Card scanners and other single-purpose devices

A general-purpose sheetfed scanner is built to accept a range of paper and does not fully commit to any part of it. A single-purpose device commits: a business-card scanner, a receipt-only scanner, a cheque scanner, a photo scanner. Each one narrows the mechanism to a single media size and thickness, and that narrowing is a real engineering advantage — it just comes attached to a specific way of becoming useless.

What committing to one media size buys

A feed path only has to be forgiving if it does not know what is coming. The separation problem described in how a document feeder works is hard mostly because a general feeder must handle a thin slip and a stiff card in the same batch with one set of roller pressures and one gap. A device that only ever sees cards can set that gap for a card. A device that only ever sees till slips can set the guides narrow, space the rollers for a short item, and tension the transport for coated paper.

The optics can commit as well. A fixed-width sensor sized to the media wastes nothing, and a fixed focal distance can be set for one paper thickness rather than compromised across a range. Fixed also means fewer moving parts, which is why these devices are often small, quiet and mechanically simple.

The practical result is that single-purpose devices frequently handle their one media better than a versatile machine does. A card scanner that will not accept anything larger than a card genuinely does not jam on cards. That is what happens when a mechanism is allowed to be specific.

What you give up

• No headroom whatsoever. The moment a slip is wider than the slot, longer than the transport allows, or stiffer than the rollers were tensioned for, the device is not slow at the job — it cannot do the job. There is no degraded mode.
• Fixed capture, or nearly. Many of these devices expose far fewer settings than a general scanner, sometimes a single quality option or none. If the fixed default is wrong for faint paper, there is no exposure control to reach for.
• A short feed path. Short is good for speed and bad for control: a short item in a short path can be released by one roller set before the next has taken hold.
• The software is often the product. This is the important one.

The category is defined by its software more than its mechanism

A card scanner is usually not sold as a scanner. It is sold as a way to get cards into an address book, and the mechanism is a delivery vehicle for that. The same is true of most receipt-only devices: the pitch is the filing and extraction that happens afterwards, not the capture.

That has two consequences worth pricing in. The first is that settings you would expect to control are frequently not exposed, because the application was designed to be pressed once rather than configured. The second is that the scan may never exist as a file in an ordinary folder — it may go straight into the application’s own store, exactly the arrangement described in when the bundled software owns your scans. On a general-purpose scanner that is a default you can work around. On a committed device it can be the entire operating model, with no path that writes a plain image anywhere.

Before buying one, the question to answer is not how well it scans. It is: where does the file land, and can I get it out without this application.

When a single-purpose device is the right answer

There is a genuine case for these, and it is narrower than the marketing but real.

• The volume is high and completely uniform. Somebody who handles a lot of one media and nothing else is exactly who the commitment was designed for.
• The device does something a general scanner cannot. A mechanism built for stiff card stock or for a media size a sheetfed path refuses is not a convenience version of a general scanner; it is a different capability.
• It is a second device, not the only one. A committed device alongside a general-purpose scanner is a reasonable arrangement. A committed device instead of one is a bet that your paper will never change.
• You are avoiding a compromise you have actually hit. If your general scanner jams on one specific media every time, a device tuned for that media solves a measured problem rather than an anticipated one.

If none of those describe you, the honest comparison is not against another scanner at all — it is against not buying one.

The failure mode to price in

A single-purpose device fails by outliving its software. The mechanism is simple and tends to last; the application that gives it a purpose is small, tied to one company and to specific operating system versions, and frequently the only way to operate the hardware at all. When it stops being maintained you have a working mechanism that nothing can talk to. The general version is the driver is the part that expires, and it hits committed devices harder because there is rarely a generic fallback.

Two things reduce the exposure. Prefer a device that also appears to the system as an ordinary scanner, so a generic tool can drive it if the application goes away. And prefer one that can write a plain image or document file to a folder you choose. Neither is glamorous; both decide whether the thing is still useful in several years.

Testing one before committing

The test is the same shape as any other with one addition: test the edge of the media range, not the middle. Feed it the widest, longest, stiffest and thinnest items you own within its stated range, because a committed device’s whole proposition is where that range sits. Then verify where the file went and whether you can open it with something that did not come in the box. The broader routine is testing a scanner with your own paper, and it applies with more force here, since a device that cannot be reconfigured has to be right on arrival.

Whether the resulting files are adequate as records — and for how long they need to exist — depends on rules that vary by country and by the kind of entity keeping them. That is a question for your own tax authority or an accountant in your jurisdiction.