Posts: 121
Threads: 21
Joined: Jun 2008
In the course of testing a fix to my HP97 emulator, I ran across an interesting glitch on the real HP97. Maybe someone out there can verify it.
1. Calculate N! using 70 - creates display overflow.
2. With Print mode in TRACE - Press SCI
This displays on printer:
"9.999999999 99 ***"
FIX displays the same as above.
3. Now, Press ENG
This format prints
" 9.999999999+9 ***
Actually the last 9 has a glitch - looks like a badly shaped 7
printed over it
Can anyone confirm?
Thanks,
Mike
http://www.limpidfox.com/hp97.htm
Edited: 9 July 2008, 11:51 p.m.
Posts: 291
Threads: 43
Joined: Jun 2007
Hi Mike,
Indeed, I got the same results as you did on my 97, with a display overflow generated through any means. I also noted that non-overflow values which incorporated the 99 exponent (such as 8.00+99) printed correctly.
For what it's worth, my 19C had no such issues, printing overflows correctly in all display modes. :)
Best regards, Hal
Posts: 122
Threads: 9
Joined: Jul 2007
Mine does not show this behaviour.
It always shows 9.999999999+99 ***
Or have I misunderstood your instructions?
Posts: 2,309
Threads: 116
Joined: Jun 2005
I can easily imagine two different reasons why this bug might appear on some HP-97 units and not others.
There were two firmware changes made during HP-67/97 production. ROM 0 (stored in the ROM/Anode Driver chip) was revised to fix incompatibilities with some revisions of the ACT (processor) chip. The original chip in the HP-97 was p/n 1818-0225, and it was replaced by 1818-0267.
ROMs 1 and 2 were revised as a pair to correct mathematical errors. Both the HP-67 and HP-97 originally used p/n 1818-0228 and 1818-0226, which were replaced by p/n 1818-0550 and 1818-0551, respectively.
It is possible that one of these changes either fixed or introduced the bug. The ROM 0 change is the more likely candidate. However, most of the printing code is actually in ROMs 5 and 6 in the 97.
HP had two vendors for the "PICK" chip that controls the printer in the Topcat series and 19C, AMI and Mostek. There are known to be functional differences between the two, so this is another possibility.
Posts: 1,477
Threads: 71
Joined: Jan 2005
Mike,
Very interesting find!
I tried this on two 97S machines and two 97 machines and all four calculators have the problem. I also tried this on a 91 and it doesn't exhibit this behavior.
It appears that (and sounds like) the print head sort of stutters in it's motion. It wonder if this was a known bug at some point before now, it's really obvious when it happens.
-Katie
Posts: 256
Threads: 4
Joined: Sep 2007
What are the serial numbers involved?
Regards,
John
Posts: 291
Threads: 43
Joined: Jun 2007
My 97 (with bug) is 2104B7022D (Brazil)
...hal
Posts: 121
Threads: 21
Joined: Jun 2008
Let's see - wow - interesting that it's been found on other 97 models as well.
Thanks to everyone for checking this out.
My HP97 serial number is 1703A00109.
Mike
http://www.limpidfox.com
Edited: 14 July 2008, 12:28 a.m.
Posts: 153
Threads: 7
Joined: Jun 2008
Mike --
My '97 exhibits the same behavior. Its S/N is 2001S92329. Changing the DSP setting from 2 to 9 returns the number to non-buggy format.
Here is a whole-tape photo and a bug closeup.
To those who did not find this bug in their '97: You might find it if you turn on the '97 in MAN mode, then follow the original procedure exactly as given.
-Joe-