equipment is no small feat! the information about this stuff is so obscure and requires experience (not just textbook knowledge). Not to mention the environment. This is not a cleanroom and alot can go wrong! Well done! BTW, I know Sam knows this but for aspiring engineers in their parents garages, DON"T MESS WITH HF! You can lose you hand very very slowly and without knowing. No amount or gloves will protect you (certainly not nylon gloves). I kind of wish he mentioned it in his video because it is a very serious silent killer
I have been wondering about this for AGES. I am soooo happy i found your page. Thanks to Wired for featuring you. And thank you! For putting this up. People like you really change the world by sharing online your knowledge. I might never make my own chips, but dammit i might!
In case you ever decide to give it a try, someone several years before also did it (just individual transistors though), and used some sort of ceramics oven for the heating step, with a ceramic wick of some sort (not fiberglass, but something else decently common) to lead high purity water into the oven for oxidization purposes.
Your work is very inspiring! Back in the early 80's, when Signetics was still a company, my cohorts designed IC's using a Lilith minicomputer. It had a whopping 10 Megabytes of storage and 2 Meg of ram. The programming language was Modula-2. This extremely efficient system allowed solving complex problems using a very modest amount of hardware. (1 Vax MIP)
@@mntrmntr Atari ST520 debuted in 1985. 512 k was pretty mindblowing in those days. Most homecomputers up t then had 64kb. Commodore 128 was launched in 1985 too.
@@Rxke The first I owned in the late 80s had only 16k. Good incentive to learn assembly language. Rich kids with their colorful commodores & spectrums only used theirs to play games.
Worked on my PhD in Electronics back in the 90s. This was more highly integrated than anything I managed back then. This is some dope (see what I did there) shit to manage all alone! Much kudos!
Will you try to simplify the process and make documentation so in theory anyone can do it? I would totally buy a textbook on this I feel your the only one who has complete knowledge of how the entire process is connected
@@GodzillaGoesGaga -- Yeah, and not dumping them in the yard. That was where MOS went wrong. CSG bought out MOS and inherited the pollution problem, then when a company bought out CSG, they abandoned the facility and the EPA deemed it a superfund site.
This is great! I've worked in chip design since the late 1980s. I remember the 4145 parameter analyser with the 5.25" disk drive very well! Your model must be at least 30-35 years old.
Incredibly cool! I once worked at an enterprise that produces integrated circuits and I understand that it is almost impossible to make even the simplest chip or transistor at home.
This is very very impressive! Hats off to your skill set and effort in achieving this. It would be a game changer to be able to create your own chips in the backroom of your house. We have FPGAs that can do digital, but personally, I think this would be a great idea for a start-up company to do analog chip design for all the rest of our needs. There are PCB houses galore. A chip service could possibly do well if the details were worked out. Fantastic work!
Not to kill the vibe but who needs one man making ICS that are 60 years behind in current technology. Again I think it's impressive and amazing. but it's not a business model making IC's at home.
@@laius6047 with a very small amount of funding, someone like this could easily produce current technology custom chips. Custom chips are EXTREMELY expensive to have produced as they are made in small quantities and set up like this could resolve that. You are deeply underestimating the possibilities due to a lack of understanding of the industry.
@@Brett3am "Small funding" got him a great home lab you can't do anything on the industrial scale with "Small funding" when it comes to chips, he can be a senior at TSMC 💯
@@smithwilliams5637 which is exactly why I said small quantity chips could be produced like this... the opposite of industrial scale. I know what I am talking about in this realm. Please re-read what I said.
I just saw that Atomic Semi(Sam Zeloof's new startup company) was co-created with Jim Keller. Two true and absolute legends joining together. I don't honestly know what's more impressive: inventing/con-inventing K8, x86_64, Zen at AMD, or building transistors and a whole semiconductor fab in your garage. Can't wait for what the future holds, I just hope he doesn't get too busy to make videos! Although I could understand why...
If I could make a suggestion on what to do next, a 555 timer IC would be great! It's one of these simpler but funamental ICs that are absolutely everywhere, and if it turns out we can make these at home - and they're reliable and functional - it'd be a great step for the maker community!
I'm not sure if home chips are all that useful for practical purposes, but a super neat project and a great way to teach and learn how computer parts work from the ground up
Just use FPGAs. This is super hardcore and looks like he easily has hundreds of thousands of dollars of equipment necessary to take a crack at this. Most people wouldn't even get as far as he did even with this equipment
This is awesome! I wanted to go in to grad school to do materials research for semiconductors but chose a different career path. I may wind up trying this because it's still an interest, thank you for your videos on this!
You have to consider the shotgun analogy here. The shotgun analogy goes like this, you destroy far less information shooting a billboard with a shotgun than a postage stamp. The postage stamp you completely obliterate where as the billboard just has some scars on it. We use the same idea making electronics for deep space applications. Our modern high tech doesn't work up there due to cosmic radiation. So we use large pitch electronics that can take the hits.
hey Sam: great intro with your own „moore‘s curve“! the incline of your progress let‘s me dream… where will humanity be in your universe in about a 40 year‘s time ❤😊
I think the retrocomputing crowd would be more interested in clones of the custom chips used in various old-school computers. Things like graphics or sound chips that were only ever used in one kind of machine.
Wicked! It's such an unknow process that it seems like magic when they show off new chips. I completely had no clue how they were made and barely understood how they worked. So this is great fun learning
"like a bought one!", as Dave would say. Could you also make power transistors (mosfets) this way, and how good would they be? I've been wondering where the main differences are in manufacture of logic transistors and high power ones, apart from size obviously. You'd need to create much thicker gate oxide, right?
@@T3sl4 Thanks for the reply. To be clear, do you mean a power transistor would benefit from small feature sizes as well? I was thinking exposing larger features (in 1 or 2 axis) shouldn't be a problem. But then if you just wanted higher current, you'd want to keep the channel length the same and increase the width, so the shape of it becomes more unreasonable. Is that what you meant? Basically I'm realizing that I know way too little about how power transistors are actually layed out haha. I dont know that much about semiconductors in general tbh, and I've only ever heard people discussing transistors for digital logic, and hence miniaturization.
@@Basement-Science Right, basically you take the shape he's built, and stretch out its width. Resistance drops proportionally. But now you need a super long die. So fold it over on itself a few times, and now you have a sinuous pattern that fits in a nice compact square. But if you're building that on a grid (because pixels, or whatever the finest unit of construction is), you can only fold as many times as there are pixels to work with. So, finer features can make a wider geometry, up to whatever the ultimate limiting factor is (feature size, breakdown voltage, etc.). Or alternately, just connect all the individual transistors in parallel, and figure how much space that would take up. Power transistors benefit even more from miniaturization, because in the vertical structure, the gate trench blocks some electric field in the drift region, increasing the field towards the channel and reducing breakdown for the same drift height; tighter spaced channels can use thinner drift region for the same voltage --> lower Rds(on) * Coss figure of merit. (I don't think this is too relevant here; RIE isn't so big a deal I think, but the subtle doping profile required to handle higher voltages, I think requires epitaxy(?). Which still maybe isn't that outrageous for someone this well equipped, but the worst part really is the process gas -- if you thought HF was bad, halo- and organo-metallic gasses are pretty awful too, and add to that, those of toxic elements like arsenic!)
Great stuff! I'm an electronics designer. Great to see that this kind of thing is within the reach of the average guy like me. I really like things that take our reliance off big corporations. Maybe you would like to try making an op-amp or 555 timer etc. These are basic elements that can be used for many applications. Looking forward to more of your videos
That is very astonishing! I am a physicist specialized in advanced technology, microelectronics and quantum physics and quite a DIY enthusiast of about anything and your work is an actual dream come true: a full chip made at home! I would like to congratulate sincerely with you 😁
Not that long I assume, I remember creating diodes with the described process, just less repetition because there is no on-chip interconnect in a practical course when studying physics, back in the 90s. I think it was a full day combined, from getting the explanations, through doing the process, to some final testing that confirmed your success. Once a process is well understood and well documented for students in future generations to learn it, it always becomes much easier to reproduce as long as you have access to the required tools and materials. The most complicated part in my opinion is learning how to safely operate the kind-of-open very high temperature oven and how to safely work with hypoflouric acid, since these are the steps in which you can really injure yourself instead of just ruining a test-sample. They are also the reason why there are no commercial "make your own microchips" toys on the market, and why most people will not try to do this at home even though they know how to do it well enough. I mean how to be safe around the oven can be learned of course, not really different than blacksmithing and other heat-related hobby activities, but dealing with hypofluoric acid will always stay very scary to me, even aftet successfully dealing with it in laboratory conditions a few times.
it's not about humongous gpu or something. It's more about custom-made highly specialized diy projects-on-chip instead of microcontrollers on a board, or some custom radio transcievers, something like that. Maybe a single-chip controller of the micro copter (size of a fly). Or if you can include piezo ceramics in it, maybe even some nano robots
@@jskratnyarlathotep8411 if chip making was made readily available like soldering, we would have stuff that used to be thousands of dollars for a couple hundred bucks. Just look at RTL-SDR. A full 100khz to 1.7 ghz receiver for 25$. In the 90s something like this would cost you thousands of dollars and would be the size of an oven.
I really want to see you make an ALU or if that is too complex for your IC at least an adder. Some flip-flops (maybe multiple) for basic ram chips would be cool as well. Of course the holy grail would be a (simple) microprocessor but even seeing an adder or flip-flop on a custom IC would be extremely impressive.
High density ROM and RAM were actually much more important than the microprocessor. A 1980s or 1990s personal computer without modern ROM and RAM had been impossible, while the same without a microprocessor would pose no major problems. A fast and simple processor can be built in TTL or one of the compatible CMOS families, just like it was in the 1970s. So ultra cheap kilobytes and megabytes from nano sized transistors were the key for the personal computer revolution, not the microprocessor.
This is honestly amazing... the word is out, and we are at the edge of something big. Today, Sam is literally the only person in the world to make chips in his home, but tomorrow, the retrocomputing community is going to jump at the opportunity to remake their old chips, and after that, it's going to be like 3D printing... Can't wait.
@@Intelwinsbigly we still use single transistors for all kinds of applications like LED lighting, amplifiers, solid state power switching, etc. I think this guy would do well trying to make IGBT's or Silicon Carbide transistors which are enormous but enormously important. Also, the big companies have stopped making RF frequency transistors and won't sell the IP to Rochester.
Incredible. I worked at Motorola in a fan doing these same processes. You are a very dedicated individual to undertake this lab work. Indeed I will watch your videos.
linux users when they understand that their cpu is prebuilt
most underrated comment
Unironically why I watched this
lmaoo 🤣🤣🤣🤣
bruh my kind of people where the fuck have i been my whole life
when you're running arch linux on a risc-v architecture in a fpga and still feel like you are getting robbed by tsmc and asml so you start your own foundry using dirt behind your garage...
Wow! You could make videos about all the details, and we'd watch every minute :)
Yes I know I would!
Details, like for example at 3:52, how come the middle section of SiO2 and Poly stays intact after etching ? (maybe I missed something)
Very much so this is just way too short for such an amazing accomplishment
YES!
Well well well! Yeah, we certainly need some APPLIED SCIENCE here.
I'm extremely surprised how not only he knows how to make the 2nd gen chip at home by himself, but what's even more impressive is that he explained it all basically in under 6 minutes only.
If you've done a degree in electronics at universities, some universities have professional grade equipment for making integrated circuits, so that is one way the knowledge can be obtained.
"home" bruh, did you see all the equipmement he have?
Eh, I’m fine with it. It is more self explanatory to layman like me than something like “home lab made silicon chip”. Actually, that’s more explanatory, but I’m still fine with homemade.
I still don't understand it
@@TheSonofseventhson he probably has 50k worth of equipment
This is amazing. How do you repeatably index the projection?
Thanks!! It’s a microscope so I look thru the eyepiece and move the XY around, the projected image is visible on the chip. First I project it using red light for alignment, then switch to UV for exposure once everything’s right
@@SamZeloof Your project is really cool! I can only imagine your homemade silicone chips improving! Do you think if making a mold(the mold blocking the unwanted rays) for the 2:06 part where instead of a projector, X-rays!(with a x-ray machine of sorts) Just an idea because the wavelength of x-rays is much smaller allowing for much smaller transistors(hopefully). And even possibly making it front and back sided on a single chip(theoretically for more complicated tech)? I am looking forward to your next homemade chip videos even if its next year... or the next! I am now subscribed!
@@blacdiamond2334 we are getting into dangerous territory here. theoretically, yes, but he's not limited by his wavelength yet. Remember, he's not in a cleanroom. There's only so much you can do in a garage. Also, working with xray is very very dangerous and not worth it at all. Also, there's a reason the industry skipped xray and moved into EUV (cost, risk, DUV is good enough, etc...)
@@himura128 All that gear - he's in a garage?
It's photolithography when you do it in a lab and ghetto-lithography when you do it in the house.
As someone working in a large semiconductor fab it is surreal to see the work that we do carried out (albeit on a much smaller scale and much larger scale, if you catch my meaning) by one person in his garage. This is impressive as hell.
I used to work IT for a company that sold high vacuum equipment - it's definitely impressive, especially given how expensive high vacuum is.
You mean you arent? Gee I always thought people working in those big semiconductor fabrication jobs would have mini manufacturing plants in their home like Breaking Bad but instead of Meth ya'll dealing silicon wafers.
@@ChristopherJones16 The last thing I wanna see at home after a 12 hour shift in the CR is more silicon wafers...
@@Yeeren I guess it’s different in the silicon world. In the machinists world they all got lathes and 5 axis milling machines in their garages. Figured the more nerdier the industry the more diehard they are. Perhaps they overwork everyone in the silicon industry? Well we know it’s no secret the entire tech industry is overworked when they hang nets outside Apple buildings, and have yoga studios and massage parlors at work next to the Smoothie Bar.
@@ChristopherJones16 I suspect it’s partly a matter of scale. This video was amazing, but the tools used in this video are so different from what someone would use on a job, that the process is actually pretty different. It also means that what you can get out of it is really different.
Unlike with machining, in the integrated circuit world you are much better off just buying whatever you want, because general purpose micro controllers, PLDs, and FPGAs are so cheap and flexible.
What I saw in this video is absolutely awesome, but it’s awesome because it’s being done as a single man operation. Not because it beats a cheap FPGA in most uses.
Almost 20 years of chip design and this is one the coolest things I've seen so far.
"Are you kids smoking reefer in the garage?" "No Mom, we are baking a phosphorous dopant at very high temperatures!"
What mom says reefer? Lmao
@@elihernandez330 mine and his
"and reefer too!"
@@elihernandez330 Who bakes a phosphorous dopant in his garage? ;-)
"No Mom, we are baking a phosphorous dopant at very high temperatures!" - I mean I'd say the same thing too :D
Dude. You just brought the computer industry back into the garage... ;) Seriously very cool.
With the latest 2/3nm processes being so out of reach of the average amateur geek, it is brilliant to be able to go back to basics again. And I love that you are showing everything, it is so much more interesting to actually see the process being done physically, seeing the etching process, washing, layering, etc.
can't wait for homemade 74hc610s
Not even actually at 2 nanometers
I worked in a small semiconductor fab in the summers of 1972/73, your setup brought back some good memories of what it was like back then. We grew our own silicon from a melt. All seems a very long time ago now.
i like to ask how hard it be to make cheap from 1990? because a lot of retro system from late 70s and 80s needing replacement chips and all that
Just a question, he's doping a P-type wafer with an N-type solution, Does the mixture of these two impurities doesn't ruin the efficiency of these gates?
Back when you could work in fabs as a summer job
I highly doubt that there are any entry level jobs in that industry anymore. Only job that is walk in ready at a fab is the lunch room cleaner
how old are you sir?
Interesting... Being semiconductor professional, I think that this could be a good example for college labs, where students can craft their own transistors and do manual layout. With some extra addition of steppers to have good alignment. After that, one can transition to shuttle wafers for higher integration. This gives so valuable hands-on experience! Great job!
Yes in our class all we got to do was a single layer "antenna"... and it was almost entirely hands off... I think someone setting a leaking canister of HF on thier foot a few years back spooked them out of letting anyone do anything serious in the lab as undergrads... which is sad.
That would mean actually teaching something, too dangerous. Let's just read some useless materials instead and teach some nonsense math that we will never go in depth anyways because computers already do that for us or something.
And also let's charge a ridiculous amount of money to basically teach people to push a button.
You know that schools today don't teach you anything useful right....
@@TheCuteZombie you get it
@@ethanspaziani1070 it started like a medium for sharing knowledge, where people had to work hard to learn and get knowledge.
Now it's just a Ford's style production line where the only purpose is to get a piece of paper for validation, no point in actually sharing knowledge. An empty shell of a glorious past.
But who I am to question the social machine.
This is without a doubt one of the most impressive uses of the term "homemade" that I have ever seen! Can't wait to see how your process improves over time!
Hopefully he can design a ALU or even basic CPU. That would be an incredible achievement
@@burntoutelectronics - True indeed! Heck, he's already surpassed by far the density of the ICs used in the Apollo space program! (if memory serves)
I was going to say something like, "homemade... well, I am not sure I can do this at MY home"... :D Awesome work!!
My home shop is not that well equipped. Thought I might see a garden rake or a bicycle hanging from the ceiling. 😂🤣 amazing contentNEW SUBSCRIBER.
I could not agree more. Amazing. Can you say "passion"?
This is fantastic because so much of technological advancement is about building machines that then build smaller machines recursively - so being able to build a chip from scratch is a real talent and excellent for humanity too.
You never know, you may discover something that was missed in the early development of chips that proves invaluable to modern chips.
Great work!
Technically not from scratch, but I guess we crossed that bridge a LONG time ago(see that one guy who tried to make a TOASTER from scratch).
@@vak2586 "If you want to build a toaster from scratch, you must first invent a universe"
@@vak2586 lmfao what
This is the most INCREDIBLE DIY project I've ever seen! Also the lab is crazy!
he could be enrichening Uranium in the lab ...😜
@@fidelcatsro6948 sure! haha!
check out david hahn
@@LoneWolf-wp9dn yes the boy who built a nuclear device in his home backyard or garage!
It is impressive and if you would like to see more at, (and sometimes above) this level, you can watch Ben from "Applied Science".
Sam, your ingenuity is amazing! I was in making chips in the late 70s and 80s, I recognize some of your equipment. Keep up your great work!
Do you still make now?
@@akashpaul909 read the damn comment again
@PRETTY GIRL TSMC hired him
How do you deal with alignment of the layers at each exposure step?
He shows the exposure setup can be used to observe the system too, I wonder if he can use a different light wavelength for alignment?
Yup the lithography machine is a projector bolted to a microscope, so I can look thru the eyepiece to align chip layers. Use red light to align, UV to expose. but the image shifts and distorts when changing between wavelengths due to chromatic aberration so you need to do some calibration
@@zoidbergVII He is not down to say any 7nm yet so an ordinary microscope and micro X-Y controls are more than enough.
I guess his process is similar to Huygens Optics, see Maskless Wafer Stepper videos. There is an explanation of those steps there.
@@SamZeloof this microscope alignment process deserves it's own video like all the other steps! Fascinating stuff!
You really know your stuff. I work on silicon diode radiation detectors which are quite easily ruined by any contamination or slightly wrong process conditions. Getting these mosfets to work in a garage is really impressive, especially with only one guy doing it!
Btw, we also have the 4145A semiconductor analyzer in everyday use, old but gold :D
Someone : " Can you build a computer for me?"
Sam :" Sure I can"
Someone :" wait why don't we go into the computer store?"
Sam : "You said you want me to build a computer"
To build a computer, first you must create an -Apple Pie- Universe
@@benholroyd5221 Yeah my comment is based on this statement
go mining for the raw ore, then come talk to me ! j/k This is awesome !
@@jasonrubik someone is doing it ua-cam.com/users/PrimitiveSkillsnetvideos
@@ThienPowChong I have been subscribed to him for years. I hope he is able to build an integrated circuit soon !
I absolutely love how youre helping reduce the complexity for the masses to have a general enough understanding of the equipment and processes to make the building blocks of computers. You have successfully torn down hours of frustrating and difficult research on my end to achieve my goals. You have massive respect from me
lithography is but one part of designing useful chips.
One of the really hard par is also how to connect those transistors into useful circuits.
By no means is his work trivial, it's so freaking impressive to see how he does it, but having to hook all those transistors into patterns that can execute jobs is also another field in it self and can take years to release a newer and better version of your old processor.
I think there's even more great videos coming from this guy I just found like 10 minutes ago :D
@@svampebob007 well arent you redundant and lacking In any education useful to me
@@trentcoble3793
Yeah I'm so sorry that computer electronics is not available in Tennessee.
I should have picked advanced buck fucking? that way you could have picked a college in Alabama or South Carolina since you seem drawn to educated people. Maybe that way you'd be closer to a real university over seas where I got my education.
Idk that might be better then just relaying on other people's education, no? because idk about the whole "relay on people's UA-cam comments" to learn think, don't you think? ah right no you don't think for yourself!
But hey (don't confuse it those words with buck or Hay) I don't want to be too redundant as to write too many words for big 'old Trent :)
@@svampebob007 I honestly really don't think i ever want to figure out what is wrong with your brain
@@trentcoble3793 you commented as if lithography was a mystical, alien technology that you had just been granted special access to, but then start crying in response to a comment pointing you to the further manufacturing marvels of processor design.
"You have successfully torn down hours of frustrating and difficult research on my end to achieve my goals. You have massive respect from me" transformed into... "I honestly really don't think i ever want to figure out what is wrong with your brain" this change in your replies in response to a simple comment makes it clear which one of you could be suffering from brain damage. what is wrong with you Trent?
2025, Sam: Oh, I finally made it to 5nm.
Intel: here's our new 10nm+++++++ process.
By 2030 Sam has developed a practical single nanometre transistor and has perfected general purpose Quantum Computation
He will be working for Intel or AMD before 2025, mark those words.
@@GeomancerHT Intel and AMD are too busy sitting in their high castles.
@@GeomancerHT More like TSMC or Samsung.
@@GeomancerHT Probably faster to build his own company and replace them…
Dude, that's an insane level of focus and commitment. Super, SUPER, impressive.
Just wow man. I can’t remember the last time I was this impressed by a UA-cam video.
ditto
Then you are subscribed to the wrong channels ;)
Those DIY chips just show again, how far the semiconductor industry has advanced in the last 60years or so.....
a girl also did this ... Jeri ellsworth.
Making Microchips at Home - Cooking with Jeri Part1 - UA-cam
@@AnalogDude_ That totally sounds like Jeri. She'll tackle anything. Fun to hang with too.
@@CommodoreGreg yeah, she is amazing, would love to meet her and discus ...
@@AnalogDude_ now, what if the two collaborated.... ;P
@@damiengates7581 hmm, the chemicals involved with it.
Dude's out here starting up the next Intel in his garage 💯
@MichaelKingsfordGray Who hurt you?!
@MichaelKingsfordGray Dude are you ok?
@MichaelKingsfordGray take your pills bro, or stop drinking.
@MichaelKingsfordGray Lost comments
Amd*
A man of so many talents. I work in a plating field and am a guitar amp tinkerer. This is a little bit everything done right.
I didn't even knew such a layer of diy could be achieved... Astonishing!
This is super impressive! I've worked in the semiconductor industry for quite some time now and would not have thought that this is possible. Thank you for proving me wrong!
This really makes me wonder what drives the cost of commercial IC manufacturing. I know the machines are expensive, but if he can scrap together some older stuff, it seems like there must be more to it for modern processes.
@@thealienrobotanthropologist Excellent question. As usual it is complicated. Overall cost per chip has actually fallen steadily, but the cost per factory has risen sharply.
Bigger wafers need bigger machines and better alignment, which drives up the cost of the machines, but bigger wafers can contain more chips and thus increase production throughput (process time is dependent on number of wafers, not on number of chips).
The trend towards smaller structures allows to put more chips on the same area, but also increases production cost, since the machines have to have better alignment and finer control over the process. Occassionally you need completely new processes because you hit a physical limit. So, overall the amount of chips per wafer increases quadratically with finer structure, but the machine cost also increases roughly quadratically with finer structures.
Negatives and positives of both trends largely cancel out.
In the end it comes out as cheaper chips because of automation and economy of scale. Automation allows higher production speed, more uniform quality and better control. Economy of scale is the reason that only a few super big companies are left at the leading edge of semi-conductor manufacturing.
If you build a new factory you have to buy machines within the last few generations, because they just don't build the cheaper old ones anymore. (Why would they, if they can sell an expensive one?)
And now for the big super secret sauce that no one talks about: there are still hundreds of older factories whose machines have paid for themselves twice over and which just crank out cheap, boring, routine chips for which nobody asks what feature size they have. Those factories also increase their automation and efficiency, but nobody talks about them... because micro-meters are boring - we want nanos!
@@KonradTheWizzard But I can't think of any piece of equipment that needs to be expensive enough to explain the cost of a new fab. Nothing in the video looks particularly expensive to build. HEPA filters, bunny suits, ESD protection, UV free lighting, etc. aren't that big of a deal financially. Optics of the caliber needed for the state of the art are expensive, but not enough to explain the fab cost. What makes a 5nm fab more expensive to build than a 50nm, 500nm, or 5000nm fab? If they kept the production volume the same, I'd expect the capital costs to be similar.
@@thealienrobotanthropologist Nothing in the video is particular expensive (on an industrial scale), because it works on relatively manageable scales. Let's work with your example scales and use etching as an example:
50µm: no problem, dunk it into some acid that is suitable for the job and wait till enough material is gone. Maybe stir a bit.
5000nm (5µm): not a big problem, you have to stir evenly, control the temperature a bit and be cognizant of the timing.
500nm: liquid etching is becoming really difficult, you have to precisely control the flow (speed and direction) of the acid, temperature has to be precise to at least 1K, timing must be precise to a fraction of a second. Using a plasma instead starts to look like a really good idea now. Not all liquids will even work at that feature size (see below).
50nm: forget liquids. Liquids will not enter such small features and if they are inside they will not leave. So you have to use plasma etching. Doing plasma etching on this level is much more expensive than a simple acid bath, it also requires timing down to milliseconds, high vacuum, flow control and a very controllable microwave.
5nm: the feature size is rapidly approaching a statistical basketball game of whether the ions (proverbial ball) will even find the feature wells they are supposed to go into (basket on a court in a different city). Steering the flow becomes more like black magic than science.
That's just one example. Every process has hundreds of problems that need to be solved at every smaller feature size.
Occassionally I spend a few hours with people who researched for decades on how to overcome one particularly tricky problem that I didn't even know existed at a specific feature size. It is fascinating for me to listen to... and was expensive for the companies spending the research dollars...
I hope that gives you an idea.
I love the Moore‘s law chart, that was a good laugh!
Dude! You are awesome! There are countries in the world that can not do what you can do. Respect!!!!
I'm a computer sciecne student and people think I'm nerdy because I assemble and fly FPV drones and stuff, but you make me feel like an imbecile lol. You're AMAZING! Please do more videos like this, it's fascinating.
I know right, I custom designed a bunch of electrostatic genies this year that make around 100 KV and this video made me feel like a caveman....
You're a student still my guy. You start somewhere. You end up wherever the journey takes you. Flying FPV drones is a good pursuit. It will teach you all the mechanics of flight plus you learn to troubleshoot and fix stuff every time you crash. Maybe even CAD and 3D Printing. Just follow whatever interests you.
watching people smarter or more capable than yourself keeps yourself groundet avoiding Dunning Kruger effect. So thats nice. More people should do that.
@@tami6867 fun fact: the dunning-kruger study (and ones replicating) uses a flawed methodology that introduces auto-correlation into it's conclusion; without correlating non-independently measured data, the effect disappears. alas, avoiding hubris has not such a simple remedy.
Absolutely incredible! I've been involved with IC's since the 4004 and even recognise some of the HP gear too, but you do seem to have used $500,000 of equipment to make a $1 chip ! I would never have thought it possible to do this type of fab work AT HOME. Just brilliant. _(Edited 2 months later to say :- The $500k I loosely mentioned refers to the prices when some of the historical gear featured was 'new'. I deal in this stuff today, and the same equipment sells for 1-5% of its original cost and generally it all works like new, and some have bang up to date calibration certs. But its bargain equipment now. Sometimes the gold inside is worth more than the used price of the whole machine !)_
That may have been the equipment cost new but an used HP4145A goes for about $3500 on eBay now a days.
@@anotherdave5107 Still in the realm of impossibility for many of us.
hes developing not producing
@@techalyzer - some ppl buy a $80K car when a used $5K would get the job done. How we spend our money reflects both needs and wants. ( and he may be getting equipment donations from supporters, labs etc. )
not to argue here but first "so does every chip manufacturer everytime they made a chip." n second "kudos to the guy who makes single handedly a f*** microchip on his garage" this videos are amazing men wow
I just wanted to say that your work is awesome. I'm a software engineer and I've always wanted to look at fab'ing, but always felt like it was passed me. Stunning to see what you've accomplished in your garage. Keep it up!
Don't feel that! You can really fabricate your own PCB at home (not like this that you need a lab), and while the scale is larger, it's a very similar process. Have a look at it!
the question is, what do you wanna fab? as an SW-Eng I guess you want to start going one step down to HW -> so PCB edging and soldering your own board is much much easier and doesn't require his fairly expensive equipment (at least by what I can tell in the video). If you want to make it even easier on yourself: start by designing your own boards and circuitries and simply order the PCB online (very cheap these days)
cause to be honest: there is not much to gain from making your own transistors ;)
@@thetomster7625 being able to make your own computer from scratch isn't useful... you sure?
@@diego032912 that comes down to how you define the word "useful"... having some self-made blinky-lights circuitry (which is about the integration level he can get to) is not much use, since you can buy a chip exponentially more powerful for much less of the money it costs him to run his lab for say, an hour. Point being: its very impressive, what he does, but useful? come on...
If you want to make your own computer (and one, that deserve the name) start by writing your own programs, compilers, design your own boards, edge your own boards and solder them in your garage.... all of that is much less costly and way more useful than making your own semi-conductors.
This makes me so happy because it blows up this fear I have that if we have some kind of natural or man-made disaster (a big war) our society won't rebuild. With knowledge and skill like this, even stuff like this is possible in your own garage at home. Of course you have some pretty sweet gear. Awesome video.
Even if we went back to the middle ages from a disaster.. I speculate we wouldnt be in the "middle ages" for long.. It would probably take civilization a year to go from middle age to the 1800s in terms of technology and production(pre-industrial revolution). Look at what the Professor accomplished on Gilligan's Island : D . This would make for a great experiment/reality tv show. Take a 100 engineers, doctors, chemists, farmers, and miners, drop them on an island with nothing and see what they can accomplish in one year. I imagine it would be most impressive.. Look at what one man accomplishes in his youtube channel "Primitive Technology" (his channel is pure Gold). So have no fear. Society will bounce back quickly.
Hahahhahaah. Same
To be fair, his garage is well equipped with chemicals and electronic equipment from other sources.
yes! this!
the fear that the common man couldn't continue. the truth is, we would and seeing this makes me happy. even if WW3 happens, and modernity ends, we will probably leave enough information behind to grow again
I don't want to be that person but... The manual process used in this video was discovered in the 70s, almost 55 years ago. The structures modern devices use today are 1/10000 of the size which is shown here, which absolutely did take decades of equipment tuning to achieve. 90% of high-end wafers today are exported from TSMC fabs in the Taiwan, whose independence as you might know is in disagreement with the country with the big red flag. The reason why the red-flag-country won't ever touch Taiwan is because there is a big red button in the TSMC's Ceo office that when pressed, will set the whole world at least 15 years back in the silicon manufacturing progress. That's why the US and NATO keep very fresh tabs on what these two countries are doing.
The World: Silicon Shortage
This Man: Hold my Beer
you win the internet today lol
Exactly my thoughts. There are exactly 3 companies producing chips for the entire planet. None of those are in the USA. And China is being a cunt. The world needs independence from them.
@@ericbenjaminjr only 3?
@@janospeh9503 Intel, TSMC and Texas Instruments I believe. TSMC is by far the largest by production and I think Texas Instruments only makes simple chips for calculators and other low power devices. There’s a few other smaller players, and a few that only operate in mainland China
@@jerry3790 Don't forget ones like Samsung, GlobalFoundries, Micron, and SK Hynix. Pretty much all the biggest companies do have fabs in the US as well. With the exception of China, most of these companies operate multinationally.
That’s pretty much a pro lab in my opinion , and what you’re doing is awesome and is opening doors for this to become common knowledge , thank you for sharing. I would really appreciate listing all the equipments to replicate such a great lab.
Very impressive Sam! You have definitely made a lot of progress looking at the characteristics, and also in particle control.
Darn! My definition of ‘Garage lab’ just got violently upgraded…
That is an impressive accomplishment!
Well done!!!
Incredible!
I've been in electronics for 40 years and never thought home built chips were possible.
That's exactly what the chip manufacturers want you (and competitors) to think.
Obviously they are light-years ahead of the general public unfortunately.
@@Jell0zz agree
@@felizr_comSame, open source hardware is clearly feasible just not at a meaningful scale or jigh performance. I can see this being a possibility for as close to guaranteed secure communication as you can.
@@karatefrosch17 I do think that open source hardware is worth for life long project, at least I have something to be proud of when I was old.
@@Jell0zz well, but a lot of that computing goes to bloatware. One can imagine garage made chips being used for regular business applications, and not hyper-realistic open-world war sims.
Zaloof's Law > Moore's Law
Moore's law doubles, Zaloof's quadruple :) before year 2030 Zoolaf's Z10500 vs Intel i25500 and Intel loose by 20% would be hilarious
Was laughing at this one in the video 🤣 Don't know why - but it is funny to beat the industry by doing this at home. TechTechPotato channel should feature this - great work regardless 👍💪
Unreal! I was taught these concepts in school 15 years ago but could never practically visualize what it might mean to make such features (doping, wells, etc.). Also it would be amazing to de-centralize some of the semiconductor industries and do some grassroots innovation on things like custom sensors and FETs. Keep up the great work!
Honestly, it's got more value for instructional and maybe development purposes than for commercial purposes.
"Boutique" ic manufacturing is too easily replaced by programmable logic for everything except analogue, and the analogue can be handler by D/A and A/D converters, or the (admittedly rare) "programmable analogue" chips, which just about always will price out this sort of thing.
Education & development, in contrast, finds more value in turn-around time, direct experience, and total fab cost than in per-chip cost, and thus has great use for this sort of thing.
@@absalomdraconis Analogue computing may actualy pick up quite a bit in the near future, it turns out to be very effective for machine learning.
@@TheAkashicTraveller I saw that recently too. Can't wait to see more!
An industrialized version of his low spec process could be used to mass produce chip types that are too simple for the extreme factories that make extreme density modern chips. Things like basic logic and analog chips could be made from redesign or licensing from the ones that were introduced in 1965 to 1975. Also power electronics could benefit from the large feature size, such as input circuits for power supplies or DC to AC inverters for solar power etc.
@@johndododoe1411 Well I've got good news for you U.S. patents only last 20 years unlike copyright so anyone can actualy do whatever they want with those. A neat usecase if this goes ahead would be replacement chips for retro devices like computers and games consoles.
People can you imagine, what could Sam do with enough time and resources. The real world ironman vibes i am sensing here.Great job man.
Sam will be prepared if he gets sent back in time or if the apocalypse happens. He’ll speed run the digital revolution
i've been dreaming all my life to be able to produce silicon chips , i just adore what you did , please keep up
This is amazing. It's on my bucket list to spin an ASIC, but doing it in your garage on your own equipment is just... Inspiring. Wow
It would be cool if you can "homebrew" a 6502 compatible microprocessor.
C64 SID would be cool too.
Lol well, this would require him to create a logic which is very tedious process that only looks simple.
Give it a few years and it will go this way, I think we will have enthusiasts gradually recreating all the missing chips, maybe a few years yet.
@@alexa.davronov1537 not really. Visual6502 has reverse engineered the original masks. He could copy the masksdirectly without even understanding how they work in detail. But bonding so many pins could be difficult.
@@Gameboygenius Seriously? Well, he would need not only a single mask, but many. Layers that connects individual transistors require separate steps in production....
There are many metallic layers on top of silicon die which require their own resist/etch/PVD cycle. He also would need to debug it ...
Bro this is fascinating, I think about you and your work once about every two weeks, I haven't seen any of your videos for over a year, yet you still come to my mind. I love what you do, it is very very much appreciated. Thank you!
Wow, that's awesome. I can't describe well enough how much of a paradigm shift this is for me to see that someone can actually build these from scratch in their own home. Huge kudos to you for putting the effort into learning all this stuff. I wish you all the best for future endeavours. This is so cool. Cheers!
It really is cool, but a good part of equipment used in the process contains modern chips, so it's not _really_ from scratch.
He's opening doors for "millions".
@@Hisu0 By that logic, he also didn't mine the silicon himself, or create the tools to mine the silicon himself. He really didn't do anything at all, I guess.
@@Cackl_
It's not about credit, it's about validity of a claim. If you say (or imply) you made something from scratch, it means you made it without using the same [level of] technology. Otherwise, I can 3d print a hammer and claim I "made it at home" or "built it from scratch" when it's obviously not what I was doing.
Similarly, it's not expected that you'd mine silica, because chip-grade silicon is available on the level of technology preceding chipmaking.
It's really great that you are providing this information to us. Chip manufacturing is such an enigma and I think an open source chip is something that people deserve to have. Thank you for your service!
No. People do not need open source mask programmable chips. That's a fallacy.
What people need are field programmable gate arrays (FPGA). And these exist and you can use them without having to manufacture them yourself. A considerably more efficient way to implement a digital circuit.
Home made mask programmable chips have only two uses, and that is for the purposes of fun and education only.
@@deang5622 He did not say it would be useful in commercial applications, just that people deserve to have it. And I agree.
@@NuggetInAJar Then you obviously don't know anything about mask programmable integrated circuits and the equipment and highly dangerous chemicals involved. I do know. Because I have made them in a laboratory setting.
So you are agreeing to something without understanding what's involved.
Let me tell you now. There is no way on earth any government is going to allow the use of Hydrofluoric acid in a residential home setting. Never in a million years.
So stop spewing out nonsense.
And go learn more about it, much more about it before making such completely idiotic posts.
I would argue the knowledge is equally as important. @@deang5622
😪@@deang5622
I had been following your progress for the last 2 years, and I have to say that I never expected you to get this far, you are an inspiration for many entire communities. Great work!
Where have you been, need more more ... of this successfull builds. Regards
Finally, custom ASICs directly from your garage!
Seriously though, I do believe that we could have small affordable machines to manufacture custom semiconductors at home in the next 20 years or so, just like 3d printers.
Wow, I put a shelf up the other day and was proud of myself... You my man are a different level!
hahaha
@@user-xf7kk4tq8v 🤣
Bravo, that is absolutely amazing! I am 58 years old, and from NJ (Python developer) and opposite of standard, you are my inspiration young man. Great job!
Everyone here is amazed watching the process and I am here thinking that putting this lab together is a much more impressive feat that this (relatively) simple process. Knowledge about chip manufacturing and semiconductor devices is very accessible in this day and age. Installing (and potentially fixing) these ebay bought (and donated) equipment is no small feat! the information about this stuff is so obscure and requires experience (not just textbook knowledge). Not to mention the environment. This is not a cleanroom and alot can go wrong! Well done!
BTW, I know Sam knows this but for aspiring engineers in their parents garages, DON"T MESS WITH HF! You can lose you hand very very slowly and without knowing. No amount or gloves will protect you (certainly not nylon gloves). I kind of wish he mentioned it in his video because it is a very serious silent killer
I have been wondering about this for AGES. I am soooo happy i found your page. Thanks to Wired for featuring you. And thank you! For putting this up. People like you really change the world by sharing online your knowledge. I might never make my own chips, but dammit i might!
In case you ever decide to give it a try, someone several years before also did it (just individual transistors though), and used some sort of ceramics oven for the heating step, with a ceramic wick of some sort (not fiberglass, but something else decently common) to lead high purity water into the oven for oxidization purposes.
@@absalomdraconis Who did it?
Sincere congratulations, this is the best DIY project I've ever witnessed. I will watch everything if you post more details.
Your work is very inspiring!
Back in the early 80's, when Signetics was still a company, my cohorts designed IC's using a Lilith minicomputer. It had a whopping 10 Megabytes of storage and 2 Meg of ram.
The programming language was Modula-2. This extremely efficient system allowed solving complex problems using a very modest amount of hardware. (1 Vax MIP)
2 megs of ram in the early 80's? you sure you didn't misspell "huge" as "modest"?
@@mntrmntr In those days, 2 megs was huge. Currently we are talking micro controllers with 2X that! It is a joy to be able to make fun of that :)
@@mntrmntr Atari ST520 debuted in 1985. 512 k was pretty mindblowing in those days. Most homecomputers up t then had 64kb. Commodore 128 was launched in 1985 too.
@@Rxke The first I owned in the late 80s had only 16k. Good incentive to learn assembly language. Rich kids with their colorful commodores & spectrums only used theirs to play games.
@@mntrmntr 4k here... :) Wrote my own games. They were terrible heehee!
Worked on my PhD in Electronics back in the 90s. This was more highly integrated than anything I managed back then. This is some dope (see what I did there) shit to manage all alone! Much kudos!
One of the most impressive videos I have seen in such a long time.
Will you try to simplify the process and make documentation so in theory anyone can do it? I would totally buy a textbook on this I feel your the only one who has complete knowledge of how the entire process is connected
I think the problem is that a lot of his equipment is quite specialized and not widely availible
There's a discord called Silicon Overlords where a bunch of us work on this stuff. That might be a good place to start.
@@SIGSEGV1337 You got a link?
@@SHADOW-id6vw Yep, that's what I'm doing, everything from scratch. Oven, vacuum metal deposition, etc...
This is already simplest way to do it.
Just, whoa. I was thinking "hey, I could do this", up until you brought up the aluminium sputtering.
Al sputter is not too bad if you can find a cheap piece of vac equipment to steal pumps and valves from, like a dead NMR, SEM, leak checker, etc.
Dude.. You are brilliant, young man!!
"Coverage of the whole Waffer" *stares at the one corner*
I saw that too!
I've designed ICs before and this is absolutely amazing! I never thougth anyone would make chips at home! I am stunned! Fantastic work!
Same but damn I hope he is taking precautions with those very carcinogenic chemicals.
@@GodzillaGoesGaga -- Yeah, and not dumping them in the yard. That was where MOS went wrong. CSG bought out MOS and inherited the pollution problem, then when a company bought out CSG, they abandoned the facility and the EPA deemed it a superfund site.
This is insane, you've taken hobby computing to a new level.
while people write comments here now, now he is making a groundbreaking z3
This is great! I've worked in chip design since the late 1980s. I remember the 4145 parameter analyser with the 5.25" disk drive very well! Your model must be at least 30-35 years old.
This is so freaking amazing. I find the whole IC fabrication process crazy and it is awesome to see someone doing it at home!
Incredibly cool! I once worked at an enterprise that produces integrated circuits and I understand that it is almost impossible to make even the simplest chip or transistor at home.
This is very very impressive! Hats off to your skill set and effort in achieving this. It would be a game changer to be able to create your own chips in the backroom of your house. We have FPGAs that can do digital, but personally, I think this would be a great idea for a start-up company to do analog chip design for all the rest of our needs. There are PCB houses galore. A chip service could possibly do well if the details were worked out. Fantastic work!
Super rad stuff, great work as always! Like everyone else mentioned, I'd watch pretty much anything you have to share on the subject :)
It absolutely blows my mind that no one has started funding you at a VERY serious level! This is mind blowing stuff!
Not to kill the vibe but who needs one man making ICS that are 60 years behind in current technology. Again I think it's impressive and amazing. but it's not a business model making IC's at home.
@@laius6047 with a very small amount of funding, someone like this could easily produce current technology custom chips. Custom chips are EXTREMELY expensive to have produced as they are made in small quantities and set up like this could resolve that. You are deeply underestimating the possibilities due to a lack of understanding of the industry.
@@Brett3am "Small funding" got him a great home lab you can't do anything on the industrial scale with "Small funding" when it comes to chips, he can be a senior at TSMC 💯
@@smithwilliams5637 which is exactly why I said small quantity chips could be produced like this... the opposite of industrial scale. I know what I am talking about in this realm. Please re-read what I said.
Jim Keller discovered him and they founded a company together.
Love your work on Semiconductors, you make it seem so simple, like it is within our reach to do this work ourselves.
Power on ahead.
I just saw that Atomic Semi(Sam Zeloof's new startup company) was co-created with Jim Keller. Two true and absolute legends joining together. I don't honestly know what's more impressive: inventing/con-inventing K8, x86_64, Zen at AMD, or building transistors and a whole semiconductor fab in your garage. Can't wait for what the future holds, I just hope he doesn't get too busy to make videos! Although I could understand why...
If I could make a suggestion on what to do next, a 555 timer IC would be great! It's one of these simpler but funamental ICs that are absolutely everywhere, and if it turns out we can make these at home - and they're reliable and functional - it'd be a great step for the maker community!
I'm not sure if home chips are all that useful for practical purposes, but a super neat project and a great way to teach and learn how computer parts work from the ground up
They would cost way more homemade than just buying.
Just use FPGAs. This is super hardcore and looks like he easily has hundreds of thousands of dollars of equipment necessary to take a crack at this. Most people wouldn't even get as far as he did even with this equipment
Tha'ts true.
Or an LM3909.
Damn impressive! Here I am still etching PC boards for my prototypes and there you are etching your own silicon chips! I am thoroughly impressed.
please know damn is a curse word
This is the absolute coolest channel out there!!
This is awesome! I wanted to go in to grad school to do materials research for semiconductors but chose a different career path. I may wind up trying this because it's still an interest, thank you for your videos on this!
This is incredible work, never knew you could do this outside of a clean room! Awesome work would definitely support.
At this scale, cleanroom controls don't need to be very stringent. As features get smaller it becomes far more critical.
You have to consider the shotgun analogy here. The shotgun analogy goes like this, you destroy far less information shooting a billboard with a shotgun than a postage stamp. The postage stamp you completely obliterate where as the billboard just has some scars on it. We use the same idea making electronics for deep space applications. Our modern high tech doesn't work up there due to cosmic radiation. So we use large pitch electronics that can take the hits.
Wow, that is seriously impressive work! I definitely want to see more on this - bonding, in circuit use, chip design, etc.
Absolutely amazing! Shows what is possible with dedicated focus.
hey Sam: great intro with your own „moore‘s curve“!
the incline of your progress let‘s me dream… where will humanity be in your universe in about a 40 year‘s time ❤😊
This is absolutely insane and amazing.
On a side note I have a future goal idea for you: make an interchangeable clone of a 6502.
I think the retrocomputing crowd would be more interested in clones of the custom chips used in various old-school computers. Things like graphics or sound chips that were only ever used in one kind of machine.
New SID chips would appeal to musicians too.
This would be such an incredible video to collab with Ben Eater on
@@parad0xheart yeah, I mean, how about making a 555 or something like that before getting into processors. lol
Wicked! It's such an unknow process that it seems like magic when they show off new chips. I completely had no clue how they were made and barely understood how they worked. So this is great fun learning
"like a bought one!", as Dave would say.
Could you also make power transistors (mosfets) this way, and how good would they be? I've been wondering where the main differences are in manufacture of logic transistors and high power ones, apart from size obviously. You'd need to create much thicker gate oxide, right?
Not very good ones: these are large-feature lateral MOSFETs, whereas power transistors are vertical (VDMOS) and medium to fine feature (
@@T3sl4 Thanks for the reply.
To be clear, do you mean a power transistor would benefit from small feature sizes as well? I was thinking exposing larger features (in 1 or 2 axis) shouldn't be a problem. But then if you just wanted higher current, you'd want to keep the channel length the same and increase the width, so the shape of it becomes more unreasonable. Is that what you meant?
Basically I'm realizing that I know way too little about how power transistors are actually layed out haha. I dont know that much about semiconductors in general tbh, and I've only ever heard people discussing transistors for digital logic, and hence miniaturization.
@@Basement-Science Right, basically you take the shape he's built, and stretch out its width. Resistance drops proportionally. But now you need a super long die. So fold it over on itself a few times, and now you have a sinuous pattern that fits in a nice compact square. But if you're building that on a grid (because pixels, or whatever the finest unit of construction is), you can only fold as many times as there are pixels to work with. So, finer features can make a wider geometry, up to whatever the ultimate limiting factor is (feature size, breakdown voltage, etc.). Or alternately, just connect all the individual transistors in parallel, and figure how much space that would take up.
Power transistors benefit even more from miniaturization, because in the vertical structure, the gate trench blocks some electric field in the drift region, increasing the field towards the channel and reducing breakdown for the same drift height; tighter spaced channels can use thinner drift region for the same voltage --> lower Rds(on) * Coss figure of merit.
(I don't think this is too relevant here; RIE isn't so big a deal I think, but the subtle doping profile required to handle higher voltages, I think requires epitaxy(?). Which still maybe isn't that outrageous for someone this well equipped, but the worst part really is the process gas -- if you thought HF was bad, halo- and organo-metallic gasses are pretty awful too, and add to that, those of toxic elements like arsenic!)
wow, this video is gold at room temperature.
What happened to you ? What are you doing today ?
the opps got him
He runs a startup called atomic semi
@@HyperFXMotionDesign buy the stocks
My same thought
Sunspots cause membranes to become insane. Wanna play a game I’m bored
This is absolutely insane. Amazing job!
Great stuff!
I'm an electronics designer. Great to see that this kind of thing is within the reach of the average guy like me. I really like things that take our reliance off big corporations.
Maybe you would like to try making an op-amp or 555 timer etc. These are basic elements that can be used for many applications.
Looking forward to more of your videos
The fact that this is even possible at home is mind blowing.
That is very astonishing! I am a physicist specialized in advanced technology, microelectronics and quantum physics and quite a DIY enthusiast of about anything and your work is an actual dream come true: a full chip made at home! I would like to congratulate sincerely with you 😁
This is absolutely insane! Can't even imagine how long it must have taken you to figure out the whole process.
Not that long I assume, I remember creating diodes with the described process, just less repetition because there is no on-chip interconnect in a practical course when studying physics, back in the 90s. I think it was a full day combined, from getting the explanations, through doing the process, to some final testing that confirmed your success.
Once a process is well understood and well documented for students in future generations to learn it, it always becomes much easier to reproduce as long as you have access to the required tools and materials.
The most complicated part in my opinion is learning how to safely operate the kind-of-open very high temperature oven and how to safely work with hypoflouric acid, since these are the steps in which you can really injure yourself instead of just ruining a test-sample. They are also the reason why there are no commercial "make your own microchips" toys on the market, and why most people will not try to do this at home even though they know how to do it well enough. I mean how to be safe around the oven can be learned of course, not really different than blacksmithing and other heat-related hobby activities, but dealing with hypofluoric acid will always stay very scary to me, even aftet successfully dealing with it in laboratory conditions a few times.
Instant sub. What you're doing is simply amazing. Probably the most incredible DIY project I've seen
This is the best yt channel Ive seen this yr
Only a matter of time before you have your own chip foundry. Probably be making GPU dies before consumers are able to buy them reliably again.
Nah that's gonna take a lot of expensive equipment lol, he just made a thousand transistors on a chip while GPUs have like billions on the same area.
it's not about humongous gpu or something. It's more about custom-made highly specialized diy projects-on-chip instead of microcontrollers on a board, or some custom radio transcievers, something like that. Maybe a single-chip controller of the micro copter (size of a fly). Or if you can include piezo ceramics in it, maybe even some nano robots
@@jskratnyarlathotep8411 if chip making was made readily available like soldering, we would have stuff that used to be thousands of dollars for a couple hundred bucks. Just look at RTL-SDR. A full 100khz to 1.7 ghz receiver for 25$. In the 90s something like this would cost you thousands of dollars and would be the size of an oven.
@@prakharmishra3000 maybe some organic semiconductors could do that, because silicon requires really nasty chemicals
@@jskratnyarlathotep8411 it will be available eventually, in a couple years or decades. You will be able to make your own chips.
Fantastic!
ㅋㅋㅋㅋㅋㅋ 뻘짓님 다음 먹거리입니까?
ㅋㅋㅋㅋㅋㅋㅋㅋㅋㅋ
And, he was never heard from again....
They got 'em!
This video just popped up on my feed at random and wow. It blows my mind how someone can make an IC at home like this. You defiantly know your stuff.
I really want to see you make an ALU or if that is too complex for your IC at least an adder. Some flip-flops (maybe multiple) for basic ram chips would be cool as well. Of course the holy grail would be a (simple) microprocessor but even seeing an adder or flip-flop on a custom IC would be extremely impressive.
High density ROM and RAM were actually much more important than the microprocessor. A 1980s or 1990s personal computer without modern ROM and RAM had been impossible, while the same without a microprocessor would pose no major problems. A fast and simple processor can be built in TTL or one of the compatible CMOS families, just like it was in the 1970s. So ultra cheap kilobytes and megabytes from nano sized transistors were the key for the personal computer revolution, not the microprocessor.
Its easy u can make at home not just ALU bt whole computer.
Watch 8bit computer video series by ben eater
8 bit breadboard computer by ben eater :-
ua-cam.com/video/HyznrdDSSGM/v-deo.html
ua-cam.com/play/PLowKtXNTBypGqImE405J2565dvjafglHU.html
He could just make logic ICs and make a CPU out of those!
This is honestly amazing... the word is out, and we are at the edge of something big. Today, Sam is literally the only person in the world to make chips in his home, but tomorrow, the retrocomputing community is going to jump at the opportunity to remake their old chips, and after that, it's going to be like 3D printing... Can't wait.
An 8086 has over 20,000 transistors.
A 386 has over 200,000
A Pentium 1 has over 3,000,000
Not quite at the edge.
@@Intelwinsbigly we still use single transistors for all kinds of applications like LED lighting, amplifiers, solid state power switching, etc. I think this guy would do well trying to make IGBT's or Silicon Carbide transistors which are enormous but enormously important. Also, the big companies have stopped making RF frequency transistors and won't sell the IP to Rochester.
@@chrisfuller1268 LOL, love how you go on a tangent about everything other than processors, which is what both dj and I were talking about
@@Intelwinsbigly yeah I think an fpga is cheaper
@@misham6547 most definitely.
And just like that, he was never seen again.
Best case scenario: He got a job offer at Nvidia
Worst case scenario: he'll never be heard from again 😔
@@greenrocket23 He founded a company with Jim Keller. I think it's even better than your best case scenario.
Incredible. I worked at Motorola in a fan doing these same processes. You are a very dedicated individual to undertake this lab work. Indeed I will watch your videos.