[TI-59/58] Percentage calculation.

[Dieter]

(06-01-2016 10:52 AM)hp41cx Wrote:  08 93 .
09 00 0
10 01 1
11 82 HIR
12 42 42

OK!
In a real TI-58C.

Very interesting. OK, the 58C has a ROM that's different from the 59. Since it appeared quite a while after the 58/59 were released this could be a kind of bugfix.

I did some more tests on TI59.EXE for Windows, using a short test program:

Code:

000  76 LBL
001  11  A
002  32 x<>t
003  01  1
004  82 HIR
005  01  01
006  32 x<>t
007  82 HIR
008  41  41
009  82 HIR
010  11  11
011  92 RTN

This loads the first HIR with 1 and multiplies this by the input before the register is finally recalled. So if everything works as expected the output should match the input.

Here is what I got:

Code:

input          output
---------------------
1       [A]         1
3       [A]         3
4,5     [A]       4,5
635,7   [A]     635,7
-17     [A]       -17

0,1     [A]        10
0,0123  [A]       123
0,00005 [A]    500000

So in numbers with negative exponent the sign of the latter is ignored. 0,1 = 1E-1 becomes 1E+1 = 10, and 5E-5 becomes 5E+5 = 500000.

But wait, it's getting even better:

Code:

input                   output
------------------------------
100000      [x^2]        1  10
            [1/x]  ,0000000001
       [÷] 10 [=]        1 -11
            [A]          1 -11
            [x^2]        1 -22
            [A]          1 -22

So very small values are handled correctly while for larger ones the sign of the exponent is ignored. Further tests showed that the limit seems to be 5E-11. See below.

Now let's enter values below 1 in EE mode.

Code:

input          output
----------------------
1 EE -1  [A]     1 -01
5 EE -5  [A]     5 -05
3 EE -99 [A]     3 -99

So in EE mode everything sees to work fine!

Now let's try ENG mode

Code:

2nd ENG

input          output
---------------------
123     [A]   123  00
1234    [A] 1,234  03
0,123   [A]   123 -03
0,00005 [A]    50 -06

INV ENG       0,00005

So ENG mode seems to work fine, too.

All in all it turned out that the problem with the ingored exponent sign only occured if the number was displayed without an exponent. That's why 5E-11 still shows the problem (display is rounded to ,0000000001) while 4,99E-11 has no problems (display is 4,99 -11). Is this strange, or what?

BTW, I just discovered some other odd behaviour, not sure whether it's the same on a real hardware 58/59/58C:

Code:


input         display
---------------------
123 EE 15 =   1,23 17
    EE 5      1,23 75
       9      1,23 59
       =      1,23 59
    EE 0      1,23 90
       3      1,23 03
       =                    // this was originally missing
INV EE        1230

So even after a number has been entered or calculated, pressing EE seems to re-edit the exponent.

Dieter