Hacker News Reader

Comment by moring | Hacker News Reader

Does anyone have an idea whether there is a potential hobbyist market for home-grown chips of this size? I know that I would definitely be willing to spend around $100 for a much smaller chip, like quad-gates or slightly above (1200 transistors is way beyond what I thought is possible in a garage), more the larger the chip gets, and ultimately including getting involved in the design of the chip -- but then of course paying for all failures resulting from that.

The upside, compared to things like Europractice or MOSIS being lower one-time costs and lower turnaround time, but higher per-chip costs, which matches very well with the "hobbyist" target.

I long thought that this would be a nice niche market, but I never found any clue about whether these are realistic numbers or if anyone else would be interested in that.

View on HN

I would guess the market would be pretty fragmented, with only so much demand for each chip. Right now, for example, some people buy FPGA based replacements for Commodore 64 "SID" chips, because the originals do fail. Some amount of purists might prefer a "real" chip. But how many? There are also simpler chips in this space that are in demand. Old Kaypros have a couple of custom gate array chips that are hard to replicate, for example.

Or maybe the retro chip synthesizer crowd would buy a product, but again...how many?

Bad wording on my part. I meant a market for ultra-low-volume fabbing of your own chip designs, not of fixed designs.

Are PLDs still a thing? They can implement most anything a really low density chip like this could.

Perhaps analog ICs would be a better market?

That's was what came to mind for me as well. For digital you simply can't beat the cost of a mass-produced FPGA over having custom silicon fabricated (at small scale). Not to mention the additional design costs, EDA tool licensing, etc.

Now for hardened applications, these large nodes have their uses. But that's a very niche case.

PLDs evolved to FPGAs, which are far more powerful for that price. Homegrown chips would only be interesting for someone who regards ASIC design as an end, not a means, hence hobbyists.

There's also some overlap with PAL and GAL chips, which you can still buy, and might still be acceptable to some hobbyists.

I think the market is larger than we think (at least for a couple more gens mature chips) , but very fragmented.

The analog desinger that wants a few hundred copies of a circuit using matched transistors.

Ferrari willing to spend 10K a piece for a silly automotive chip

I can see a market for very flexible small scale fabs

The security paranoid who want a simple CPU they can trust at an optical scale they can verify (ie imageable to compare with what they ordered)

> 1200 transistors is way beyond what I thought is possible in a garage

what (prior) breakthroughs enabled Zeloof to achieve this?

All the research that was founded by tax dollars in the 70s before all the research was privatised (like nor more scientific publications how it was done etc).

absolutely believe this 100%. and even if isn't publicly funded, under capitalism wage laborers are shorted when they get paid a wage and their code becomes the private property of a capitalist. so yeah, i agree with your framing 100%.

my question was more: what is the short laundry list of the exact developments/research that allowed Zeloof's garage-size/at-home chip-making?

I have a superficial understanding of photolithography, but I'd guess the stuff he's doing is described pretty well in undergraduate textbooks, and he got loads of equipment very cheap as it's obsolete for commercial or even academic use.

Further, capital costs will rapidly start rising as well as the required knowhow as the feature size shrinks, so I think he's already quite close to the limit a one person hobby project can achieve.

All this being said, this is an extremely impressive achievement. There's a long way from reading how stuff is supposed to work, to actually doing it your garage. My hat's off to Mr. Zeloof.

I could see a market if he built this into a small box, like the size of a photocopier.

Reply on news.ycombinator.com