Sheetfed, flatbed, or overhead: choosing by the paper you have
There are only three ways a machine can look at a page: pull the paper past a fixed sensor, hold the paper still and move the sensor under it, or point a camera down at it from above. Everything sold as a receipt scanner is one of those three, and the choice between them is decided by the most awkward item in your pile rather than the most common one. Work out what your worst piece of paper is first, then pick the mechanism that survives it.
The sheetfed scanner moves the paper
A sheetfed scanner holds the sensor still and drags the sheet past it, usually with a set of rubber rollers. This is the mechanism behind almost every device marketed for receipts, because it is the only one that can take a stack and work through it unattended. Feed a tray of slips, walk away, come back to a folder of files.
The price of that convenience is that the paper has to be strong enough and regular enough to be dragged. A sheetfed path is unforgiving about torn edges, staples, sticky tape, heavy folds, and anything that has been in a wallet long enough to go soft. It also puts the paper under mechanical stress at exactly the moment you can least afford to lose it. If the item is fragile or irreplaceable, a feeder is the wrong tool, and knowing which items those are is most of the skill. The mechanics of the feed path — and why receipts specifically give it trouble — are worth understanding before you buy: see how a document feeder works.
The flatbed scanner moves the sensor
A flatbed puts the item face-down on glass and slides the sensor underneath. Nothing pulls on the paper, nothing bends it, and the item is held flat against a known reference plane, which is why flatbeds produce the most geometrically honest image of the three. They will happily scan something torn in half, something with a staple still in it, or something so faint that you want to try three different exposure settings on the same sheet without it going through rollers each time.
The cost is throughput and desk space. Every item is a separate lid-lifting, positioning, previewing operation. For a shoebox of slips that is untenable; for the twelve documents a year that genuinely matter it is the right answer. Many people end up wanting both mechanisms and buying a device that has a feeder on top of a flatbed — which is what most office multifunction units already are, discussed in the multifunction printer as a scanner.
The overhead camera moves nothing
The third mechanism is a camera looking down: a phone on a stand, or a dedicated overhead unit with an arm. Nothing touches the paper at all, so nothing can be damaged, and there is no maximum thickness — you can capture a bound book, a curled roll held down with two fingers, or a receipt so shredded that it exists as four pieces arranged on a table.
What you give up is the reference plane. A sensor sliding under glass knows exactly how far it is from the page; a camera on an arm does not, and any curl, tilt, or unevenness in the paper turns into distortion in the image. Lighting becomes your problem too, because the room is now part of the optical path — glare from a window or a ceiling light lands on the glossy face of a slip and takes a line of print with it. Overhead capture is the most flexible and the least repeatable of the three, and it rewards a fixed setup over a hand-held one.
The decision rule that actually works
Sort your pile by what defeats each mechanism, not by what suits it. Take a real, representative handful of what you need to scan and ask three questions of it.
- Is any of it too fragile, too thick, or too irregular to be pulled through rollers? That part needs a flatbed or a camera regardless of what you buy for the rest.
- Is there enough of it, arriving often enough, that per-item handling would stop you doing it at all? That part needs a feeder.
- Is any of it longer than a sheet of standard paper? Long till rolls have their own constraints in every mechanism, covered in scanning very long receipts.
Most people find the answer is a feeder for the volume plus some fallback for the exceptions, and that the fallback can be the phone already in their pocket rather than a second purchase.
What the mechanism does not decide
The mechanism determines what paper you can capture; it does not determine what you get out. Two devices with identical feed paths can produce very different files, because the image processing, the compression settings, and the software that receives the scan are separate variables. A well-built feeder attached to abandoned software is a worse long-term purchase than an ordinary feeder with an open output path — see the driver is the part that expires.
It also does not decide legibility. Whether faint thermal print survives depends far more on the colour mode and exposure you choose than on whether the sensor moved or the paper did.
Where this leaves you
If you have a recurring stream of ordinary slips, a sheetfed feeder is the mechanism that matches the job, and the flatbed or camera is your exception handler. If your volume is low and your items are varied, buying nothing and using a camera on a fixed stand is a legitimate outcome — when not to buy a scanner is a real answer, not a cop-out.
What none of this settles is what form a record has to take to be acceptable to anyone else. Whether a scan can stand in for the paper, and for how long the paper needs to exist alongside it, is set by rules that vary by country and by the kind of entity you are. Put that question to your own tax authority or to an accountant who works in your jurisdiction, and choose hardware on the assumption that you may need to keep the original for a while.