Hacker News new | past | comments | ask | show | jobs | submit login
Decap of a tiny chip that plays Christmas songs (twitter.com/kenshirriff)
137 points by Eduard 30 days ago | hide | past | favorite | 40 comments

Years ago we took one of the happy birthday tune cards and connected it to the floppy drive hinge on a PC, the developer was convinced we had installed a TSR program to play it every time he inserted a floppy, not realising it was a hardware hack. Weeks of fun until he found out.

As an 80s kid, these melody cards were among the first things that piqued my interest for what later lead me to studying electrical engineering / computer architecture and now doing web development.

Funny how small experiences can shape a primary school kid's path. Similar experiences include those mini play-with-your-pinkie pianos, putting a 9V battery on the tongue, and the effects of putting a paperclip into an AC power socket.

> putting a 9V battery on the tongue, and the effects of putting a paperclip into an AC power socket.

Removing insulation from a live phone wire with my teeth (I had decided to put an extension in my room without asking for help...) was one of mine, that definitely made me more careful. Went just fine until I closed the circuit with my tongue by accident just as someone called. Not pleasant.

> Removing insulation from a live phone wire with my teeth (I had decided to put an extension in my room without asking for help...) was one of mine, that definitely made me more careful. Went just fine until I closed the circuit with my tongue by accident

I read this and thought "so what?".

> just as someone called.

Ahh there it is. Yep, that is indeed not pleasant.

Yeah, I'd done it once or twice before, and it went well since nobody called, and so I did not for a moment think about the difference. It helped strongly reinforce to make sure stuff is disconnected, not just "not harmful right this instance", before touching it.

The 90v(?) AC that comes through the phone line to make it ring was intended to be powerful enough to ring a physical bell. Back before the exchange was automated, home telephones had a hand-cranked generator to produce that signal. According to my dad, kids used to take those out of old phones and use them for all kinds of shenanigans. (Like shocking their friends, siblings, and the dog, I'm guessing.)

I'm certain someone will come up with the idea of pairing a smartphone with a taser at some point, and closing the loop on this.

I'd be unsurprised if someone is already selling something similar as a sex toy.

I had basically the same experience. I had messed with phone wires a lot, and they were always harmless, until someone called one day. I messed with them more carefully from then on :P

I am so glad I am not the only one who did this. I was messing around with the phone wires because my sister was hogging the phone and I had important business on a MUD. I found if I put a large resistor across two of the wires (I can't remember which now) the call quality would degrade enough that she would stop and I could get back on the internet.

I did this too! I picked up the wrong wire - I was installing a phone extension for my Nan - and stuck the wire in my mouth to strip it, as you always do as a kid. I was sitting cross-legged on the floor and got "blown" backwards across the room just like in a movie. I'd been shocked a dozen times from the 240V mains due to prior misadventures, but that was the worst shock I'd ever had. I couldn't taste anything for four days afterwards.

>and the effects of putting a paperclip into an AC power socket.

I believe that (in earlier years when electric protections were non existant or just a huge fuse) is one of the origins of survivorship bias ...

I remember the first attic of a house of a certain age to learn that it used to be a thing to not have insulated wiring. Instead, the wiring used a series of insulators as standoffs/guides, but bare exposed copper carrying live current. I was flabbergasted that more homes didn't burn to the ground.

Yep, and at least here (Italy) up the '60's or maybe even '70's many older houses were "retrofitted" with what was called "piattina", which you can see here attached to a switch and to a socket:


it was two wires (no earth/ground) and was fixed using tiny nails that were driven in the plastic slip between the two wires, the "luxury" version were these (on the right):


but the most common ones had not a real insulator, they were simply steel nails dipped in some rubbery paint.

In theory the nail should have never come in contact with any of the two wires, but expecially where there was some corner preventing the nail to be hammered perfectly straight or because of sloppy installation, after a few years (with the rubber/insulation becoming stiff and microcracking) it happened.

Look up “Knob And Tube Wiring” on Wikipedia for details (which claim that the wires should have been insulated, but I believe I’ve seen the same as you report, in a California house built in the 1950s).

Without wanting to be that guy, I’m really glad I grew up in a country with shuttered power sockets.

LOL, so did I. That does not stop a 9-year-old child from putting a paper clip there, I can promise you. The 9-year-old brain is ideally-developed for defeating this type of precaution.

Eventually I just found it easier to attach wires to a plug, put the plug in the socket and then just short the wires together.

What resources are there for learning about how to read die shots? I'm surprised that one is able to glean so much information from it.

The author of the tweet (Ken Shirriff) is particularly good at explaining things. A couple of his posts showing that:



I focus on 1970s and 1980s chips, which are simple enough to understand. For that generation, Mead and Conway's Introduction to VLSI Systems is a good place to start. https://ai.eecs.umich.edu/people/conway/VLSI/VLSIText/PP-V2/... Old books on chip design are also useful, and can usually be obtained for cheap. Also see The Layman's Guide to IC Reverse Engineering: http://siliconzoo.org/tutorial.html

I'm afraid, none.

It takes one to make chip, to read one later

What are some top schools for getting into chip manufacturing? Is this the right way at looking about getting into the field?

I'm afraid none in the West. Speaking, and writing Taiwanese Chinese is something really needed to get onto the cutting edge of device fabrication.

An average US college will tell you about VLSI, logic, teach you Cadence, and designing something on the level of 8051, and process engineering enough to work on something from 350-180nm era with aluminium interconnect.

The problem is this is hardly enough to make you employable. And without employment in few serious semi companies left standing today, you will never see modern chip manufacturing.

Semi industry has really high entry barriers, and it has really poor remuneration. The industry largely coasting on greying cadres from eighties-nineties in the West, and very cheap PhDs in Asia.

> I'm afraid none in the West. Speaking, and writing Taiwanese Chinese is something really needed to get onto the cutting edge of device fabrication.

You should take a look at http://opencircuitdesign.com/ You don't need "to get onto the cutting edge of device fabrication". You need to search for a VLSI design class.

A VLSI class, even from an average university will contribute very little to your employability. As I said, Big Semi barely if ever employs entry level professionals. And besides Big Semi, there is really very few companies left.

Even startups <$1m are full of seasoned professionals coming from abroad, seeking PhDs in US universities. I think it's more characteristic for semiconductor startups to hire much more senior engineers than the industry average to convince their investors, not the other way around.

Nah, sorry that is just not true.

Try this one for the US: https://sunypoly.edu/research/albany-nanotech-complex.html

Or this one for Europe: https://www.imec-int.com/en

Both world-leading semiconductor technology research institutes (coincidentally both got the first EUV-tools) with plenty of students.

I know IMEC as I was thinking going there at some point.

Both IMEC, and SUNY are heavily graduate research oriented. There are some undergrad opportunities there, but one's chances to get there are probably less than getting into MIT.

The "death valley" in between graduation, first employment, and a first real substantial job in semi industry is super-duper real, and acute problem. It's a problem in Asia, and 10x of than in US.

I don't know, just apply?

Obviously it is important to get hands on experience even as an undergrad. Ideally you can do that at your university, otherwise there are plenty of opportunities for internships. At least in Europe I had absolutely no problem finding one. It can't hurt to also look for second tier companies.

I largely quit trying to get into semi 10 years ago because I was was able to get better salary, and quicker just by pumping wordpress websites. And not like 10%-20% bigger, but few times as much.

My first full time job in line of "make us a WP website" was CAD $65k. An intern stipend at a semi lab was $16k cash a year, with a very competitive recruitment like "show us you was in top 10% of your group"

I mean, what do you mean by “chip manufacturing”? I’m sure there are plenty of people designing chips at Intel/AMD/Nvidia/Apple/etc. that are good at reading die shots, so the real question might be, what career path will get you to work closely with these people and learn from them.

Unless you actually mean working at a fab? Which strikes me as more process engineering than typical HN interests.

I did mean chip design in my question. I‘ll (partly) admit to being a helicopter parent, so I am just curious about various career paths for a kid who shows interest in electronics, and what one particular career path (i.e. chip design) entails these days.

I’m also very aware the epicentres of “top schools” will likely shift in the next 15-20 years, that kids’ interests and talents are labile, and that true self-determination is the key to a successful and happy life.

Got it. Somewhat non-intuitively from the outside, hardware is far less glamorous and lucrative than software. If you really like hardware, then software that drives hardware is the place to be — you still get to play with and learn about the hardware, but you’re not in the mines, as it were. And — big one — your skills largely transfer.

At any rate, if it’s not common knowledge these days, get the kid a cheap second-hand bench-top oscilloscope. As long as the basic functions work, it doesn’t matter if it’s from the 1980s. But that thing is your eyes. And the knobs are fun.

Thanks for your take on this!

Your comment triggered warm memories of playing with my grandfather’s Soviet oscilloscope as a kid. It had knobs so clicky that you’d feel them your knuckles! :)

I am conflicted on this myself. Even most basic coding keeps being more profitable than EE.

I went to live, and make money in the mainland China for a few years. It was a hard bargain with the conscience.

I simply cannot fathom how the industry can be done in the West now, and the labour price has nothing to do with that for the reason it's higher in China for actual, non-software engineering.

An average semi engineering course in US/Canada will teach you a bit of everything, including physical device manufacturing.

India is much more VLSI, and applied skills heavy (Cadence,) thus they dominate there.

None of US multinationals will hire a fresh "average" US semi engineering graduate. Maybe MIT + 1 more top schools, but otherwise fresh grads have no chances on the job market against already experienced people coming from Asia to do their PhDs in US.

Right! My point is that they will hire fresh software engineers/other disciplines, and then with some networking and being on an adjacent team, you get access to informal learning opportunities and internal transfer opportunities, so that over the longer arc of your career you can get into whatever field you want.

Interesting. There is a chip that was once one of my friend's favorites for his consulting work... it was essentially a 6502 with huge mask-programmable ROM and a tiny (like 128 byte) RAM -- it was designed for greeting cards, where if you bought 100K+ quantities of passivated, unencapsulated die, they cost between 1 and 2 cents.

> I dissolved the transistor-like package in boiling sulfuric acid

Well, well. Here's something cool to add to my list of home lab tools!

Now someone needs to code an emulator for it.

I was expecting Doom. :/

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact