HP 15C Combination Curiosity (bug?)

[Eddie W. Shore]

Let: R1 = 45, R2 = 5, R3 = 3

Problem 1:

COMB(R1, R2) + COMB(R1, R3)

Code:

Code:

LBL A 
RCL 1
RCL 2
Cy,x
RCL 1
RCL 3 
Cy,x 
+
RTN

HP 15C Collector's Edition: 1,235,949 (no flash)
HP 15C Limited Edition: 1,235,949 (no flash)
HP 11C: 1,235,949 (no flash)

Problem 2:

COMB(R1, R2) × COMB(R1, R3)

Code:

Code:

LBL B 
RCL 1
RCL 2
Cy,x
RCL 1
RCL 3 
Cy,x 
×
RTN

HP 15C Collector's Edition: 1.7337e10 (no flash)
HP 15C Limited Edition: 1.7337e10 (no flash)
HP 11C: 1.7337e10 (no flash)

Problem 3:

COMB(R2, R3) × COMB(R1 - R2, R2 - R3) ÷ COMB(R1, R2)

Code:

Code:

LBL C
RCL 2
RCL 3
Cy,x
RCL 1 
RCL 2 
-
RCL 2
RCL 3 
- 
Cy,x
×
RCL 1
RCL 2 
Cy,x
÷
RTN

HP 15C Collector's Edition: 0.0006 (flashes)
HP 15C Limited Edition: 0.0064 (flashes)
HP 11C: 0.0006 (no flash)
(similar test on a HP 32SII: 0.0064 (no overflow error))


Problem 4:

COMB(R1, R2) ÷ COMB(R2, R3)

Code:

Code:

LBL D
RCL 1
RCL 2
Cy,x
RCL 2
RCL 3 
Cy,x 
÷
RTN

HP 15C Collector's Edition: 122,175.9 (no flash)
HP 15C Limited Edition: 122,175.9 (no flash)
HP 11C: 122,175.9 (no flash)


Problem 5: Same as problem 3 but store R4 = R1 - R2, R5 = R2 - R3

LBL E

Code:

RCL 1
RCL 2
-
STO 4
RCL 2
RCL 3
- 
STO 5
RCL 2
RCL 3
Cy,x
RCL 4 
RCL 5 
Cy,x
×
RCL 1
RCL 2 
Cy,x
÷
RTN

HP 15C Collector's Edition: 0.0064 (flashes)
HP 15C Limited Edition: 0.0064 (flashes)
(work around did not appear to work)
HP 11C: 0.0064 (no flash)


HP 15C: I don't know what causes the display to flash on problem 3 and 5. When I enter the key strokes manually, I don't get the flashes.

HP 15C Collector's Edition: 00935 (2023 - Philippines)
HP 15C Limited Edition: 01535 (2011 - China)
HP 11C: Serial # 2416A74279 (USA)

[Gene]

Edited from original posting! This is what seems to be the case as of the posting. Reserve the right to change or modify if something changes!



Hey Eddie !

For problem 3, the values you posted have an extra zero in the answer for the 15C CE and 11C compared to the answer for the 15C LE. There is an extra zero.

Congrat's to Eddie Shore for finding this.

What happens:
If you run a program with GSB label, the result will be returned correctly but in some circumstances (apparently doing some combination calculations), the display will be blinking.

If you go a GTO label and then R/S for the same code, there will be no blinking.

Good news:
1) Correct result is returned - no sign the a wrong result bug.

2) Blinking is NOT a result of flag 9 being set.

The code Eddie gave does not cause an overflow and flag 9 is clear. I tested this by putting in another label and called the routine that causes a blink by a GSB in the code.

LBL C
Eddie's example that causes a blink with the combination function
RTN
LBL 0
GSB C
RTN

Display is blinking.

Change the code to:
LBL 0
GSB C
F? 9
GSB 9 (which does not exist)
RTN

Code stops at the RTN with blinking display. Flag 9 (error blinking flag) is not set. If it were, the code would have tried to branch to label 9 and given an error.

In fact the blinking stops when you press the blue shift key in this situation. When there is an overflow and flag 9 is set, it stays blinking until you clear flag 9 or press ON, backarrow, etc. Flag 9 is not affected by this.


Bad news:
1) The display should not blink. We knew in advance that blinking was a possible problem - see below and an earlier post in this forum.

I would guess that the running display blinking is not quite turned off properly in one place somewhere and this example times it just right. It is a cosmetic thing but misleading as one would think an overflow had occurred.


REPOST below.

Here is a repost of one of the known differences between the 15c original and the 15c CE. There is a thread in this forum describing what was known prior to shipment. It appears this issue Eddie has found fits into this section:

Display

The original HP-15C had fixed display contrast. The contrast on the HP 15c CE can be adjusted by following the steps on page 63 of the Owner's Handbook.

The behavior of the flashing "running" display is also slightly different on the HP 15c CE. Sometimes the "running" text will flash at a different speed than on the original HP-15C. It may even not flash at all, for instance, during loops of complicated MATRIX instructions (50 repetitions of 8x8 determinants which takes about 20 seconds to run is one example), but given the processor speed of the HP 15c CE, this is unlikely to be noticed with any user program. It does not affect any computational results.

Key handling during “temporary displays” is different on the HP 15c CE. Temporary displays are when you press [g] [MEM] to temporarily display the memory allocation, when you press [f] [PREFIX] to temporarily display the full mantissa, or when, in complex mode, you press [f] [(i)] to temporarily display the imaginary component. In these three cases, while the temporary display is up, key presses are accepted on the HP 15c CE rather than ignored as they were on the original HP-15C. If you press a single digit key, the digit you pressed will be displayed and flashing. Pressing any additional key will stop the flashing. Since this behavior is not intended on the original HP-15C, the best way to avoid this difference is not to press a key during a temporary display of course.

[Eddie W. Shore]

Very interesting to know, Gene. I'm glad that the overflow error is not being caused. I will remember to run labels using the GTO, label, R/S from now on. This method helped solve (for lack of a better word) the flashing problem with Problem 3 on both the Limited Edition and Collector's Edition.

What happens if I try the GTO, label, R/S method with the PSE instruction?

Code:

Code:

LBL E
1
ENTER
9 
+
PSE
PSE
PSE
5
ENTER
6 
×
PSE
PSE
PSE 
1 
RTN

The 15C Collector's Edition works correctly (the PSE is fixed!).


Not so much with the Limited Edition:
f [ E ]/GSB E: 3 (?) displays, it blanks, it shows 30
GTO E R/S: 3 flashes (the 10 is ignored until it again), and then 30 flashes
In neither case 1 is shown.

Thank you Gene for looking into this!