5 ms·

Assumption: the dream primary value of something like this is the ability for individuals to fab chips on their own. Like 3D printing, it’s for rapid iteration

by mcdow 2y ago

Assumption: the dream primary value of something like this is the ability for individuals to fab chips on their own. Like 3D printing, it’s for rapid iteration in prototyping. Then once you have a design you have one of the big players manufacture it in the traditional manner. If my assumption is true, how is this better than FPGAs?

TeMPOraL 2y ago
> Then once you have a design you have one of the big players manufacture it in the traditional manner. Why? Assuming this is ignoring a good chunk of individual interest. It's similar to people mentioning ordering PCBs instead of making your own: sure, making a thousand copies of a PCB is now cheap enough on the margin to be accessible. But what about making five? Or just one? Not every human sees a hobby as an investment into business. Not everyone does projects with a sellable product in mind. Many just want to test their ideas, have fun, scratch their own itch, build something so it exists, and not to sell it. The primary value of a home fab to me would be to enable fabbing a single task-specific chip (or a tiny amount of them) for any random need I have, whenever it occurs.
- kragen 2y ago
  Every other YouTube video I watch tells me two or three times that Their Sponsor JLCPCB (or, equivalently, their competitor PCBWay) will fabricate five copies of your PCB for US$10 and ship them to you in a week, though the non-promotional price seems to be more like US$80. Tiny Tapeout, MOSIS, and CMP do similar things for chips, at a much higher cost and longer timescale.
- zokier 2y ago
  > It's similar to people mentioning ordering PCBs instead of making your own: sure, making a thousand copies of a PCB is now cheap enough on the margin to be accessible. But what about making five? Or just one? For example JLCPCB is currently offering 5 PCBs (2L, 100x100mm) for $2 + $1.52 shipping. That's why people are saying that making your own PCBs is not economical.
eschneider 2y ago
FPGA is clearly more _practical_ if you're trying to bang out some commercial functionality. Still, making your own chip fab is cool to do in it's own right. :)
- kragen 2y ago
  FPGAs mostly aren't useful for analog.
  - eschneider 2y ago
    That's fair.
  - nrclark 2y ago
    By themselves, no. But FPGAs are widely used in mixed-signal systems with an ADC on one side and a DAC on the other.
    - kragen 2y ago
      Yes, for sure. FPGAs are a great way to move around and crunch large amounts of digitized data. But there are other times where that doesn't help much because you're doing something like a submillimeter analog front end or something.
zokier 2y ago
> If my assumption is true, how is this better than FPGAs? For starters, you can make analog/mixed-signal chips
marcosdumay 2y ago
> Then once you have a design you have one of the big players manufacture it in the traditional manner. That doesn't even work very well with 3D printing. You have no chance at all of transferring something from 10µm chips into a commercial fab.
- kragen 2y ago
  Not the actual chip layout, no, but you can very plausibly get the circuit design to work. If you're doing this you presumably want something you can't do on an FPGA, which probably means analog, so you probably don't want the super-high-end process nodes like 10nm, 7nm, or 4nm anyway; those are generally not going to be very useful for analog signals. Right now Tiny Tapeout seems to be the best option.
koeng 2y ago
Analog I guess. I'm trying to make a chip for DNA synthesis, and so need physical contact with the real world, with electrodes, where electricity from the circuit will cause localized pH changes, which you can use for precise control of biological reactions. FPGAs can't do that kind of analog work.
- fragmede 2y ago
  Sounds fascinating! What sort of ADC and DAC do you need to do for that where traditional techniques fall short?
  - qwezxcrty 2y ago
    maybe the actual electrodes? They may need to be microfabricated?
    - koeng 2y ago
      It is the actual electrodes. They need to be post-fabricated with gold or platinum. The typical aluminum isn't compatible with DNA synthesis chemistry.
- nrclark 2y ago
  It's true that most FPGAs have limited built-in analog capabilities. But good DACs and ADCs aren't too expensive, and an FPGA can control them with exceptional precision. Does your process have some kind of input / output that can't be handled that way? Dedicated DACs/ADCs will almost always offer better performance than the ones you'd find on a microcontroller or even an ASIC.
  - koeng 2y ago
    You don't even need DACs or ADCs, you just need an SRAM cell leaking current through an electrode. The process is entirely separate from that. By analog, I don't mean the signal, I mean the chip is interacting with the world physically. I need the electrodes physically touching chemicals. Those chemicals are incompatible with the aluminum or other normal metals on those connections - you pretty much need platinum or the metal from the chip will screw up the chemical reaction. You also need large amount of input/output - a good start on a chip would be about 1,000 to 10,000 electrodes. I think it is going to be difficult to put that many on an FPGA.

by mcdow 2y ago

TeMPOraL 2y ago

> Then once you have a design you have one of the big players manufacture it in the traditional manner. Why? Assuming this is ignoring a good chunk of individual interest. It's similar to people mentioning ordering PCBs instead of making your own: sure, making a thousand copies of a PCB is now cheap enough on the margin to be accessible. But what about making five? Or just one? Not every human sees a hobby as an investment into business. Not everyone does projects with a sellable product in mind. Many just want to test their ideas, have fun, scratch their own itch, build something so it exists, and not to sell it. The primary value of a home fab to me would be to enable fabbing a single task-specific chip (or a tiny amount of them) for any random need I have, whenever it occurs.

kragen 2y ago
Every other YouTube video I watch tells me two or three times that Their Sponsor JLCPCB (or, equivalently, their competitor PCBWay) will fabricate five copies of your PCB for US$10 and ship them to you in a week, though the non-promotional price seems to be more like US$80. Tiny Tapeout, MOSIS, and CMP do similar things for chips, at a much higher cost and longer timescale.
zokier 2y ago
> It's similar to people mentioning ordering PCBs instead of making your own: sure, making a thousand copies of a PCB is now cheap enough on the margin to be accessible. But what about making five? Or just one? For example JLCPCB is currently offering 5 PCBs (2L, 100x100mm) for $2 + $1.52 shipping. That's why people are saying that making your own PCBs is not economical.

eschneider 2y ago

FPGA is clearly more _practical_ if you're trying to bang out some commercial functionality. Still, making your own chip fab is cool to do in it's own right. :)

kragen 2y ago
FPGAs mostly aren't useful for analog.
- eschneider 2y ago
  That's fair.
- nrclark 2y ago
  By themselves, no. But FPGAs are widely used in mixed-signal systems with an ADC on one side and a DAC on the other.
  - kragen 2y ago
    Yes, for sure. FPGAs are a great way to move around and crunch large amounts of digitized data. But there are other times where that doesn't help much because you're doing something like a submillimeter analog front end or something.

zokier 2y ago

> If my assumption is true, how is this better than FPGAs? For starters, you can make analog/mixed-signal chips

marcosdumay 2y ago

> Then once you have a design you have one of the big players manufacture it in the traditional manner. That doesn't even work very well with 3D printing. You have no chance at all of transferring something from 10µm chips into a commercial fab.

kragen 2y ago
Not the actual chip layout, no, but you can very plausibly get the circuit design to work. If you're doing this you presumably want something you can't do on an FPGA, which probably means analog, so you probably don't want the super-high-end process nodes like 10nm, 7nm, or 4nm anyway; those are generally not going to be very useful for analog signals. Right now Tiny Tapeout seems to be the best option.

koeng 2y ago

Analog I guess. I'm trying to make a chip for DNA synthesis, and so need physical contact with the real world, with electrodes, where electricity from the circuit will cause localized pH changes, which you can use for precise control of biological reactions. FPGAs can't do that kind of analog work.

fragmede 2y ago
Sounds fascinating! What sort of ADC and DAC do you need to do for that where traditional techniques fall short?
- qwezxcrty 2y ago
  maybe the actual electrodes? They may need to be microfabricated?
  - koeng 2y ago
    It is the actual electrodes. They need to be post-fabricated with gold or platinum. The typical aluminum isn't compatible with DNA synthesis chemistry.
nrclark 2y ago
It's true that most FPGAs have limited built-in analog capabilities. But good DACs and ADCs aren't too expensive, and an FPGA can control them with exceptional precision. Does your process have some kind of input / output that can't be handled that way? Dedicated DACs/ADCs will almost always offer better performance than the ones you'd find on a microcontroller or even an ASIC.
- koeng 2y ago
  You don't even need DACs or ADCs, you just need an SRAM cell leaking current through an electrode. The process is entirely separate from that. By analog, I don't mean the signal, I mean the chip is interacting with the world physically. I need the electrodes physically touching chemicals. Those chemicals are incompatible with the aluminum or other normal metals on those connections - you pretty much need platinum or the metal from the chip will screw up the chemical reaction. You also need large amount of input/output - a good start on a chip would be about 1,000 to 10,000 electrodes. I think it is going to be difficult to put that many on an FPGA.