Hacker News Reader

Comment by Animats | Hacker News Reader

If engineering isn't near the factory, it's not as effective.

Here's one of the most generic electronic components - a 1K resistor.[1] These sell for about US$0.0015 each. DigiKey has a list of many suppliers.

There are a few old-line US resistor makers in there, including Bourns and Ohmite. They're price competitive with Chinese companies. But when you look up their engineering job locations, none are in the US or UK.[2] Plants are in Mexico, Malaysia, Taiwan, and Hungary.

To get prices down, engineers have to be very familiar with what goes on in manufacturing. If you separate engineering from manufacturing, you get overpriced designs.

Not that many people who went to a good engineering school in a first-world country today want to spend their lives inside a big factory in a low-wage country. But that's what it takes to make stuff.

[1] https://www.digikey.com/en/products/filter/chip-resistor-sur...

[2] https://jobs.bourns.com/go/Engineering/9254400/

View on HN

Being an engineer means mastering your production tool. For everything to do with physical production, you need to be close to the means of production to gather essential information on quality, capacity, operator feedback (machine and quality operators are invaluable sources of information.), etc.

Most information is not digital or hardly digitizable.

I don't completely agree with the article's classification of ARM as a hardware company. ARM produces VHDL and resells licenses, but does not produce any chips. It's closer to a software company than a TSMC.

I'd go one further and say you have to be at least a journeyman in whatever tools your process is using.

The difference between someone designing a part in cad and someone designing the tool paths for the machine that makes the part in cam is night and day.

We could discuss this at length, but I completely share your point of view. Anyone can design a part that's impossible to produce.

The real added value is knowing how something is actually going to be made, in how many stages, with what tools, what controls will be carried out, with what tools, what the acceptance and rejection criteria are, and how these criteria have been determined, are essential points.

> The difference between someone designing a part in cad and someone designing the tool paths for the machine that makes the part in cam is night and day.

I remember, in a CNC programming class, the instructor calling out one of the students on a lathe program: "One millimeter increments?! What material do you think you are using? Styrofoam?!".

That class is where I feel in love with the ASR-33 teletype and its cadenced hum. It was punching the tapes we feed into the CNC machines. I wish I could have bought that machine when it was retired not too long after my class.

> "One millimeter increments?! What material do you think you are using? Styrofoam?!"

I'm an EE not a MechE but I'd be truly curious to know if there are any MechE programs where a fresh graduate would have ever heard the term "feeds & speeds".

In a similar vein, I learned to solder in EE but not because of any of my course work. We were lucky enough to have an aerospace electronics manufacturer situated on the north edge of campus. The IEEE Student Society worked out a deal with them where EE students who wanted to learn could come and do a 3-hour crash course with the techs. I could solder before I did the course, but my ability to solder well improved dramatically as a result of those 3 hours of training. And, even more importantly than learning to solder, I learned a ton of things about solderability: what makes a circuit board easy to solder and what makes it hard to solder under different manual and automated manufacturing techniques (wave soldering, paste + pick & place, etc).

> if there are any MechE programs where a fresh graduate would have ever heard the term "feeds & speeds".

This was an extracurricular activity, and the MechE's were in their fifth year or so. I was in my first year (semester, really) and I was suggested I take the course because I was already a reasonable programmer and there was very little materials in the course, but it was more about programming the machines (simple loops, no real decisions, etc).

I was doing 0.1mm increments in my code because I "felt" steel wouldn't be soft enough for more, but I never got any real training on that before second year.

My brother, who graduated 2024 in MechE is aware of the term “feeds & speeds”.

Mechanical engineering is a pretty broad discipline covering everything from micro fluidics to structural requirements of sky scrapers. It’s a good skill to have but I’m not sure that awareness of operating a milling machine is critical for success after graduation.

>I'd be truly curious to know if there are any MechE programs where a fresh graduate would have ever heard the term "feeds & speeds".

Germany has dual degrees where you both learn a trade and get a degree. If you are doing this for Mechanical Engineering you definitely will learn this.

Degree programs also have required internships and there are definitely courses which you can take during your degree. I would be surprised if there weren't a majority of mech eng graduates who would know the basics of milling.

Did the instructor's comment mean that 1 mm increments are very large, and he expected the student to be doing much finer work?

What about using simulation(or estimation using software) to understand the manufacturing process well enough to design for it, without being in the factory?

Simulation is a tool. Determining whether the results of a simulation are consistent and valid requires a great deal of experience. Almost anyone can use fluid mechanics simulation software and obtain beautiful multicolor images. Only an expert will be able to determine whether they're worth anything, and when it's time to run a mock-up to validate calculations.

Simulation can help train new engineers, helping them to understand complex physical phenomena, but it cannot replace field experience.

The organization of the production site, the employees and the quality culture are an integral part of the production tool, and cannot be simulated.

For a while, we had the "maker movement" and "maker spaces", and people were learning that stuff. But that all tanked when TechShop went bankrupt.

There are still maker spaces around, but most of them are now more into sewing, paper folding, and hot glue than CNC machining. Few go beyond a 3D printer. The ones that do tend to have some kind of subsidy from a larger educational institution.

We used to have shops in schools back when funding for education went to education instead of administrators.

That's not been the case for a generation so no one under 70 even knows what we've lost.

Totally anecdotal but to your point, the engineering jobs in my hometown followed the manufacturing jobs in leaving town in the 1990s after NAFTA.

Engineering seems to be returning as domestic manufacturing increases thanks to foreign auto companies setting up shop across the state, replacing what the US companies left behind.

> spend their lives inside a big factory in a low-wage country

Some gladly would if paid handsomely by the local standards, that is, adequately by the US standards.

The bigger problem is raising children away from your native culture.

You should probably recommend them to Apple recruiters, since they regularly have shortages of bilingual top tier talent willing to work full time at major factories.

Even with extremely generous FAANG salaries in areas with cost of living less than a quarter of Cupertino.

I spent a decade in a low-wage country and the number of people who were "glad" to spend even half that much time there -- paid handsomely! -- could probably be counted on one hand. Virtually the only people who stay longer than 3-5 years are ones who end up founding their own businesses there.

Raising children away from your native culture isn't the deal breaker you imply. There are international schools (though, with eye-watering fees) and expat enclaves in most places I've been.

But very few (effectively zero, though I did come across a handful of exceptions) companies treat these employees the same way as the ones back in the home country. If you don't rotate back to HQ in ~3 years then you're in a career dead-end. So you've got this situation where you need people who are ultra-ambitious -- willing to throw away all their existing social networks to go work in a foreign country for years on end! -- but that means those same people aren't going to want to stay past their expiration date. And companies know that, too. A lot of them make it an explicit part of the deal. I met one high-level guy (regional CTO I think?) at Coca-Cola who was Indian and the deal with corporate was he'd do 3-years in a low-income country (not India) but then he'd get transferred to the US. Met some people in the oil industry who had similar deals. Do 2 years in Vietnam then you get to go to Malaysia or whatever.

So everyone fully admits that these countries are in fact "worse" in every way?

I’m not sure what “admission” you’re looking for? That life is generally better in wealthy countries? Wow big surprise

> So everyone fully admits that these countries are in fact "worse" in every way?

Considering money is the universal currency? Yes. Worse in every way.

Churches ask for donations. Women marry rich men. You can buy politicians. You can buy more expensive healthcare treatments.

Can you do that in a worse SEA country?

There are plenty of industries where product engineering is done at a different company or place than product manufacturing.

For example, consumer electronics, industrial machines and robotics, telecom and medical devices.

The fact that it exists doesn’t mean it’s the best. You see it all the time that businesses do something just because everyone does it too. For example, the current AI investments, collatoral debt obligations in the 2000s, and conglomerates in the 1960s.

>the current AI investments

As someone whose done both hardware and AI, the current AI investments are at worst a repeat of the 2000 dot com boom.

They aren't wrong, but they may be premature with how terrible our compute substrate is.

Taiwan is kinda nice though? And just a short hop to Japan for holidays or even long weekends.

Taiwan is the only Asian country that has western sensibilities. Gay? Who cares.

Very weird comment. Article talks about hardware talent in UK. Your comment tries to prove "engineering jobs" are not in the US by providing the job listing of a single supplier, when everybody knows that there are a huge amount of hardware talent in the US working at great companies that deliver amazing products. Your comment seems to equate "manufacturing jobs" to "hardware engineering jobs" which apparently isn't correct.

I wonder this is also what doomed some, but not all, of meta's hardware efforts like its watch from 2023 (Portal was successful and a solid build but that was more pacakging than manufacturing and died for a different reason: Priorities).

In general though it seems where design/code<->hardware feedback loop needs to be very fast in some cases, it is a non trivial separation of concerns.

Many of the legacy component vendors retain domestic US manufacturing to supply military parts. They come with a premium price and aren't usually worth using commercially but that is one of the last backstops preventing all knowledge from disappearing.

Reply on news.ycombinator.com