I had already planned to write about "parts" when iFixit's D5100 tear down press release popped up in my email this morning. But before we get to that, let's back up a bit.
Back when we designed the original Quickcam in 1994, one of the design goals (in the Marketing Requirements Document, otherwise known as the MRD in product management speak) was "fewest number of parts necessary to get the sensor signal into the computer's CPU." To refresh you, the Quickcam was a simple tethered digital camera that plugged into your computer. Software on the computer provided still, video, and conferencing capabilities.
The design goal was specific for a few reasons. First, the company creating the Quickcam, Connectix, was a software company. We were worried about three things we weren't used to in the software business: how many parts we had to stock in inventory to build cameras, how much hand assembly time it might take to manufacture it, and the overall product margin. We felt that by reducing the complexity of the product we could minimize the first two things and maximize the latter. It actually became a bit of a game for the engineering side to reduce the number of parts. If I remember correctly, the final count was 28 (though I include the lens mount and lens itself as one, as we received it that way from the manufacturer).
I was reminded of that the other day when I first saw the BigShot project. BigShot was developed at Columbia University and represents a work in progress. The idea is to create a simple, build-it-yourself camera that allows you to learn some science while constructing your functional camera. If I'm not mistaken, the BigShot comes in 18 parts (though they label the two digital boards as one part each). And you can see how that brought up the Quickcam memory. Simple. Direct. To-the-point.
Reducing parts volume doesn't necessarily make for a less capable or durable device. In some ways, it's better because the device can be more easily repaired should something happen to a part. But parts reduction also forces designers to think strongly about function. No features that only 1% of the users will ever try: you've got to make sure that the parts you do use actually perform the functions that everyone needs done. It's always interesting to me how such design focus usually results in something unexpected. In both the Quickcam and BigShot, the "unexpected" bit has to do with power. The Quickcam didn't have a power supply, for example. Instead, Jon Garber came up with a wild idea when I insisted on no AC wall wart: use bits coming from a serial port on the computer much like you use water to drive a wheel to generate power. There was just enough energy derived to power the sensor and the communications. Likewise, the BigShot has a power surprise: you can hand crank it to power it for a photo if you don't have a battery handy.
Which brings us to the D5100 tear down. Ifixit jokingly refers to 4 billion screws, but that's less an exaggeration than it seems. The number of hand-assemble points (screws, cable plug-ins, etc.) on modern DSLRs is crazy. The overall parts count reaches well above 2000 for even the low-end DSLRs. Part of this is the propagation of old designs. Late film-era SLRs were already getting complex by the addition of automated features (metering, focus, frame advance, etc.). Even back in the N8008 days cameras had a CPU. By the time we got to the F5, we not only had a CPU, but both flash and RAM memory as well as serial communications (through the ten-pin connector).
The question that needs to be asked is this: how much of all that complexity we've added to a camera is really necessary? For example, how many of you have actually connected something to the USB, video, microphone, and HDMI ports of your DSLR? That's four connectors that aren't getting a lot of actual use. Of course, if you took those things off the camera, your competitor will ding you for not having those items.
This is one reason why I've been arguing for modular cameras. All that communication (USB, video, microphone, HDMI, as well as things that we don't have but will such as Bluetooth and WiFi) should be optional and upgradable.
One strong trend that appeared not long after the first DSLRs appeared and continues to this day is the demand for a "digital FM2n." Simple. Direct. Purposeful. To-the-point. How many parts do you think would be in it? 400? 500? Certainly not 2000+. What's that say about costs, profit margin, and assembly time? It really startles me that the camera makers don't see and attempt to satisfy this pent-up demand.