Camera Repair Course (2019)
learncamerarepair.com
learncamerarepair.com
Two particularly good ones that I remember were the VCR and the SLR camera. I brought in the camera -- I think I paid ~$25 for it on eBay, and I still have the shutter mechanism somewhere. Makes for a good fidget toy.
This exercise really made me appreciate just how far you can get with good design and clever engineering. Even though I knew roughly how the camera worked and what its advantages were, we inevitably came across some assemblies that we couldn't immediately identify. Eventually someone would work out that it was part of the metering or autofocus system, and everyone had fun shining light through it to try and gain an intuitive understanding of how it worked.
And you can absolutely do this! Most of those assemblies were clever because they were so devilishly simple. The autofocus, for example, redirects a small patch of light from a particular location onto a separate mirror or prism that can be placed out of the way. Some optics expand the light onto a 1D CCD or a pair of detectors, and then some rudimentary circuitry can be used to drive the lens motors and close the feedback loop.
The modern techniques might be conceptually similar, but it's so cool to see the problem solved without a massive sensor and a DSP that's arguably a supercomputer -- Just some discrete components.
It feels very: (1) I can't believe this actually works, (2) I can't believe they managed to assemble this tiny thing in a cost-effective way, and (3) you're trying to tell me this Rube-Goldberg machine was relatively reliable too!? I know our modern monolithic slabs of glass and aluminum and potted silicon and millions of lines of code can probably beat these old devices on almost any quality metric that anyone actually cares about, but there's something charming and magical about the way we used to make things.
I have a Canon EOS 55 from the late 90's, and in many situations it has better autofocus performance than my Fujifilm mirrorless from 2017.
I guess contrast-based autofocus is what really needs the DSP, since you need to do edge detection / FFTs on more patches of image for the technique to work well, as well as PSF analysis if you want to guess the lens defocus direction. With phase it's just a few cheap 1D correlations, and you get accurate lens direction information for free, which explains why it feels so much faster.
Unfortunately we didn't have access to an electron microscope, so a modern mirrorless camera would have made for a rather less interesting teardown.
New, higher end mirrorless cameras still use phase detection just like the DSLRs of old. Sony still uses old-style sensors but mounted directly on the chip, while Canon uses pixels that only accept light from some directions to calculate phase information. Both of these tactics end up with very similar results.
Though it is augmented with DSP.
Something that really slowed it down is that Canon filed a patent for Dual Pixel AF which was the first on sensor pdaf system to outperform previous systems, and then proceeded to not make any serious mirrorless camera using it. All the worse is that DPAF was a fairly minor invention that other companies didn't do first mostly because they didn't have the manufacturing capabilities yet - basically using a microlens to split the light instead of a divider to obscure it.
So Sony had to find ways to make it work without infringing on the patents and without artifacts.
> Mirrorless cameras with phase detection have actually existed for a long time - actually since 2010
I guess that depends on your timescale. The first SLR camera was 62 years ago, with the Nikon F (1959-1973, 14 years span). The first [SLR] camera with phase detection autofocus, the Canon EOS 650 (incidentally my first SLR [1]) was released in 34 years ago, in 1987. It wasn't until 2002 (Canon EOS-1Ds, 19 years ago) that there was a DSLR that could capture images on par with film.
Even if Canon deliberately suppressed mirrorless PDAF (which I don't dispute), phase detection autofocus has been around much longer than digital cameras, and at least until recently DSLRs have had better autofocus performance than mirrorless.
To be clear, I'm not cargo culting DSLRs here. My point is that beyond the change from film to CMOS sensor, DSLRs are not significantly different from the film cameras they replaced.
[1] I mentioned to a coworker that I was thinking picking up photography; the next day she gave me a bunch of film photography gear that had been sitting in her attic for the past 20+ years. Unfortunately, the EOS 650 ended up getting stolen.
Mostly my point is that mirrorless cameras adopted PDAF as soon as making a useful pro-level mirrorless camera was possible.
Also, I'd argue that mirrorless cameras are basically a digital version of rangefinder cameras - they make a lot of the same tradeoffs optically and indeed there is a continuous history from digital rangefinders to modern mirorrless cameras.
A weird little side effect of most people moving to digital cameras. Digital cameras have their upside. I took a photo development course in my twenties. I don't recall if I completed it. Probably not.
I have a serious medical condition and it puts one at risk for lung bleeds. Years ago, on an email list for the condition, someone who kept a health journal for their child was able to track their lung bleeds and determine they were being triggered by darkroom development work.
I've had several somewhat serious photographers in my life, one of whom set up a darkroom in their home. I'm glad to have digital cameras available for me because I can't safely do darkroom work. The chemicals are a danger to me.
But it's weird watching old style cameras become something of a dying art and it makes me think of that bricklayer who, late in life, found his skills in demand for doing historic restoration.
This looks like a really excellent resource for anyone interested in this hobby.
Most commonly asked questions were "did you see anything 'weird' wink wink?" Yes.
"did you look at _all_ the photos?" You didn't have to, but the printer running unadjusted was crap since it kind of tried to level the exposure and color tint to some average. I always looked at every pic and adjusted according to how I imagined the photographed wanted it to look like.
Loved taking those machines apart and see how they worked on the inside. Also, for a low-wage manual worker that didn't get a lot of training on the job, one of the chemicals we filled up with from large canisters (for C-46? E-46?) had a warning label it would react with ordinary house-hold cleaner chems and form mustard gas. That was kind of stressful.
Presumably you're referring to the bleach stage (or blix stage which is bleach + fix).
I'm pretty sure regular bleach is more common in households than C-41 bleach.
Edit: Bleach + ammonia makes mustard gas, and ammonia is contained by some window cleaners. I think it's a lot more common for someone to mix bleach and window cleaner (thinking it works better) than to mix window cleaner with photographic chemicals.
Sorry about your health conditions, but I'm pretty sure that modern chemicals are much safer than they were in the past. Black and white chemistry is largely benign; the worst was apparently formaldehyde in older C-41 kits which has since been replaced. It seems like too much trouble to make color prints by hand anyways.
In general it feels like the world is heading towards un-understanding. The human world has become so big that we have relinquished our knowledge.
Oh sure, there are the John Deeres's, Monsanto's, many many many of us programmers here, actively or passively reducing how knowable, how manipulable the real systems of the world are to end consumers. Computing for so long gave enormous powers, but lately it feels like it winnows & narrows down what is possible, what is understandable, what is visible.
But I also feel like this was part of the direction, part of what globalization does, part of what productization does. Talent & skill are no longer dispersed widely, but concentrated, in very large enterprises that make gobs & gobs & gobs of money, that own the seemingly ever complexifying means of production & supply chains behind it. We have lots of great means in the world, but as our consumer expectations rise & produce quality increases, there's less interest in what the small local shop can do, with it's couple camera gurus, who once would have been a formidable force.
Trying to articulate the values of knowability, of understandability, of systems that we can be involved with is very very hard, especially after access has already faded, especially as the visible, humanizing upsides have decayed. This was mentioned by @adamhjk in his recent thread on Copilot, & FOSS[1], and this is one of the prime things that keeps me up at night: the world is becoming manufactured, and it doesn't need us, but if we could be here, if our souls were given some self-determinism, I think the world & our souls would both be much richer places.
The problem is the deliberate secrecy of knowledge; holding information to oneself in order to gain an advantage over others.
You don't need to understand gravity to benefit from the gravity. Likewise, the average person doesn't need to understand integrated circuits, atomic physics, or any other high technology. The harm comes from those who hold information to themselves in order to exploit others.
Passive components are still (relatively) easy to make --- hand-winding inductors is still common in amateur radio, for example; and even diodes are possible (crystal radio), but transistors and other semiconductors are definitely beyond the average hobbyist.
(If you have glassworking skills, vacuum tubes are just barely doable --- that's how they were invented, after all.)
On the upside, hopefully soon-ish they well. Open chip design has become a very quick moving area lately. We're starting to see more fabs open up to having chips made.
But the makers probably won't be building the chip fabs to make those chips.
> It's magic with a DIY layer on top.
It'd be lovely to see some folks try to DIY a fab. How hard do you think it would be to go build a 1µm chip fab yourself (what the 486 launched on)?
The fab is hard (probably! not sure!), but we're getting pretty good at spreading the knowledge of how to design chips.
Your point is well taken that we can't know everything. I like the above quote, "magic with a DIY layer on top". In the 3d printing scenario, I think we're closer to full-knowledge than we might originally credit. With something more complex like an OS, I'd rate it down to more a question of knowability: no one may know Linux entirely or even let's say (a much vaster) Freedesktop stack, but there's a couple thousand people on the planet (maybe decreasing, ffffff) that, if they had some reason to figure something out, could go & dive in & understand all the relevant subsystems to the challenge at hand reasonably well within a day. Computing drags in so so so many things, there is so much going on, just mountains of code... but our tools to navigate & observe & experiment & view are highly highly refined, are very very good. Computing is imminently knowable, everywhere (except when institutionally designed not to be).
Personally I really think the web as a medium is probably one of the best ways to introduce people to information technology, to bestow agency & learn about systems. There is a consistent, malleable, programmatic (inter-)system, focused on media delivery (immediate results) with well defined rules. But I wrestle with this difficulty you present: even with the best DevTools (quite likely one of the greatest understanding/learning-machines on the planet), under the surface lies of it all is an incredibly complex browser, constituting a vast magical system. One can be a complete expert at web development, but few of those experts are browser experts. Usually there's no call to go below the abstraction boundary, to understand the magic. But it is a barrier. The web is by far my most preferred rich environment, the most hands-on, practical, tactile thing I could imagine putting before a new IT recruit, but it's with faith not reason that I would justify putting them to such a highly abstract machine first. And, alas, there aren't even that many browsers left to compare across- not that most were open source or viewable anyhow, not until very recently.
But here it is on HN and you don’t have to know anyone or anything more than the phrase “camera repair” can go in the Google.
Today interest is enough to get started along the learning curve. Sure it would be great to have access to an experienced expert on the workbench by the door. But that was always an extremely limited resource.
The philosopher Jean Baudrillard speaks on this topic a bit in his last work, "The Agony of Power." Highly recommended reading for all those in tech
Sure, the average person probably knows/understands less of the sum of 'human knowledge' than the medieval man. But the average person has more knowledge, both learned and learnable, than ever before in human history.
Most people have never been particularly knowledgable or skilled. The average medieval peasant was no more able to make steel than the average person today.
Ideas and theories that were groundbreaking at their time, such as Newton's calculus, Shannons information theory, and Einstein's theory of relativity are now undergraduate (or even high school) courses. As part of my B.Sc in Computer Science, we studied public key cryptography (in particular RSA), which conceptually didn't exist until the 1970's. The average undergraduate probably knows more about cryptography than Alan Turing ever did.
In the 1800s, we didn't have biology, chemistry, and physics, we simply had "natural philosophy". The sum of our scientific knowledge was so meagre that there was no need for that distinction. Today, not only is physics distinct from biology, but quantum physics is distinct from geophysics. In medicine, don't just have doctors, we have neurologists, paediatricians, and radiologists.
> Trying to articulate the values of knowability, of understandability, of systems that we can be involved with is very very hard
The world today is more knowable than ever. When I was in primary school 20-something years ago, if I wanted to find information I had to go to the library and hope they had a book on the subject. These days, if I'm at the pub stuck in an argument about who scored the first try in the 1987 Rugby World Cup (Michael Jones), it takes about 10 seconds pull out my phone and find the answer. I think your concern is the opposite, that there's too much to know. It's so knowable that we spend at least 10 years at school just learning things.
Until recently, apart from (and due to) the lack of knowledge, it was incredibly hard to transfer and obtain information.
Before ~1455, when Johannes Gutenberg developed printing press, the reproduction of knowledge was incredibly labour intensive. It took more than a year for a monk to create a copy of the bible. Gutenberg's printing press shortened that to a week. Today, I can download a 40 megapixel copy of the Gutenberg bible (2 GB) in less than a minute. [1]
In 1820 only 14% of the worlds population was literate, it wasn't until the 1960's that more than 50% of the world could read. Even now, in Australia, 20% of Aboriginal people are illiterate, and up to 65% are functionally illiterate.
That said, we shouldn't rest on our laurels. We still need to continue to strive for our right to the freedom and access of information and knowledge.
TL;DR I think you have a romantic idea of the history of information and knowledge.
[1] For what it's worth, I'm not religious, but it's a good example of the history of information
But I was born into that world. The world that has happened during my lived life has been one where a greater and greater amount of the total knowledge has been withheld, retained, kept proprietary.
And worse, a large amount of this knowledge is knowledge that is impossible for anyone else to ever figure out. Almost all manufacturers keep their processes & ingredients secret. Want to know what this Tyvek thing is that you wrap your house with? We'll give you some marketing blurb, but mostly you have to trust us. With cloud computing, with DRM, there is no observable world: where-as with Tyvek we could buy some & put it under microscopes, tear it apart, run mass-spectrometers on it, try to reproduce it, we couldn't even take our chips out, give them an acid bath, & see what they do any more, because vast amount of the functionality is behind a firewall. That knowledge is physically retained, withheld, guarded inside a data center.
And that knowledge is vast, is the very core of the business of many of the biggest companies of the world. Powering products that we interact with for hours a day.
This domain of unnatural knowledge defines more and more of the world about us. It shapes society more actively in our day to day than our relationship with the natural world, our knowledge of the sciences. In spite of being the most active force shaping the world about us, humans, it has felt to me, have less and less ability to peer into, to see, to understand the world about them. While our total knowledge grows & we have a bigger pool of knowable things- truly a wonder- as you argue, I argue that the past handful of decades have seen far more knowledge created which is un-discoverable, un-findable than at any other point. And it is knowledge of the basic world about us, of how the basic elements of our world work. This to me is an anathema.
Given the opportunity, DuPont and Amazon would hide their proprietary knowledge just the same. Maybe one chemicals are easier to reverse engineer than cloud platforms, but the intention is the same.
Since then I've gotten more serious about film, picking up a Leica M2 and learning home development and building my own scanning rigs. The Leica Ms, at least the all-mechanical film-based ones [2], really deserve all the praise they receive. The serial number on mine dates back to the late 1950's, and it will, hopefully, with proper maintenance and regular usage, outlive me. I love my digital Fujifilm gear but am under no illusion that they will be usable in ~10 years time.
I once struck up a conversation with an older photographer using a film Leica. He said that he's optimistic about the future of film considering its resurging popularity among younger people, but is concerned that no one is picking up the repair trade, leaving maintenance in the hands of the mother companies who have already began discontinuing repairs for some of their models [3].
Here's to hoping mechanical repair receives the same renaissance as film photography and vinyl records.
[1] Dubbed "the poor man's Leica," now fetching ~$100 on eBay.
[2] From the screw mount models up until the M7, which gets hamstrung without a battery to operate the shutter.
[3] Admittedly, their digital models, but it would be best to avoid a monopoly in the hands of a company already known for their boutique pricing.
Any film photographers in Sydney Australia, reach out. (Particularly any that can get a Seiko 0 shutter adjusted.)
I learned from this Fix Old Cameras video https://youtu.be/P57xV0mz8mI
I used it on a Mamiya 80mm f2.8 lens set for C-series TLR’s two years ago. Still working.
Only caveat is to also clean the aperture blades if they are part of the shutter. I didn’t the first time so it two tries to get everything smooth.
It turned out there were a couple of things wrong. The main one was that lubricants had gummed-up the shutter blades and the slow-speed escapement (that's two separate things wrong). The gummed-up blades had caused mechanical interlocks in the lens to get out of whack and jam the blades open.
Soaking the entire mechanism in Shellite (lighter fluid in US) for a day or two unstuck the shutter blades, and some poking-around freed the mechanism. Working the escapement repeatedly has it moving nicely; a bit slow but close enough.
Be very careful with lubrication. Some left the factory with none at all, and those places that were lubricated were things like the cocking lever; not the escapements or clockwork. The lubricants used are those used for watches: often whale oil or modern synthetic replacements, not stuff from Bunnings or Supercheap Auto...
It is recommended as a first step in a Fix Old Cameras video linked in another comment.
Many recommend alcohol, but it often contains a bit of water and doesn't easily dissolve oils and greases. I've found Sheelite (lighter fluid) to be about the best.
For those of a certain age, Freon was the go-to solvent. Ozone layer be damned!
(Are the panels out of alignment? There’s a secret service menu to fix it easily! Do you have discoloration creeping in at the top & bottom? Remove the cover and just push down on the little plastic thing next to the lamp!)