Inside the Hardware Product Everyone Loves to Hate
Dec 20, 2019
Editor’s note: As our Entrepreneur-in-Residence at Eclipse Ventures, Ben brings a founder’s perspective and his passion for hardware design to our discussions. Here, he continues his smashingly popular product-teardown series.
There’s a single product category that is universally despised by nearly every technology-using human on Earth: the printer. While I’ve certainly had my fair share of disagreements with printers over the years, I’ve come to be increasingly curious about a set of intersecting qualities that don’t seem to exist anywhere else:
Technology. Printers are surprisingly complex machines. Manipulating 70 micron thick pieces of paper and accurately applying permanent ink is among the most sophisticated pieces of physical technology in the average home.
Cost. Consumer inkjet printers (even ones with scanners and copiers) are dirt cheap. Like sub-$50 cheap. For context, shipping them is usually more expensive than buying them. ????
Hatred. I’m pretty forgiving of technology but even MY heart rate rises the second I realize something needs to be printed. In my anecdotal conversations, this is true for approximately 10 out 10 humans.
As with any product that doesn’t quite make sense, the best way to understand is to take it apart. With my startup hat on, I’m always wondering why a small company hasn’t yet been able to unseat the big players in the printer space (Canon, Brother, HP, Epson, etc.). This post makes it clear why that is so very difficult.
Shopping for a Popular Inkjet
Whenever I tear down a product (I have abit ofan addiction), I start with scouring eBay for broken versions on the cheap. But this time, I ran into a curious problem.
I started with looking at popular, low-cost color inkjet printers with good reviews on the various tech-review sites. I landed on the Canon PIXMA MG3620. MSRP is $79.99 but retailers (even Canon’s own site) sell it for $49.99. Not bad. Off to eBay to save a few bucks on a broken one!
On first pass, the used/open-box PIXMAs are as cheap as expected ($10–25) but I started noticing a curious pattern:
This is not necessarily an earth-shattering revelation, but it begins to reinforce the picture of a very bizarre set of economics where the printer itself is significantly cheaper than the cost to ship it. Whelp, off to Amazon for a brand new printer for half the cost thanks to free Prime shipping.
The new printer arrived in all its glory. Right off the bat there’s a handful of (sometimes) costly items that products in this price point don’t always have to contend with:
Large, full color corrugated box
Six large, molded polystyrene shipping bumpers and other sheets of shipping material
Two inkjet cartridges (one color and one black)
Accessories: USB cable, CD-ROM (can’t believe they still include these), several bound manuals, power cord
Lots and lots of packaging trinkets (little molded and die-cut pieces of plastic, wire ties, etc.)
While it’s always hard to estimate BOM (bill of materials) costs here, it would shock me if these items were less than $10 (25 percent of price I paid).
One thing of note from just the packaging: There are an INCREDIBLE number of features this product has to have just to be competitive. Print/copy/scan, wireless connection, full-bleed/full-color printing, USB, modem/fax, software system to manage remotely, web interface, and connections to a pile of cloud software platforms.
After cracking open the covers, the plethora of components required to make these little machines tick is even more shocking than expected:
It’s important to remember that each of these complex plastic parts has a multi-part injection mold tool, some have pad-printed components (i.e. the green logo on the Start button), and nearly all must interface with: LEDs, electrical components (like wires and PBCs) and of course TONS of mechanical parts (springs, gears, bearings, etc.).
Turning the printer around and going after the power supply, we see the first clue of how the economics can possibly make sense:
Nearly all printers take in 120v-240v AC mains power to avoid the yucky power bricks sitting on the floor. In normal hardware startup operations, this is hard and expensive due to the safety and certification risk of dealing with dangerous high-voltage AC power. This is why nearly all products from small companies have those dreaded “wall warts” that step down dangerous AC to safe DC power (nearly all of these are bought off the shelf).
Even in big companies, Underwriters Laboratory (UL) and other safety agencies require any product that deals with the dark arts of AC power to be carefully tested. This is expensive and time consuming. So Canon (and most other manufacturers) have come up with a neat trick to save cost/time: build a single internal “power supply” module that is shared across a ton of SKUs, then certify the module itself and reuse over and over again. This saves cost versus buying off-the-shelf wall warts and saves time by avoiding slow regulatory testing for each unique printer SKU. Very smart and something we’ll see over and over again with this product.
Removing the left-side cover, we get our first glimpse of the control board and gear train for the paper handling system. Parts, parts, and more parts! In addition to the number of complex, injection-molded plastic parts, which I expected, there’s a surprising number of stamped and die-cut metal parts. One example is the stamped/molded tray that holds the printed circuit boards (PCBs). This sub-assembly is likely inside nearly all Canon inkjet printers, even if certain features (like scanning) aren’t. Similarly, with the WiFi module: if the product has WiFi capabilities, that part is populated; otherwise, it’s the same sub-assembly missing that component.
I’m starting to build a thesis that nearly all of Canon’s inkjet printers reuse identical components (both between printer SKUs but also over some multi-year period of time). Only certain exterior plastic parts are tweaked to differentiate between price points, geographies and performance.
The pattern of modularity continues to get more apparent, especially when looking at the PCBs:
The WiFi module is another example of custom components that are pre-certified and used across a large swath of Canon’s SKUs. Similar to the power-supply module, WiFi cards are regulated by the FCC and take time/money to certify. If this chip was mounted directly to the primary PCB, each SKU would have to be certified separately. Canon’s solution of creating and certifying a separate component and adding that to the product via a cheap ribbon cable saves time and money.
Handling and printing on paper is extremely technically difficult. It only becomes clear how this is done once most of the bulky plastic parts are removed:
In the image above, we can see how all the subsystems of the printer fit together to shoot ink onto a moving page at high speeds. The red arrow highlights the peristaltic pump responsible for powering the inkjet head cleaning system. This pump and related components use suction to remove dried up ink and clean off the heads onto the cleaning pads (both near red arrow and green arrow). Interestingly, the suction pump is driven by the DC motor that controls the paper feed (another cost-saving method). The yellow arrow shows the main feed roller that’s driven by the primary DC motor and the related gears (blue arrow). All wires and circuit boards meet up at the pink arrow.
Moving paper through a printer requires extraordinary accuracy. To my initial surprise, this isn’t accomplished with a precise stepper motor; that would be far too expensive for such a cheap product. The paper feed mechanism is powered by a cheap DC motor that drives a set of cheap plastic molded spur gears. Normally, this drive-train configuration would be far too inaccurate to feed print on a piece of paper at 22PPM. But a nifty custom optical encoder (blue arrow) allows Canon to:
lower BOM cost on a cheaper motor
maintain accuracy of the feed mechanism at the final stage of of the gearbox, and
keep, change, or reuse the encoder assembly separately from the motor (and vice versa).
By far, the most impressive part of an inkjet printer is the system that deposits ink (I know, surprising). Ink stored in the small tanks/cartridges is deposited one at a time by the millions by an insanely tiny piezoelectric elements that make up the print head (green arrow). To reduce cost, the elements are actually controlled in the printer, rather than in the inkjet cartridge, using the circuit board that sits on the print cartridge (purple arrow) connected via some nifty pins and pads (yellow arrows). The cartridge slides horizontally across the paper on the stiffener rail (blue arrow) and powered by a gear motor with a rotating rubber belt (red arrows). Needless to say, doing this accurately and cost effectively is incredibly difficult. Some mind-blowing cost engineering has gone into these components.
Enabling Business Model?
Many argue that this amazing engineering is all possible due to the business model Canon/HP/Epson/Brother/etc. employ: sell the printers at a loss and make all margin on the ink — the so-called “razors and blades” model. There is no doubt some truth to this but after digging into this printer, I actually question how much money is lost on the initial sale of the printer.
Most companies are beleaguered by SKU proliferation (an increasing number of unique products usually due to colors, sizes, specs, etc.). After taking apart this product, I theorize printer companies actually benefit from SKU proliferation by selling up with a dizzying array of printers and complicated model numbers all the while reusing the same components across hundreds of different SKUs over several years. The net effect of this is printers that actually cost far less to manufacture than you might guess, as the components are made by the tens of millions using tooling developed and purchased many years ago. This creates a wide swath of SKUs that have similar COGS (cost of goods sold) but wide ranges in retail prices.
If this theory is true, we have finally stumbled on the real reason there are no good printers: a startup building a 10x better printer from scratch would have to spend an enormous amount of capital and will only be able to monetize it across a single SKU to start. The bar for a capable printer is incredibly high, the cost is insanely low, and it’s a mindbogglingly complex problem to solve well. As with many tech markets, the structure of the incumbent companies makes it very difficult for new entrants to succeed.
That being said, if you want to try to disrupt the printer market, I’d love to meet you! There’s a lot of printer hatred out there that could be turned into love ❤️
Follow Eclipse Ventures on LinkedIn and Twitter for the latest on the Industrial Evolution.