HP-71B Versions and Bugs

[J-F Garnier]

Hi HP-71B fans!

I updated my page about versions and bugs of the HP-71B, especially regarding version 1AAAA. I found a lot of informations in the HP IDS (Internal Design Specification) for version 1BBBB. Actually, I didn't review the 3000 pages, but I made a binary comparison between the 1AAAA and 1BBBB ROM. Each time I found a difference, I checked the corresponding IDS pages and very often comments explained the reason of the change.

I don't have the IDS for version 2CCCC (it exists, let me know if you have it!), and I was not able up to now to make the same study for 1BBBB-2CCCC differences. However, I checked the fixes of some 1BBBB bugs, and give a description of an interesting bug, still in 2CDCC ROM, related to "funny functions" (Math FNROOT, INTEGRATE for instance) and user-defined functions.

Many thanks to Gordon Dyer, for the lending of his 1AAAA machine some times ago.

J-F

[Valentin Albillo]

Hi, Jean-François:

Thank you very much indeed for your excellent report on
HP-71 bugs, impressive work both illuminating and useful. As for other 71B bugs not mentioned, may I remind you the
one I discovered for the Math ROM, details of which can be found
in my Datafile article HP-71B: "Math ROM Baker's Dozen (Vol. 1)", namely that trying to find out a real root (or ascertain that none can be found) of x² + 1 = 0 by executing:

     FNROOT(0, 0, FVAR * FVAR + 1) 

does *not* terminate (at least not this side of geological times) as it should. I've seen this happen in 71B BBBB and CDCC, always with Math 1A (is there any other version ?). It doesn't happen with the HP-15C equivalent SOLVE routine.

Perhaps you would consider interesting to find out what FNROOT is doing in this particular case and why it doesn't exit in a reasonable time. I for one would be very interested to know and would be very obligued if you could shed some light on this subject. Certainly only a Saturn assembler guru like yourself can succeed at this endeavour. :-)

Thanks again and best regards from V.

[Rodger Rosenbaum]

Quote:

---

Hi, Jean-François:

Thank you very much indeed for your excellent report on
HP-71 bugs, impressive work both illuminating and useful. As for other 71B bugs not mentioned, may I remind you the
one I discovered for the Math ROM, details of which can be found
in my Datafile article HP-71B: "Math ROM Baker's Dozen (Vol. 1)", namely that trying to find out a real root (or ascertain that none can be found) of x² + 1 = 0 by executing:

     FNROOT(0, 0, FVAR * FVAR + 1) 

does *not* terminate (at least not this side of geological times) as it should. I've seen this happen in 71B BBBB and CDCC, always with Math 1A (is there any other version ?). It doesn't happen with the HP-15C equivalent SOLVE routine.

Perhaps you would consider interesting to find out what FNROOT is doing in this particular case and why it doesn't exit in a reasonable time. I for one would be very interested to know and would be very obligued if you could shed some light on this subject. Certainly only a Saturn assembler guru like yourself can succeed at this endeavour. :-)

Thanks again and best regards from V.

---

Valentin, I happened to notice that if you execute OPTION ROUND POS first, then it will terminate