strange HP-97A

[HP-Collection]

Hi folks

Found this 97A in my collection and wondered about the 14-pin socket on the back. As you can see it's directly wired to the mainboard. Does anyone has an idee for what this could be?
Second I wondered about the small ICs on the right bottom. They are not the same colour. Maybe a sort of memory that has been modified because of the socket?

[Accutron]

The differently colored RAM/ROM ICs are not abnormal.

[BobVA]

It appears to be mostly tapped into connector XA2P1, in particular:

Code:


1  KBA
2  KBC
3  KBD
4  KBE
5  STR
6  RCD (update - courtesty of Katie's post noted below)  connected to the via above pin 4
9  SG
10 SH
13 SF
14 SE
15 SA

The KB lines are the four that signal which of the four keyboard "rows" have been pressed.

The SA-SF lines are a subset of the LED segment lines. The ones present are the decimal point and the segments that would form an "F" if they were all illuminated. (SD, SB, SC are missing).

The red and black wires are probably power, and I'm not sure of the others.

The STR line advances the digit selection on the display and, I think, also serves as the column strobe for the keyboard, so by watching it and the KB lines I think you can tell what key has been pressed, but I'm not sure.

So, I think that connector could read the keyboard. (I don't know if those connections could are sufficient to simulate a key press.) As to why it would want to sample a subset of the display segments, I don't know. Perhaps whatever the designer wanted go know could be inferred from just those lines?

Interesting, though! Thanks for posting this!

Bob

Update - from an excellent article by Katie located here (that I should have googled first! :-) )

Quote: Inspection of the Keyboard PCA Schematic Diagram in the HP-97 service manual shows that a 1-of-14 counter/decoder chip (labeled as the cathode driver since it servers that purpose too) is used to scan the rows of the keyboard matrix.
These 14 rows times the 4 columns (KBA, KBC, KBD, KBE) account for all 64 keys on the HP-97. The 14 outputs from the cathode driver chip are normally high and only go low when the row is being scanned. If a key on that row is pressed it will complete the circuit to one of the 4 column lines.
So, pulling a column line low at the moment when the needed row is selected is equivalent to pressing the corresponding key.
In this way, keyboard input can be simulated by an external circuit. The cathode driver chip is clocked by the rising edge of STR (strobe?) line and reset by the falling edge of RCD (reset cathode driver?) line. Knowing this it is straightforward to program a microcontroller chip to simulate key presses using these 6 control lines.

So it appears to be able to write (and presumably read?) keystrokes. The subset of LED segments selected is still puzzling, though.

[Katie Wasserman]

(08-06-2016 05:44 PM)BobVA Wrote:  So it appears to be able to write (and presumably read?) keystrokes. The subset of LED segments selected is still puzzling, though.

Cool find! Bob, I completely agree with your analysis.

The subset of the segments are almost exactly what's needed to determine the number displayed using just 4-bits, with one exception. I don't see how the numbers '6' and '8' can be distinguished. It's possible that the service manual schematic is wrong and it has the labels for segment B and segment F reversed. If so, then this is exactly the 4 bits needed to decode the decimal digit and with a bit of logic one could not only simulate the pressing of the keys but also read the number on the display all through the 14 pin connector.

So this 97A was likely used as a calculation engine. It would be very interesting to know what was hooked up to it.

[BobVA]

(08-06-2016 06:50 PM)Katie Wasserman Wrote:  ...It's possible that the service manual schematic is wrong and it has the labels for segment B and segment F reversed. If so, then this is exactly the 4 bits needed to decode the decimal digit and with a bit of logic one could not only simulate the pressing of the keys but also read the number on the display all through the 14 pin connector....

Good theory! It made me look at the pictures HPC posted again. The last wire on the right (brown) is connected to a trace that *might* go to XA2P1 pin 16 (SB), which would provide the "B" segment line.

HP Collector, if you have occasion to open it up again, or a picture from a slightly different angle, could you check to see if there's a connection to pin 16?

Fun puzzle!

Bob

[Katie Wasserman]

(08-08-2016 02:59 PM)BobVA Wrote:  HP Collector, if you have occasion to open it up again, or a picture from a slightly different angle, could you check to see if there's a connection to pin 16?

I pulled out the main PCB from a donor 97 that I had around and traced it out. The service manual PCB labeling of the LED segments is entirely different than what I have on my PCB, the pins are correct but the mapping from pin number to segment is permuted for many of the segments. I suspect that there are different PCB board versions to accommodate the various iterations of the ROMs.

I stick to my guess as to what's going on and the the service manual is wrong for the board shown above.