This 22-year-old builds chips in his parents\u2019 garage

[Jan Panteltje]

This 22-year-old builds chips in his parents\u2019 garage
https://arstechnica.com/information-technology/2022/01/this-22-year-old-builds-chips-in-his-parents-garage/

more than 1000 transistors on it...

[Rick C]

I assume it was the writer who mucked up the part about, "a form of crystalline silicon known as polysilicon". Polysilicon is the opposite of crystalline silicon, hence the "poly".

--

Rick C.

- Get 1,000 miles of free Supercharging
- Tesla referral code - https://ts.la/richard11209

[Jan Panteltje]

On a sunny day (Sat, 22 Jan 2022 10:08:49 -0800 (PST)) it happened Rick C
<gnuarm.del...@gmail.com> wrote in
<8a2d8da4-0a9f-4633...@googlegroups.com>:

>On Saturday, January 22, 2022 at 9:59:47 AM UTC-5, Jan Panteltje wrote:
>> This 22-year-old builds chips in his parents\u2019 garage
>> https://arstechnica.com/information-technology/2022/01/this-22-year-old-builds-chips-in-his-parents-garage/
>>
>> more than 1000 transistors on it...
>
>I assume it was the writer who mucked up the part about, "a form of crystalline silicon known as polysilicon". Polysilicon is
>the opposite of crystalline silicon, hence the "poly".

Quite possible, I do not know enough about that to make a judgement
I like his projector into a microscope idea.
Been thinking about using a DLP projector to project a pattern on photo-PCbs, people have done that,

There also is this guy who inspired me to buy the super cooler, he build, among other things, his own electron microscope.
https://www.youtube.com/watch?v=VdjYVF4a6iU
https://www.youtube.com/watch?v=7PWESWqhD8s

[whit3rd]

Only claims 100 on a chip, but the metallization doesn't connect to that many; I'm not sure how complex
the circuitry can be, but it looks like a rather simpler circuit, like a differential pair and some
current mirrors. It could hold an op amp, that uses external bias resistors (don't see
any sign of FET or CMOS or both NPN and PNP transistor types).

His basic idea, it seems, is that small-scale production is possible and instructive.
Imagine a six-man shop that can produce integrated devices with quick turnaround.
Would that be a useful addition to a maze of megacorporations with ever-changing
off-the-shelf designs, big plants, and long turnaround, but cryptic spec sheets
and big foundry toolchain complexities (and surcharges)?

For prototypes, heck yes. Bill Sloman can sell 'em some e-beam litho tools.

[Dimiter_Popoff]

Would be awesome if we could have chips made to order the way we have
PCB-s nowadays... Bill said the litho tool got lost but well, some
knowledge on how it was made must have remained.

[jla...@highlandsniptechnology.com]

On Sat, 22 Jan 2022 22:26:25 +0200, Dimiter_Popoff <d...@tgi-sci.com>
wrote:

PCBs use a lot of inkjet printing nowadays. I'd expect that a crude IC
process could be developed around inkjet.



--

I yam what I yam - Popeye

[Dimiter_Popoff]

Well I mean top or nearly top technology PCB-s, like 4 mil multiple
layers 0.1mm drills, I suppose the equivalent of that today for silicon
would be within the tens of nm. One would not be held hostage by the
whims of some marketeers deciding which processor is to have a future...

I have not been making crude PCB-s since my 20-s, back then I did some
using permanent markers etc. Helped learning the trade I suppose.

[jla...@highlandsniptechnology.com]

On Sun, 23 Jan 2022 00:18:28 +0200, Dimiter_Popoff <d...@tgi-sci.com>

I think that some of the earliest ICs were patterned with silk
screens. So were PCBs.

[Dimiter_Popoff]

I remember the day when they took a photograph of what the PCB would be
and then used the film like they now do after photoplotting (well, the
mechanical photoplotters of some 30 years ago are gone I suppose, the
Gerber format meant for them is pretty much alive though (back then
I could implement it but I had no data about the excellon format
for the drilling machines; I got hold of a punched tape and snooped
it from it.. :-).
Before that my first CPU board (early 80-s, I was in my late 20-s,
came later to the party) was photographed on that whole-room camera,
no plots yet. It took me months to "draw" it (actually I used
thin red scotch tapes and stuck them on a plastic transparent millimeter
"paper"). I was working as a turner back then, was convenient for
cutting the scotch tapes to say 1mm or may be I went below that
(scale for photo was 2:1), can't remember. Here is what remains of
that first CPU board of mine, must be still somewhere around:
http://tgi-sci.com/misc/grany09.gif

[Jasen Betts]

On 2022-01-22, Rick C <gnuarm.del...@gmail.com> wrote:
> On Saturday, January 22, 2022 at 9:59:47 AM UTC-5, Jan Panteltje wrote:
>> This 22-year-old builds chips in his parents\u2019 garage
>> https://arstechnica.com/information-technology/2022/01/this-22-year-old-builds-chips-in-his-parents-garage/
>>
>> more than 1000 transistors on it...
>
> I assume it was the writer who mucked up the part about, "a form of
> crystalline silicon known as polysilicon". Polysilicon is the
> opposite of crystalline silicon, hence the "poly".

It's not amorphous (which is a true opposite solid state to crystaline)
It's polycrystaline. So it's still crystalised silicon.
It's significantly cheaper than monocrystaline silicon.



--
Jasen.

[Jan Panteltje]

On a sunny day (Sat, 22 Jan 2022 12:06:14 -0800 (PST)) it happened whit3rd
<whi...@gmail.com> wrote in
<0cf15665-e706-4327...@googlegroups.com>:

>On Saturday, January 22, 2022 at 6:59:47 AM UTC-8, Jan Panteltje wrote:
>> This 22-year-old builds chips in his parents\u2019 garage
>> https://arstechnica.com/information-technology/2022/01/this-22-year-old-builds-chips-in-his-parents-garage/
>>
>> more than 1000 transistors on it...
>
>Only claims 100 on a chip,

<quote>
Then his homemade photolithography machine beamed on his design: a grid of 12 circuits,
each with 100 transistors (and a dancing bear), 1,200 transistors in all.
<end quote>

so 100 per circuit, 12 circuits on a 'wafer' ?

but the metallization doesn't connect to that many; I'm not sure how complex
>the circuitry can be, but it looks like a rather simpler circuit, like a differential pair and some
>current mirrors. It could hold an op amp, that uses external bias resistors (don't see
>any sign of FET or CMOS or both NPN and PNP transistor types).
>
>His basic idea, it seems, is that small-scale production is possible and instructive.
>Imagine a six-man shop that can produce integrated devices with quick turnaround.
>Would that be a useful addition to a maze of megacorporations with ever-changing
>off-the-shelf designs, big plants, and long turnaround, but cryptic spec sheets
>and big foundry toolchain complexities (and surcharges)?
>
>For prototypes, heck yes. Bill Sloman can sell 'em some e-beam litho tools.

He is now 22, wait a few more years what else he comes up with.
And then China gets into it
Maybe your own chip-making machine on ebay for $999.99 .....
:-)

[Jan Panteltje]

On a sunny day (Sun, 23 Jan 2022 01:06:30 +0200) it happened Dimiter_Popoff
<d...@tgi-sci.com> wrote in <ssi2lo$r14$1...@dont-email.me>:

>I remember the day when they took a photograph of what the PCB would be
>and then used the film like they now do after photoplotting (well, the
>mechanical photoplotters of some 30 years ago are gone I suppose, the
>Gerber format meant for them is pretty much alive though (back then
>I could implement it but I had no data about the excellon format
>for the drilling machines; I got hold of a punched tape and snooped
>it from it.. :-).
>Before that my first CPU board (early 80-s, I was in my late 20-s,
>came later to the party) was photographed on that whole-room camera,
>no plots yet. It took me months to "draw" it (actually I used
>thin red scotch tapes and stuck them on a plastic transparent millimeter
>"paper"). I was working as a turner back then, was convenient for
>cutting the scotch tapes to say 1mm or may be I went below that
>(scale for photo was 2:1), can't remember. Here is what remains of
>that first CPU board of mine, must be still somewhere around:
>http://tgi-sci.com/misc/grany09.gif

This was much faster, one day torn around:
http://panteltje.com/pub/z80/graphics_card_top.jpg
http://panteltje.com/pub/z80/graphics_card_bottom.jpg

http://panteltje.com/pub/z80/sound_card_bottom.jpg
http://panteltje.com/pub/z80/soundcard_top.jpg

http://panteltje.com/panteltje/z80/system14/diagrams/index.html
all done in the evenings and weekends in the eighties.

concepts use peeseebee why
!

[Dimiter_Popoff]

My first graphics card for this system was made using the same
wiring technology only I had got hold of some teflon wire, 0.5
or 0.7mm external diameter and the wiring was on the bottom side.
Teflon made it much easier, no burnt insulation while the soldering
tip could just go anywhere. Was a proud 320x240 pixels, 2 bits per
pixel - on a mono screen it was just enough to write my first graphics
editor and use it to design the next CPU board which got already
photoplotted (it was still a 6809 one though).
I think on that graphics card I had even put a pallete RAM in the
form of a 74170.... Two bits in, was it 4 bits out, I don't remember.

[Jan Panteltje]

On a sunny day (Sun, 23 Jan 2022 14:01:40 +0200) it happened Dimiter_Popoff
<d...@tgi-sci.com> wrote in <ssjg37$amt$1...@dont-email.me>:

stripped flatcable still works for me to this day, about 1 GHz board:
http://panteltje.com/pub/raspberry_pi_datv_transmitter_test_setup_IMG_3937.JPG
http://panteltje.com/pub/test_board_wiring_side_IMG_3921.GIF
if you look close on the right you see the RF part is not with flat cable
note all the SMDs on the big chip adaptor.
http://panteltje.com/pub/xdipo_locked_to_raspberry_pi_datv_transmitter_IMG_3938.JPG
who needs peeseebees ;-)

[boB]

Very impressive !

At 22 years old, how does he pay for all that nice stuff he's got ?
The HP gear and microscope, etc ?

Somehow I don't think he has to have a day job unless he just does
this in his spare time.

boB

[Anthony William Sloman]

Not any longer. I only worked for Cambridge Instruments from 1983 to 1991. When I last saw an electron beam microfabricator it was at the University of New South Wales (being used by their quantum computer team) and it was a Raith machine.

https://raith.com/electron-beam-lithography-systems/

Whose electron microscope column or column they actually use isn't spelled out on the website.

The machines that Cambridge Instruments sold went for about a million dollars each back then. It's not a high volume product, and it does require a lot of precision engineering, so the price won't have gone down all that much. Not really startup in a garage material. Class 10 clean areas aren't cheap either.

--
Bill Sloman, Sydney

[wmartin]

Very cool! In California, he would likely be jailed for having "gasp!"
chemicals in his garage! Certainly must be making bombs...

[three_jeeps]

He is a CMU student. Knowing first hand the amount of work that gets thrown at ECE students, I am surprised he has a lot of time to devote to his hobby.
Yea, I wonder about the $s as well.
J

[Jan Panteltje]

On a sunny day (Mon, 24 Jan 2022 11:02:11 -0800) it happened wmartin
<w...@wwmartin.net> wrote in <ssmt3l$3g8$1...@dont-email.me>:

Yes this is a common problem with society these days, even kitchen knives are forbidden?
It will all change after the next nu-culear war and making sausages on the hot plutonium.

[Brent Locher]

On Monday, January 24, 2022 at 2:07:35 AM UTC-5, boB wrote:

A rich parent who is with it will be far happier seeing his money go into scopes and electronics vs drugs and rehab. A stint in rehab for kids of wealthy parents will cost more than his lab setup

[Jan Panteltje]

On a sunny day (Wed, 26 Jan 2022 06:38:27 -0800 (PST)) it happened Brent
Locher <blo...@columbus.rr.com> wrote in
<2f68efff-30f9-470e...@googlegroups.com>:

At 22 I was working and accumulating all sort of electronics at home... rented bedsit in other city than my parents.
Build a TV and designed and build a vidicon TV camera... 1968
That landed me a job at the national TV network the next year, and more electronics accumulated at home.
Addiction to electronics perhaps.
Nothing much changed to this day :-)

[Cydrome Leader]

It's even more of a CA problem. My favorite is how it's illegal to use 100
or even 99% isopropyl alcohol there. You're seriously supposed to cut it
to 80% (or something like that) or less with water or acetone. They
actually fine companies for selling high grade isopropyl alcohol there
too. Clearly all other problems been solved.

[Anthony William Sloman]

On Friday, January 28, 2022 at 9:09:08 AM UTC+11, Cydrome Leader wrote:
> Jan Panteltje <pNaonSt...@yahoo.com> wrote:
> > On a sunny day (Mon, 24 Jan 2022 11:02:11 -0800) it happened wmartin <w...@wwmartin.net> wrote in <ssmt3l$3g8$1...@dont-email.me>:
> >>On 1/22/22 06:58, Jan Panteltje wrote:

<snip>

> It's even more of a CA problem. My favorite is how it's illegal to use 100
> or even 99% isopropyl alcohol there. You're seriously supposed to cut it
> to 80% (or something like that) or less with water or acetone. They
> actually fine companies for selling high grade isopropyl alcohol there
> too. Clearly all other problems been solved.

Sounds bizarre. Analytical grade isopropyl alcohol has to be 99% pure (or better), and chemical supply companies have to be able to ship it to chemical laboratories or fine chemical manufacturing plants. What you local pharmacist can sell over the counter may be more restricted.

Are you sure that you know what you are talking about?

--
Bill Sloman, Sydney

[Tom Del Rosso]

How can he connect it? Did he get a wire bond machine from ebay too?

The wire bond machines have always been the part of the process that I
find most impressive, because they combine precision with fast moving
parts.


--
Defund the Thought Police

[Cydrome Leader]

Yeah, I do.

top matchs are all fines or settlements for selling standard products for
industry in the shithole known as CA

https://ww2.arb.ca.gov/search/site?keys=isopropyl

[Rick C]

You mean like pick and place machines? They may not have quite as much accuracy, but they are certainly in the same ball park. I see mentions of 0.001" or 25 um. How accurate are wire bonding machines? I see one mention of 2.5 um. So maybe 10x?

Pick and Place has to maintain these numbers over distances ranging towards a meter. Wire bonding only needs to operate over a range of a cm or so. I know wire bonding is very fast, but for pick and place the long distance and the weight of the object being placed is much more in some cases. Handling a wide range of sizes is not easy either. I wonder just how long they maintain these specs as they wear?

I find it interesting to watch a pick and place operate. One I saw had multiple heads so multiple pieces could be picked up with less movement while being placed.

--

Rick C.

+ Get 1,000 miles of free Supercharging
+ Tesla referral code - https://ts.la/richard11209

[Cydrome Leader]

A manually operated wire bonder doesn't look any different from a fancy
optical microscope. I was recently told quite a bit of this work is still
done by hand. The one I saw had no wire instaled, so I didn't get to play
with it for real.