A table giving the memory sizes of the first four generations of HP handheld calculators, related desktop calculators, and the HP-01 watch may be found at:

http://www.brouhaha.com/~eric/hpcalc/rom_size.html

These are the hardware memory sizes, so the RAM register count does not necessarily match the number of user-visible registers.

Eric,
Thanks for the site, interesting!
I've been a HP user starting in 1973 when the 35 came out.
(used it to take my LSIT exam) and have followed them up to & including the 33s (if it ever ships)
Never knew much "tect" stuff though.
Danny

Just a question .....

The HP21 has a memory ? or not ?

So, why no RAM ?

Big FAT

As it says in the introduction on the page, the table only counts memory separate from the CPU chip.

The 67 is missing. Is it the same as the 65?

tm

Quote:

---

The 67 is missing.

---

No.

Quote:

---

Is it the same as the 65?

---

No.

I see that I made a mistake in viewing the categories. So the 67 is a Woodstock and not a Classic?

tm

Mechanically it's extremely close to the HP-65. The electronics, however, is basically the same as Woodstock with the exception of the display driver and card reader. So in a sense it is closer to being a Topcat without a printer.

The only memory on the HP21 is the stack and storage registers. There is no program memory or RAM chip.

Just curious: Where did the early financials (HP 80 & 70) keep the financial data (N,I,PV,PMT,FV)for their primitive TVM solvers. As you state, there is no RAM...

Quote:

---

Just curious: Where did the early financials (HP 80 & 70) keep the financial data (N,I,PV,PMT,FV)for their primitive TVM solvers. As you state, there is no RAM...

---

Good question!

The RAM size for the HP-70 should not have been listed in the table, as I have no information on it. I've been reluctant to disassemble my HP-70 as the back label is pristine. On the other hand, one column of keys is not working, so I need to repair it anyhow.

In the HP-80, some of the chips are in a hybrid module, so I can't tell exactly what chips are present. I was basing the information in the table on U.S. Patent 3,863,060. However, the code in the patent may not be the same as that of the final product. For instance, the source code listing in the patent has 1.75K of code (seven ROM chips), which the article "Microcode: Electronic Building Blocks For Calculators" (Hewlett Packard Personal Calculator Digest, Vol. 3, pp. 4-6, 1977) states that the HP-80 contains two quads worth of ROM (2 Kwords, which would be eight of the 256-word chips).

The source code published in the patent simply keeps the financial data in the stack, and will produce incorrect answers if you perform intermediate calculations using more than two stack levels between the entry of two financial variables. I do not have an actual HP-80 close at hand, but I suspect that this behavior is not present in the released software. It is reasonable to expect that the real HP-80 contains an extra RAM chip, and that it may well have more ROM code than listed in the patent in order to deal with the RAM variables. I won't know for certain until I dump the HP-80 ROM code. So for the present, I have removed the RAM information from the HP-80 table entry as well, and added a footnote about the 1.75K vs. 2K question.

The HP-80, HP-80, HP-22, and HP-27 have what I call "old-style TVM", which differs significantly from the TVM capabilities of modern financial calculators in two ways:

1) Only three variables are used as input to the old-style TVM, with one or two remaining variables calculated as output. See the table on page 35 of the HP-70 Owner's Handbook (p. 19 of the PDF file on the MoHPC CD-ROMs) for a list of the valid input and output combinations.

2) The old-style TVM does not use the sign convention to distinguish incoming and outgoing cash flows. The PV, PMT, and FV values are normally all positive.

The HP-92 was the first calculator to introduce the new-style TVM used on all subsequent HP financial calculators (HP-37E, HP-38E/C, HP-12C, etc.). The new-style TVM solves a single five-variable equation (requiring four known values and one unknown), and uses the sign convention for the direction of cash flows.

Eric: if mine was a good question, yours is a more-than-good answer! :-)

BTW, I recall a HP Journal (May 1973?) where the HP80 was introduced, and it stated that the hybrid has 7 ROM chips, instead of only 3 ROM chips in the HP35. So it seems to favor the 1.75 position, and there was no mention about RAM chips, as far as I can remember.

The reason why there was more ROM in the first financial model than in the first scientific model may fuel the "engineers vs. MBAs" discussion, what I don't intend to do... but I'm on the engineers side, just in case.

Mr. Smith.

You didn't answer my question. I think the 67 is part of the Classic family, not a Woodstock, at least according to W.A.C. Mier-Jedrzejowicz.

tm

Quote:

---

You didn't answer my question. I think the 67 is part of the Classic family, not a Woodstock, at least according to W.A.C. Mier-Jedrzejowicz.

---

Well, let's look at that question again:

Quote:

---

So the 67 is a Woodstock and not a Classic?

---

I think I did answer that. In fact, I think I gave two answers, which can be more concisely summarized as "yes" and "no".

Eric

Electrically the HP67 IS a Woodstock. Its ACT chip is interchangeable with the HP21/25/29/97. The only thing classic about it is the case/charger/battery.

I've now added HP-70 information to my memory size table and chip table.

The HP-70 has two quad ROMs (2K words of ten bits) and one RAM chip of the same type as the HP-45 (ten registers of 56 bits).

Later model HP80 machines did not have the hybrid module.

Also those inductor packs in the classic machines are not the same. The inductor connected to the decimal point has a different value... that is why there are two different part numbers.

Quote:

---

Later model HP80 machines did not have the hybrid module.

---

I imagine they would have switched to using two quad ROMs when they became available. That would reduce the cost significantly, and allow the use of the same logic board as the 45 and 70.

Can someone with an HP-80 that doesn't use a hybrid please make a high-resolution photo or scan of the logic board available?

I just looked up the HP Journal articles on the HP-80 and HP-92. The latter article confirms that on the HP-80, the TVM functions require that the known variables be entered in a left-to-right fashion, and are only stored in the stack. This explains why the HP-80 did not need a RAM chip. The HP-70 does have a RAM chip; it and all later HP financial calculators do store the TVM variables and do not require the known variables to be entered in any particular order.