HP-71B to sacrifice in the name of SCIENCE!

[brouhaha]

Since a good photomicrograph of the 1LK7 Saturn CPU chip (as used in the 18C, 28C, and the late 71B) has recently been made available, I thought it might be nice to find the earlier 1LF2 CPU chip (used in the early 71B) to desolder, and send off for decap and photomicrograph. I'm expecting them to be very similar. The 1LK7 adds two instructions for Saturn architecture level 1, so I expect that the 1LF2 instruction decode PLA will have a few less terms. The 1LK7 might be in the newer HP CMOSG process, whereas I think the 1LF2 was in CMOSC. The I/O buffers and analog circuitry might be different.

I have been hoping to find a cheap non-working 71B to sacrifice, but I've only seen TAS listing for working, for high prices, and untested, for nearly as high prices.

For a unrelated reason, I just went through boxes of calculators that I brought when I moved in 2012, and which have been almost untouched since. (Unfortunately I wasn't able to bring my entirecollection.) I do have four 71B, with serial prefixes of 2523A, 2635A (which had had a rough life), , and two 2845A. The 2845A both have 2CDCC ROM, and probably 1LK7 CPUs. The 2635A has 1BBBB, and probably the 1LF2.

While all four were working when I packed them away, the 2635A now shows no signs of life. I think it probably has 1BBBB with a 1LF2, but I haven't yet opened it. It is in physically pretty good shape. I'm thinking about sacrificing the CPU chip; hopefully there is little or no ESD damage to it.

Based on the appearance of the bottom plate, I think it has the older style construction, compared to the other three, so I'm not yet sure whether I can swap the housing with the 2635A. But at least I may wind up with spare hybrids from inside the 2635A, if they're not damaged.

Anyhow, I'm potentially interested in broken or beat-up 71B and accessories (especially. HPIL module) for decap and photo. The HPIL module is especially interesting because it has a custom HP microcontroller with HPIL, which provides the mailbox interface over the Saturn bus. It's may be related to the processor in the Thinkjet.

I used to have a lot of HP-IL cables, and am dismayed that I now only seem to have three. Sigh.

[KeithB]

You can't have mine!

[brouhaha]

I just discovered that my 71B with S/N prefix 2635A (metal back, 1BBBB, presumed 1LF2 CPU) actually works from an 82059B AC adapter. It fails to work from alkalines. There's no sign of battery contact corrosion. Since it's functional, I suppose I should take it apart to diagnose and repair it, rather than part it out. So I'll still be on the lookout for a broken early 71B to sacrifice for 1LF2 chip decap and photomicrograph.

In the mean time, this means that I'll be able to do some testing of the differences (or lack thereof) between the 1LF2 and 1LK7. I expect that the undocumented behavior of BCD arithmetic on non-BCD values is probably identical. It will be interesting to determine what length the 1LF2 thinks the level 1 opcodes are, and what (if any) affect they have on registers.

[J-F Garnier]

(03-10-2024 12:28 AM)brouhaha Wrote:  I expect that the undocumented behavior of BCD arithmetic on non-BCD values is probably identical.

I guess so, because some parts of the HP-71B OS depend on this behavior !

Quote:It will be interesting to determine what length the 1LF2 thinks the level 1 opcodes are, and what (if any) affect they have on registers.

I already checked that the PC=(A) opcode is correctly fetched on the 1LF2 and doesn't break the opcode flow.
I used it to identify the CPU type by software on the 71B (here).

The opcode for PC=(A) is 808C and is part of the 808x group, in the 1LF2 only the 8080 INTON and 808F INTOFF are defined.
So it depends on how the opcode decoding is done in the 1LF2, but I would expect the 808C opcode to be executed as INTOFF.

On the other hand, I always wondered why the RSI opcode is 80810 and not just 8081.
Due to this 5th nibble, an attempt to execute RSI on a 1LF2 will probably put the CPU out of sync with the next opcodes.

J-F