Installing the battery turned out to be quite easy. Both the single- and double-height batteries have a 70mm x 38mm footprint. I cut a 72mm x 33mm opening in the lower half of the case (no tragedy if the experiment failed as I had converted it to a red case and had a spare black bottom). The opening has to be 2mm longer than the battery because the battery cover pushes the battery 2mm inside the case.
I made the opening 5mm narrower than the battery and cut 3mm deep slots on each side of the double-height battery for the cover to fit into, holding the battery securely in place. I started the slots in the battery with a Dremel and polished them up (somewhat) with a hacksaw. The slots go down to white plastic; they are between two cylindrical batteries and therefore nowhere near cutting into anything important. The main thing at risk seems to be a pc board inside the battery at the contact end, which a 5mm deep slot might imperil.
In use the unit balances on the battery when set down with the lid tilted back a bit. It's too wobbily to type on that way, I put it on my lap or slip a book under the unsupported half.
I run Linux off a compact flash card so as to free up both PCMCIA sockets. Leaving the unit on and running X-windows, with nothing in the PCMCIA sockets, the battery took 5 hours and 4 minutes to go from 100% full to 5% where it suspends.
The discharge rate as registered on the front-panel LCD was not linear. It went to 30% in 2 hours 40 minutes, then stayed at 30% until 4 hours 40 minutes, then went to 5% in the last 20 minutes. /proc/apm (Linux's copy of the APM-BIOS-supplied battery data) showed the same numbers as the LCD throughout.
The standard battery (shown on the right beside the double-sized NP-750) still works fine and is held in place (picture on the left shows it installed) not by the cover so much as by a little black plastic thingy that also holds one of the backup batteries.
Incidentally I've also been using a pair of NP-750's in my Hitachi VisionBook Traveler, a 133MHz Pentium-based notebook with 48MB RAM and a 3GB drive ( http://boole.stanford.edu/travdisk.html describes how I upgraded the Traveler's standard 1GB drive to 3GB, a nontrivial surgical operation). The battery upsizing took a more minor modification than above: the Traveler has a cover over each battery that clips off to make room for a double-height battery. The manual says nothing about this, and I did have to cut short slots in the batteries at the contact end, at the same height and depth as the slots I cut for the PC110 but only 5mm long, to accommodate black plastic thingies that I didn't feel like breaking off (though that would have been the simpler solution). I held both batteries in place with a couple of big rubber bands that went round the whole unit---bit kludgy but it worked fine. On a trip to Australia last month I got just under 5 hours of steady vi editing in X-windows done on one charge!
On the right is the motherboard of an IBM (Japan) Palmtop PC110, hung out on
the line (a rod sticking out from my bookcase five feet above my desktop and
about two feet to the right of the keyboard I'm typing this on). The case,
display, keyboard, modem, and battery have all been removed, leaving
just the motherboard with the following plugged into it:
Power-supply daughterboard, with 10.5V coming from the adaptor.
2-slot PCMCIA assembly, currently holding just a LinkSys ethernet card, without which you wouldn't be able to ping it.
16MB memory board (plus 4MB onboard = 20MB RAM)
15MB SanDisk CompactFlash card (contains Linux)
Clock backup battery.
Visible in the picture are the upper side of the motherboard, with the compactflash card at the bottom, the PCMCIA unit above it with an ethernet card (hidden inside) and transceiver above it, and the power board off to the right with a cord from the 10.5V supply going into it. The empty space at the bottom right is where the battery normally goes. The green thing dangling down is the 2.7V backup battery (distinct from the coin cell at the back, I'm not clear how these backup batteries divide up their responsibilities).
The picture to the left shows the bottom of the board. The add-on 16MB RAM is at the top right.
The picture to the right shows the board relative to my desk, it's hanging way up there on the right.
The picture to the left shows that a skeletal PC110 can still be used despite having no keyboard or display. It's running as a client of my P-II-300 Linux box's X-server at home (the monitor whose outline you can just make out). The window on the left is running on Boole at Stanford (where the web page you're reading lives), while that on the right (displaying a Latex2html paper residing on Boole, http://boole.stanford.edu/parikh) is being displayed by Netscape 1.12 running on the PC110 (confusing, huh?).
I used the cellular modem a lot at LICS (Logic in Computer Science) at Indianapolis in June, and then on vacation at Martha's Vineyard. Ran up a couple of hundred dollars in cellular charges. If I took more trips it might have been cheaper to get a CPDP account, except I doubt either of those locations have CPDP. Neither location seemed to have the *DATA (*3282) TX-CEL-capable banks of data modems that we have here in the SF Bay Area. [Which have since gone away, as of mid-1999 sometime.]
Also Amanda Walker's PC110 page is at http://www.alfar.com/~amanda/pc110/