Dave Jones teardown of Canon SX-100

[Duane Hess]

Surprised to see a listing on this calc. No objections, just the number of Canon people seems miniscule. Especially compared to HP. But it is a HP site!

Back in the day when office equipment stores were office equipment stores not a Staples (glorified WalMart office stuff). Not belittling Staples, but those of us old enough know the difference between then and now. Further, you likely only knew of brands/models carried by your local stores; unless you read magazines & checked the ads. Of course, of one the local deals in my town was Canon.

Thought about a SX-320 when I was in Jr. High. HOLY CRAP! Was it expensive! I now own one & it is an interesting machine. 1000 steps, 100 memories, internal to the box. 200 memories if you utilize splitting a register into two 6 digit floating point numbers (6 digit mantissa + 2 digit exponent).

EDIT: unsplit (i.e. normal usage), each register has a 14 digit mantissa and 2 digit exponent.

However, it has some tediousness in programming. Mainly it did not list program steps & instructions in the display, such as the 67/97, 1xC, etc., etc. like many of the HP line. You absolutely needed the printer.

Below is a comment from a curator, although I don't know whom or what museum they "curated". It was a message from someone who sold me my SX-320II:

" The Canon SX-320 was Canon's attempt to enter the programmable calculator market. This market was dominated by calculators made by Hewlett Packard, Texas Instruments, and Wang Laboratories, which had well-established leadership positions in the marketplace years before Canon decided to make a go of it. Canon had made less capable programmable machines in the late 1960's and early 1970's, but they were very basic, and didn't do very well in the market. The SX-320 was intended to be a high-end programmable calculator with a lot of capability. However, as it turned out, due to the well-established competition, as well as some restrictions on what Canon could do as a result of patents held by other makers, the machine didn't turn out to be a very good seller. "

I always wondered what some of those patent restrictions were.

As far as I know, the SX-100 is identical in functions to the SX-3xx models except there was no external memory box or an external BCD interface (i.e. similar to the HP-97S). The 3xx's had external typewriter, plotter, and attachment abilities to BCD instrumentation. (I/O was bi-directional--but don't know the details) Tycom made interface box(es) for the 300 series to allow "generic" BCD interfacing, not just Canon labelled peripherals. Don't know if there were other manufacturers. Also don't know the model numbers of most Canon peripheral.
EDIT: Peripherals were daisy chained via control/interface boxes external to the peripheral.

Pretty sure the 100 was significantly slower than the 300/320/320II/350. Its hard to compare speed between algebraic & RPN machines but I wrote a program on my 320 & HP-97 trying to make the logic/number of steps as equivalent as possible. The program solved roots of equations using the secant method as described in the HP-67/97 standard pak manual. The 320 has unmerged steps, so that added a significant number of steps. But the inefficiency of algebraics compared to stack machines inflated the 320 code to about 3X the number of steps. Mainly due to the number of store/recall memory commands; merged or not. However, the 320 was about 4.7 times faster than the 97 for the tests made.

For what its worth, figured you must be a Canon interested person. Have goofed with my 320 & think I've figured out how it internally stores "full-word" register numbers and the two "half-word" (left/right) numbers when you split a register. Will post or PM if you're interested.