Integrated Circuits Can Be Easy to Understand with the Right Teachers | Hacker News Reader

Integrated Circuits Can Be Easy to Understand with the Right Teachers

hackaday.com

214 pointsszczys26 comments

This video was reasonably good, but still required a lot of background.

If you want really terrific easily understood, yet totally in-depth information about integrated circuits, starting from the physics how semi-conductors actually work at the electron level, all the way up through gates and on to an entire computer, Ben Eater's videos are outstanding.

Most people know his work from the Breadboard computer, https://eater.net/8bit, where he builds an actual functioning computer, with instructions and microcode and output, from scratch on a breadboard, along with terrific explanations.

But that is just the tip of the iceberg. All his videos are amazing, and super easy to understand. Here is his version of how nor gates work: which is way easier to understand to me than this video:

https://www.youtube.com/watch?v=sTu3LwpF6XI

Here is his whole channel:

https://www.youtube.com/channel/UCS0N5baNlQWJCUrhCEo8WlA

and his website:

https://eater.net

That video is great, thanks.

I studied as an EE. I struggled as a freshman to figure out how flip-flops work (building blocks of registers) when it dawned on me while in the shower that they only work due to propagation delay. What an eye openning shower that was...

I feel like some teacher failed you that this wasn't made more explicit.

Eh, it's a subtle point. Unequal propagation delays are how a bistable F-F finds its initial state, but once static operation is established, only the maximum toggle rate depends on propagation delay.

And astable flip-flops usually have separate, explicit timing elements rather than relying on gate charge or whatever.

I think you might be talking about deeper elements of design than the basic fact of delayed feedback through logic gates implementing memory.

I agree. I actually think my instructors were excellent. I had a mental block that needed clearing. My EE instructors went out of their way to spend as much time in office hours as needed to students that needed it. I remember spending several hours in one instructors office trying to figure out where I was going wrong. She was also confused because all of my intermediate work was correct, but I always came up with the wrong answer. Short of it was that I was entering numbers in as reciprocals of what they should have been. E.g. I was entering 1000 in my calculator instead of 1/1000. Took 2 sets of eyes and a couple of hours to spot the problem. Went from barely getting a C at midterms to not losing a point the rest of the semester. Finished with a solid B amd the next semester was perfect. Only ended up there because of my instructor's help and accessibility.

> I remember spending several hours in one instructors office trying to figure out where I was going wrong. She was also confused because all of my intermediate work was correct, but I always came up with the wrong answer

In grad school, I was a TA for some undergrad EE courses. We (the TAs) and professor corrected tests together and whenever we encountered a wrong answer, we'd always replicate the student's work to find where things went wrong so we could a) point it out and b) give partial credit whenever possible. For a class of 120 students, it took four of us two days to get through all of the tests.

I was extremely lucky to have learned ICs from a very special man who helped design/refine the concepts at PARC. Thanks to that, he was able to explain the entire course from a perspective of an "inventor".

This is _very_ similar to the way he taught, although we didn't have 3D representations, they were mostly transparencies taped together and layered. Very cool to see it done this way.

Do tell more

I understood this but I have an EE background. I feel like it would be still quite difficult to follow even for self-taught makers because some concepts were missing entirely, like the basic function of the transistors he's constantly talking about.

IC designer here. Most people struggle to grasp semiconductor circuits because they think about them in terms of voltage drops when in fact they should be thinking about them in terms of current loops. With this approach, the scales are lifted from your eyes.

This is a great insight. It's hard to break the habit of defaulting to the node voltage method when it's been beaten into your head via practice in EE labs. In the lab I used in college, something like ~90% of the ammeters were broken or wrong (mostly due to blown fuses by other clueless undergrads), but the voltmeters were always reliable. Defaulting to the voltmeter also meant you didn't have to move things around on the breadboard. I think this hands-on experience influenced the way I approach circuits even if they are on paper or a computer screen.

Not just about integrated circuits.

I found best teachers are the ones who actually solved real-world problem. Only then they understood which piece of knowledge is helpful because they actually understand it.

Teaching is not for the theorists.

How does a purposefully contaminated shard of glass wield control over electrons?

It's not a shard of glass, it's a slice of silicon monocrystal. These materials are less alike than graphite and diamond.

I think you may be proving his point that bottom-up explanations don't really make it easy to get a high-level understanding of chip features.

It's humorous exaggeration/misstatement with a wink. The same way people often refer to hard disks as "spinning rust".

I appreciate that, though I still feel that a lot of people mistake glass for crystal and vice versa.

Except old hard disks really did use iron oxide as the magnetic storage material. Semiconductors were never made from glass.

Oxygen doped silicon isn't necessarily the same thing as glass but it's not entirely far off either and has been used in semi-conductors. Actual Glass layers are also a major component in transistor and IC design, so I don't feel like this is any more stretched than the rust comment.

way to prove the point.

Neat exhibit! So sad this could be the last Maker Faire https://www.mercurynews.com/2019/05/18/attendees-bittersweet...

Easier, perhaps, I doubt it's easy for anyone, ever. Maybe if you work at some higher abstraction level. But just understanding those bipolar transistors is quite a challenge. You can measure them, but understading why what's measured happens...

Anything is though.

That’s not understanding integrated circuits at all, that seems to be just understanding how a device functions on a chip.

I don't understand your comment. They're describing an Integrated Circuit with several transistors.

Reply on news.ycombinator.com