Although its appearance may suggest otherwise, the TI-62 appears to be a close relative of a very different-looking Texas Instruments calculator, the TI-57 LCD. Relative, yes: identical, definitely not! The TI-62 is in fact a much improved version, addressing many of the deficiencies of the TI-57 LCD model.

More memory, for starters: programs can be up to 100 steps long, if all but one memory registers are converted into program usage. Conversely, a 42-step program can coexist with 10 memory registers. Since the programming model is fully merged, this makes the TI-62 a respectable machine.

Then there's a bigger, semi-alphanumeric display that is used to show program steps in mnemonic form. It can also displays 10-digit numbers with a two-digit exponent.

The keyboard is also a great improvement: the TI-57 LCD had that hideously stiff, bouncy keyboard that Texas Instruments inexplicably used in several of their early LCD calculators. (This was corrected when the TI-57 II was released, but the TI-62's "Galaxy" style keyboard is still better.)

Overall, the TI-62 gives the solid feel of a well designed, solid, useful algebraic programmable. Maybe that explains why it is no longer in production?

The example program I wrote for the TI-62, implementing the Gamma function, is not particularly efficient: it uses a copious amount of program steps and several number registers to compute the Lanczos-approximation. The fact that such a "wasteful" program fits easily into the TI-62's memory confirms that this is indeed a respectable machine.

STO 1:   2.506628283501
STO 2:  92.20704845211
STO 3: -83.17763708287
STO 4:  14.80283193078
STO 5:  -0.2208497079533

LBL G	00
STO 0	01
EXC 2	02
÷	03
EXC 2	04
+	05
RCL 1	06
+	07
RCL 3	08
÷	09
1	10
ST+ 0	11
RCL 0	12
+	13
RCL 4	14
÷	15
1	16
ST+ 0	17
RCL 0	18
+	19
RCL 5	20
÷	21
1	22
ST+ 0	23
RCL 0	24
=	25
lnx	26
+	27
(	28
RCL 0	29
-	30
3	31
.	32
5	33
)	34
×	35
(	36
RCL 0	37
+	38
.	39
8	40
5	41
)	42
STO 0	43
lnx	44
-	45
RCL 0	46
=	47
RTN	48