I've got a lot of scanners (they are useful in DTP). But, until recently, all were effectively "single sheet" devices (even those with feeders weren't particularly fast). They've been more geared to oversized scanning (tabloid as well as up to 40" wide media), high resolution (film scanners) and "one off" page scanners (keep a digital record of your real estate tax assessment, etc.)
A friend recently lent me a "really fast" (her words) scanner (Brother ADS-3600W). This was an eye-opener!
The 3600 will do ~50 sheets per minute, double-sided (so, 100 "pages" per minute). Documents must be reasonably narrow, though (letter/legal/and smaller -- 8.5" width)
The feeder will accommodate 50 sheets, though you can add to it as sheets are processed without interrupting the flow. This is probably not advised, though (see below).
It has WiFi and a wired 100/1000 NIC. A USB3 port connects to a PC. A USB port is available to scan directly to a thumb drive. The UI is via a small, color touch panel.
The only "supplies" are replacement rollers (after 200,000 sheets).
I opted for a 2400N which omits the WiFi, downgrades the USB to USB2 (which can be a bottleneck at higher scanning resolutions/page sizes) and replaces the touch panel with a set of programmable LPBs.
[My thinking: considerably cheaper (I paid $150 NIB vs closer to $400 for the 3600), I don't use WiFi and have wired drops all over the house, I'm not likely to want to have a dedicated PC to service this peripheral and 30 vs 50 ppm is a small concession for an activity that I don't plan on performing continuously. Also, a failure in the touch panel would mean the device would *have* to be operated from a PC whereas a broken button could be easily hacked!]Looseleaf sheets/manuals are easy to process. I've set up one of the LPBs to signal "600x600 dpi, 24b color" and use this to scan the front and back covers (both sides of each) as they are often produced in color even if the content is B&W. A second LPB is set for "600x600 dpi, monochrome" (it dithers well so most greys are reproduced) which I use to scan the contents of the manual.
[600dpi seems to be about the low end for reliable OCR post-processing; for documents with very small typefaces, I scan at 1200x1200. For B-size foldouts, I scan on a flatbed scanner and import those pages individually to the document during reassembly.]Both buttons route their output to an FTP service that runs, here, 24/7/365 so I can pick up the output without having to take any special steps. (I will set up a third option for SWMBO to scan to an SMB share on that same host as that's easier for her to deal with).
Each option causes ALL of the media to be scanned into a single PDF/A. I believe there are options to create a document per sheet but that's contrary to how I use it (I could always feed a single sheet in and end up with a single page PDF!)
This choice has consequences; if the scanner gags on anything (e.g., a jam -- see below), then the entire document is scrapped. If sheet 49 of a 50 sheet document chokes, then you have to rescan ALL the previous sheets, again!
[IIRC, the scanner has 256MB of internal memory; 512MB for the 3600W. I've never encountered a PDF that approaches that size]Jams have been an issue, for me, as I have been "ripping" books (I have a large paper "guillotine" that cuts off the binding, converting the book to front+back cover and a bunch of loose sheets). For perfect bindings, this often results in some glue being left on the edges of *some* pages (particularly the pages adjoining the covers -- have a look see, for yourself!).
If, like me, you try to maximize the paper size fed into the scanner (to minimize risk of something close to the binding edge being "cut off"), then there is an increased chance of this glue causing two adjacent pages to stick together along that edge (the glue often seeps *between* the pages during the binding process).
Note that the adhesion may be slight, but enough to cause those two pages to be fed concurrently. The scanner can detect this change in thickness and inhibit the feed. (if you have a Post-It note on a page, it will detect that, as well, regardless of where on the page it is encountered!)
It is surprisingly difficult to ensure that no two sheets are stuck together, e.g., by "fanning" the pages.
Another consequence of this glue residue is that it can transfer to the front/back CIS in the scanner -- where it must be removed with a weak solvent (alcohol seems to work).
Once the document/manual/book has been scanned, I fetch the (multiple) PDFs from the server. Each carries a name of the form <foo>_<date>_<time>.pdf (this can be changed) so I know the order in which the files should be reassembled.
Select the files that comprise each document (thumbnails are helpful to discover the cover of the NEXT document)
Context menu: "Combine files in Acrobat"
Acrobat: "Organize Pages" Review pages for any that stand out as "odd" Drag rear cover (recto and verso) to reposition after the last scanned page. "Save as Other | PDF/A"
Every 5,000 sheets (10,000 images), the scanner throws a "maintenance alert" to indicate cleaning is needed. Wipe out the paper dust, clean the rollers with a damp cloth. Wait to air dry.
I've not yet had to replace the roller assemblies and imagine I will largely finish my scanning before that time comes (200,000 sheets -- 400,000 pages!)
There are lots of other features -- but those likely only have value in a more traditional "office" (e.g., "scan-to-email" as a FAX emulator)