Hacker News Reader

I had a 89 Corvette...designed in the late 70's early 80's before computers were generally used to do so. It was a mass of shims and slotted holes to get things to line up.

I had a 1998 Corvette...designed in the early 90's with CADCAM. No shims. Things like a door latch and door made by entirely different companies, the three holes to attach the latch weren't shimmed or slotted, and the replacement fit exactly with no adjustment.

Cars were exactly the other example I was going to use but cut my comment short.

Anyone who has looked in the engine bay of a car from 50s/80s/00s/now can observe how tightly packed & compartmentalized everything is now. Early cars had a lot of room under the hood cuz mechanics were constantly in there tinkering, adjusting or replacing bits and pieces. It's incredible how long a car from a reliable brand lasts now, with minimal maintenance.

EVs are another step function in this change as theres even fewer mechanical moving parts and fluid lines to worry about.

Same as the old land rovers I used to work on. 2mm was considered miraculous tolerances anywhere. If something didn't line up it usually got hit until it did.

“I had a 89 Corvette...designed in the late 70's early 80's before computers were generally used to do so. It was a mass of shims and slotted holes to get things to line up.”

This has nothing to do with computers. Detroit just didn’t care about building decent cars back then. Mercedes and many others were able to build very nicely designed cars long before CAD/CAM was available.

Wouldn't you know it, the mercedes door handles from 1985 used slotted holes[1]. As much as you like them, Mercedes cannot transcend the technical abilities of the time.

[1] https://www.ebay.com/itm/174693656016

I guess my comment was more about the general build quality of US cars from the 70s. I have worked on some and it seemed they were just thrown together without any thought or care.

No argument there. They were definitely not going after the "luxury" market segment (and still don't)

I was going to argue with you but luckily I checked first :-)

My argument was that Foundations of Mechanical Accuracy (one of the canonical works of precision engineering literature) was published in the early 1900's, but no, it was actually in 1970. (https://pearl-hifi.com/06_Lit_Archive/15_Mfrs_Publications/M...)

Can't believe that I was that far off! I might have been thinking of Precision Hole Location, but even that wasn't published until the 1940's.

One of my favorite YouTube channels, "my mechanics", is a restorer who specializes in restoring beyond the original condition of the item -- he will frequently fill in holes and re-bore them when the initial machining was shoddy and led to a poor alignment of the parts.

That's fascinating, because recently I started thinking that I'm underestimating how long we've been able to craft (which may be different than engineer, or mass manufacture) things precisely.

Recently I dug teeny bit into pocket watches, and then wrist watches, which turnst out were a thing way longer than I imagined!

> Recently I dug teeny bit into pocket watches, and then wrist watches, which turnst out were a thing way longer than I imagined!

Weren't these handcrafted by highly skilled and highly paid specialists for a long, long time. In contrast to more modern mass production.

Yes; so I think that's where scope or intent of statement may come in. I think:

* We were able to craft, manually, things much more precisely for much longer than some of us thought

* We were able to design/engineer/mass-manufacture precise things for far shorter than many of us thought

What we have with CAD and robot aided assembly is making precision affordable.

Which is probably a good reason the early steam engines didn't take off. You could build something in the lab, but there was no way to affordably mass produce a usefully efficient version.

They were. With the following caveats: non, or limited, interchangeability of parts, horrendous precision of showing the correct time. All of which can be traced back to low precision manifacturing.

Modern items have stickers about warranty void if you open it, etc, other warnings like that. My 100% user serviceable coffee machine recommends you send it to them, cars encourage you to go to dealers. I don't see how that sticker is different in the 50s and now. There's a profit motive for them to fix it for you even if they don't charge it's still reputational (they stand behind their work).

Having this sticker on my device doesn't make me think that I need to send it in often, it's more that they're covering their ass.

The rest of your comment I'd agree with but a lot of things were designed around needing tight tolerances and using tight tolerances when you don't need them is not always a good idea. My earbuds won't charge if there's the tiniest bit of dust in the little recesses the charging pins go into or in the bottom of the well. Lower precision there would make them more reliable.

I think the most egregious one of these I've seen is a digital multimeter I have that says there are no user serviceable parts inside. Pretty standard, but it implies that even the sort of person who uses a multimeter isn't qualified to replace the fuse and 9v battery inside it.

Mad dad worked in airplane manufacturing. He once told a story about how every part in for something in a wing was in spec. But it was a foot to long all together. Not even close to fitting on the airplane

That was in 80s.

The entire fuselage for the last 737 NG to be built had to be scrapped, and a new one had to be fabricated by the supplier, because certain parts were so out of tolerance that the wings could not be fitted.

https://twitter.com/aeroimageschris/status/11524425725533265...

https://www.airliners.net/forum/viewtopic.php?t=1427119

edit - “Mad Dad” - that’s either a great typo or great use of alliteration.

> [...] the camera needed frequent enough service [...]

Let's hope that in the future software engineering becomes mature enough that we don't need daily security updates.

I think that needing legions of Archeologist-Programmers who will continuously tune the systems is much more likely.

Shhhh don’t ruin my retirement part time gig plans

Don't worry there's going to be a need for literal legions, and it still won't be nearly enough

"I had a problem so I thought I'll use Java - now I have a ProblemFactory"

Indeed, Any maker who has attempted certain kinds of projects (involving vacuum, mechanisms, etc.) will understand this.

Meanwhile the Swiss, making watches machined to an impossibly perfect degree since the 18th century: "I have no such weakness."

> Modern people overestimate how long we have been able to engineer things precisely

I think you're underestimating it also. For example, micrometres capable of measuring imperceptible differences have existed for many centuries, same for manufacturing machines (lathes, mills) that can shape objects to imperceptible requirements.

Cameras probably just don't require such precision.

> Cameras probably just don't require such precision.

Depends on the camera.

I think if we combine both of your comments, we get a more clear picture of reality: we've had the ability to manufacture one instance of something with precision for a while, but not the ability to do so consistently, reliably, or quickly.

i.e. tolerances have shrunk dramatically over the past century or so, along with time needed to produce things with such low tolerances.

However, while that is generally true across the board, tolerances haven't all gotten to the same place! Materials involved, costs, time to build, and need for low tolerances still largely affect all of it, of course.

And for some things, individual variations are even considered desirable.

Cameras always have been optical appliances. And as such they in deed do require a lot of precision. Especially those with changeable lenses, unless you don't mind your focal plane not being where your film, or nowadays sensor, is.

I presumed OP was talking about winders and so on. If shims were used for critical elements like optical alignment, I'm guessing that it's because this was the cheaper route to precision, rather than due to an inability to measure and manufacture precisely.

Cost is always a factor in manufacturability.

Optics absolutely had shims in this era. Talking 1930-1950 era Leica lenses for example. 50mm lenses that you can’t just swap in a replacement element without adjusting shims because they would match them and shim at factory.

I recall some lenses from the era having their actual exact focal length inscribed inside since they weren’t exactly 50mm and the variance was 0.2mm or so.

Don’t even get me started on Soviet cameras.

Recently I disassembled a Nikon lens, the first 24-120 with VR as the VR constantly engaged. The plan was to disconnect the VR unit completly and use, interim, without VR. Didn't work, as apparently apperture control (electronic in this lens) and focus (AF and manual, also electronic) go over the same platine. No VR, that worked, but also no focus or apperture.

Point of the story, the lens had some shims to get into the proper focal plane relative to the camera. The model in question doesn't have the best sharpness reputation, some people put it on the worst 10 list of Nikon lenses. The sample I had is sharp so (VR didn't engage until 28 mm). My theory now is tgat the lens is sharp, and the bad rep comes from compromises during manufacturing (Thailand instead of Japan, with all the corresponding issues of setting up high precision operations in a new site), and potentially sub-par QC. Not sure if other lenses have shims as well, I'm kind of not inclined to disassemble perfectly working samples just to find out.

So no, shims aren't per-se a bad sign but rather one of a certain calibration strategy. If those shims are everywere so, manufavturing quality propably isn't the best.

Oh, one word on 50 mm lenses. Those are among the fastes, sharpest lenses ever regardless of manufacturer. And dirt cheap compared to "pro" models. Seems to be quite well understood optics.

I do have two pre-AF Nikon lenses, and those are master pieces of precision mechanics, dating back to the late 70s and early 80s.

Lensrentals.com founder Roger Cicala publishes teardowns of high-end modern lenses and I recall that he and his team occasionally (often?) find shims used to calibrate the optics.

https://www.lensrentals.com/blog/author/roger/

nit picks on this:

Humans are capable of detecting extremely small details with their fingertips[1] anecdotally, this can be trained if you use your finger tips for precision work a lot, you can easily feel things that you can't see without significant magnification.

Additionally, I would push back on "many centuries", the Machine age came about in the mid 19th century, so if were being generous, 300 years[2].

[1] https://www.sciencedaily.com/releases/2013/09/130916110853.h...

[2] https://archive.org/details/englishandameri01roegoog/page/n2...

[2] is cited on the wikipedia page for micrometer. Cool book though, might read it.

Resistor matching is still a thing to this day.

And also tube matching for us tube guitar amp users.