Recently I've been wanting to expand on the pragmatic use of my V200 so I took it apart to investigate. First and foremost, the easiest way to increase the speed of the painfully slow OS is to overclock the hardware.
The bulk of the information found for this endeavor is listed here: http://mobifiles.bytefox.de/viewtopic.php?t=1812
Contrary to what everyone has stated there, the C9 0603 capacitor is not 47 pF but instead 50-51 pF. After desoldering the cap and respective resistor, R1, I used my cheaply purchased LC-100S to measure it. Yes, its an inaccurate LC meter, but with a little zeroing out while I was holding the alligator clips to account for my capacitance, it repeatably read under 0.5 pF w/o a DUT. The readings of the cap over the span of a few minutes were between 50.2 and 51.5 pF. The resistor in the RC oscillator was also a 0603 labeled "16b" which decodes to 1,430 ohms. DMM confirmed that it's within 1% of the specified value. I'll post photos later of this area of the board.
Google confirms ~13.7-14 MHz from this RC oscillator which puts the timing at 71.5-72.9 ns. The latter is just shy of the spec for the 256KB of RAM, a pair of CY62128VLL-70ZC's, but faster than the 4MB of FLASH ROM, a LH28F320BFHE-PBTLZ2 which has a timing spec of 80 ns. The calculator is said to run at 12 MHz, which fits in the operational constraints of the NOR Flash.
Though some have been able to overclock to ~20-24 MHz, the RAM is always accessed and because the OC is faster than it's base timings, its easy to see why previous attempts don't always yield consistent results. The RAM is in a TSOP-32 package. Mouser shows plenty of viable options for replacement. I mean, if I'm overclocking the MCU, you kinda have to do the RAM. I'll write a log up for this specifically a little latter. However, without those details spelled out, I can reliably move over to SRAM chips that have 45 ns timings. Not a huge upgrade, ~55%, but sizable nonetheless.
Since others have OC'd their calculators to nearly 2x faster on stock SRAM, I think that 2x will be quite achievable. If not, I'll have a few spare random 0603 caps.
That said, a 32-33 pF cap will give me 45.8-47.2 ns. Since I'm buying just one small SMD cap, might as well go for a crème de la crème electrolyte such as a C0G/NP0, though I will settle for 1% tollerance.
The choices, just for pedantic thoroughness, are:
Yageo's CC0603FRNPO9BN330 - 33 pF, 50V, NP0 1% costing $0.14
Murata's GRM1885C1H270FA01J - 27 pF, 50V, C0G 1% costing $0.15
AVX's 06035A240FAT2A - 24 pF, 50V, C0G 1% costing $0.30 (most expensive)
Murata's GRM1885C1H200FA01D - 20 pF, 50V, C0G 1% costing $0.14
Murata's GRM1885C1H160FA01D - 16 pF, 50V, C0G 1% costing $0.18
Buying these 5 different caps giving me possible speeds of ~21.2 MHz ~25.9 MHz, ~29.1 MHz, ~35 MHz, and ~43.7 Mhz will cost me $0.91. Luckily I need some other things from Mouser for $4-5 in shipping wont' be too bad.
Now the RC oscillator are connected directly to TI's ASIC that contains some hardware from the first gen of the calculators. I don't know what clock speed it is capable of running at, but many people have "reliably" put on a 20-22 pF cap on the stock hardware. Continuity testing and visual doesn't show any other capacitors in series or parallel so we'll see what happens in a couple weeks once I get my final order ready.