How one might make logic devices with cnc, 3d print, or laser?

AwesomeCronk

I am wanting to make homemade ICs. I know this is not feasable for years at least, but you plan now for the future.

Anyone in the experience of doping silicon and creating logic, I would like your input.
I want to create logic gates, at most 1/8" across. I want to know how, without mechanical components. Can I do it by 3d-printing conductive filament in special patterns? Do I have to use silicon doping? I don't really care about power consumtion, but 3.3v to 5v logic would be nice.

Yes, I know I could just use an fpga, but I want to eventually make the gates themselves.
Thanks for any input!

jmg

recently posted was this thread...
http://forums.parallax.com/discussion/170596/home-made-silicon-ic

That is a candidate for highest cost per gate, on the planet... (but he has some nice kit there..)

AwesomeCronk

Yep, that was a portion of the inspiration!

DigitalBob

I couldn't see the video in that other post. Was it Sam Zeloof stuff he makes some homemade I/C's

Phil Pilgrim (PhiPi)

If you have a laser cutter, why not try fluidics?

-Phil

Genetix

AwesomeCronk,

Rather than trying to make your own chips, why not build a few gates, flip-flops, and other simple logic out of discrete components.
You can then measure voltages and currents as well as watch the signals on an oscilloscope to see how they work.

If you want a real challenge try building something like an 8-bit shift register out of discrete components.

Peter Jakacki

AwesomeCronk wrote: »

I am wanting to make homemade ICs.

While you may be "inspired" by the youtuber who must have spent many tens of thousands of dollars at the very least, even if some stuff was off eBay, he still put in a lot of effort to learn by doing. That is a good starting point for anyone who is "wanting to" do anything in that you should at least do something yourself such as read books, articles, and watch videos and appreciate just how difficult it is to build even primitive and impractical chips that this guy did.

He began thinking about how to make chips as his “way of trying to learn what’s going on inside semiconductors and transistors. I started reading old books and old patents because the newer books explain processes that require very expensive equipment".

Each chip he made can do something that a ten cent 555 timer could do ten times better. But that wasn't the point, he was learning by doing, and learning was the point.

Something as simple as making single sided pcbs isn't easy, but it's do'able. Making double-sided plated through pcbs at home is not really do'able, unless of course you setup a production facility in similar manner to what the youtuber did, but that didn't look like it was at "home" (I believe he took over his rich doting parent's garage). If pcbs are hard to do then chips are probably 100 times harder to do.

There are plenty of "wanna do's", there are plenty of "gonna do's", there are plenty of "you show me do's", but there are very very few "just do's". So if you need someone to "show" you by walking and carrying you through every painful step, millimeter by millimeter, then that's not doing or what this guy did. No, he just did.

But by all means learn, starting at the bottom with simple logic circuits built out of discrete transistors, then maybe 7400 logic chips etc. Learning requires effort but learning is fun and rewarding.

frank freedman

This is probably one of the first chips you may build by hand if you want to get to the level of chip inards. https://shop.evilmadscientist.com/productsmenu/tinykitlist/652.

Or for analog, look for DIY opamps.

AwesomeCronk

Okay then! I’ll just keep working with FPGAs then.

The_Master

Perhaps it would be possible to do a limited version of a homemade IC that would be just essentially several point-contact transistors on a single piece of germanium. No doping involved, not even sputtering. Maybe deposit metal somehow, than use a razor blade on it. Just thinking out loud.

Phil Pilgrim (PhiPi) wrote: »

fluidics
-Phil

This sound interesting. What is this?

AwesomeCronk

Fluidics is the use of fluids inside microchannels or tubing to create logic systems. I looked at pneumatics for this, but pneumatic-driven pneumatic valves are EXPENSIVE!!

jmg

AwesomeCronk wrote: »

Okay then! I’ll just keep working with FPGAs then.

You could also look at the SiLego parts, which are small and combine Logic and Analog, or if the Logic speed is modest, you can use Boolean opcodes inside a MCU... a 20~30c MCU or another, $6.40, on a PCB, with pgm/debug

Phil Pilgrim (PhiPi)

AwesomeCronk wrote:

Fluidics is the use of fluids inside microchannels or tubing to create logic systems. I looked at pneumatics for this, but pneumatic-driven pneumatic valves are EXPENSIVE!!

You wouldn't use mechanical valves except, perhaps, to convert electrical signals to fluid flow on the inputs. All internal switching is done by the shapes of the microchannels and where they come together and exit. For example, here is a site that illustrates some simple fluidic logic:

https://www.symscape.com/blog/fluidic-logic

-Phil

AwesomeCronk

Yeah that’s beyond me.
On the bright side, the DueProLogic works well if support can get back to you within two months!

Genetix

Speaking of fluidics, I have always been curious as to how pneumatic or air-operated logic works.

The dentist's drill is pneumatic or driven by air.

frank freedman

FPGAs and other programmable devices ar about as close to custom silicon as you will get on the average (well more than averge) budget. The first programmable logic I used was a i1702 eprom and then the 82S100 PLA from Signetics. HIgh school student budget required friendly sales person at Hamilton Avnet at the time to program it. But if you want to play at that level, one good book is "Digital Logic and State Machine Design" by David Comer. There are many good concepts presented in there for designing low level logic. This book has been around for a long time. Haven't had time to really play with FPGA beyond the harware equivalent of "Hello World", but seems same as of old but on a larger scale. Anyone have a part number for more time in a day or even a week?

JRoark

I bet you could have some mischief with a point-contact of iron oxide on a galena base. Add in a couple of silicate insulating layers (maybe via sodium silicate solution?) and you’ve got the basics of a primitive semiconductor gate. This low tech stuff could be big fun on a rainy day.

Phil Pilgrim (PhiPi)

This Scientific American article from 1964 is an excellent introduction to fluidics:

http://miriam-english.org/files/fluidics/FluidControlDevices.html

Given a pattern, such devices could easily be made from laser-etched layered acrylic. Creating the pattern is the hard part, since some knowledge of fluid dynamics is requisite.

-Phil

Clock Loop

“Licensed under the TAPR Open Hardware License (www.tapr.org/OHL)

Since we need speed.

Fluidics; move a cavity filled with....   my suggestion,  silicon gel.

Each junction or function must be fluidically formed for the desired output.

Lasers are fast so lets combine the two.

Lasers of different wavelength, but close, and selected via the silicon gel's optical spectrum, channeled through paths filled with this silicon gel, each function or junction is fluidically SETUP with ultrasonics.
The ultrasonic part is just the programming of the junction and thus whole chip a.k.a. an fpga LIKE chip.

This works by fractal structures being made in the silicon gel with ultrasonic vibrations creating cavitation in the silicon gel leaving fractal voids that change the way light propagates throughout the material in range of the junction.

This ultrasonically configurable logic optical circuit should be built on a laser table.

Licensed under the TAPR Open Hardware License (www.tapr.org/OHL)"

Clock Loop

Material vibration.



And...

Similar effects of Sonoluminescence



Plus the experiments on microwavelength beam splitting using silicon fractal junctions.



And there you have it.

Its all about tuning, its a silicon computer / guitar.

“Licensed under the TAPR Open Hardware License (www.tapr.org/OHL)

Did you know you can also read and write data to a hard drive platter(magnetic) using lasers?   

The photo-electric effect proved that.

Its done by sensing the laser emitter junctions, or (a function junction made in device mentioned in the post above) energy potential during a write.   
With that information you can tell if you wrote to a spot that had a 1 or a 0.

Licensed under the TAPR Open Hardware License (www.tapr.org/OHL)"

Its the same reason an led can sense the light level in a room, due to the diode junction voltage falloff rate.

Cluso99

The old Xilinx FPGA (XC28164 IIRC) and the software (around 1990) allowed you to manually route the FPGA. I did this in a project to connect to the main bus of an ICL Minicomputer. So you could actually route the logic within the FPGA.
These days, while you can declare a logic diagram, you have no idea how the software turns this into logic inside the FPGA.

jmg

Cluso99 wrote: »

The old Xilinx FPGA (XC28164 IIRC) and the software (around 1990) allowed you to manually route the FPGA. I did this in a project to connect to the main bus of an ICL Minicomputer. So you could actually route the logic within the FPGA.
These days, while you can declare a logic diagram, you have no idea how the software turns this into logic inside the FPGA.

Yes, the days of 'seeing' and manually routing what was inside a leading edge FPGA are long gone.

The SiLego parts are at the smaller end of programmable logic, but they do have a visual schematic entry, that then creates cross-point fuse tables.

They used to do only One time programmable QFN parts, (tho you can iterate a socketed device using the RAM) and they now have Multi Time pgm and TSSOP packages, as well as DIP header boards, so they are easier for students and learning.

https://www.dialog-semiconductor.com/sites/default/files/greenpak_brochure.pdf

A small MCU is more flexible, but the SiLego parts are Logic, so avoid the one-place-at-a-time software timing drawbacks of little MCUs.

AwesomeCronk

Given that this discussion has moved to fpgas, has anyone used the DueProLogic successfully?
Maybe the moderators can split this.

Discovery

Something to think about. Years ago in college electrical engineering laboratory we students were building operational amplifiers from discrete components: transistors, resistors, and capacitors. It was not easy because the internal properties of discrete transistors vary significantly from one transistor to another even though the transistors were made by the same manufacturer, in the same lot, on the same day. Lots of compensation circuits were required to get the op amps to function. The exercise became easier when it was possible to buy several identical transistors in the same package.

Discovery

Cluso99

yes, the spread in transistor characteristics are enormous when you think of how tiny they are and mass produced. We forget about this when using digital ICs.