HP 19B teardown part 1

[brouhaha]

Someone here in the forum was kind enough to give me a broken HP 19B. I've taken apart first-generation clamshells (18C, 28C) before, and their construction is documented in HP Journal articles, but I haven't previously seen the innards of second-generation clamsheels (19B, 19BII, 28S) which use the Lewis (1LR2) chips with industry-standard bytewide SRAM.

I carefully peeled away the labels and Richard Ottosen drilled out the heat stakes using a Dremel drill press and a drill bit which he put in a small metal tube to alleviate problems with flexibility of the bit:

https://www.flickr.com/photos/22368471@N...193666469/

Once the case was opened, one side of the PCB was revealed, showing one normal SOIC static RAM, and the back side of the dice of two custom HP "Lewis" chips (1LR2), which are attached to the other side of the board using Tape Automated Bonding (TAB), similar to some but not all Pioneer models:

https://www.flickr.com/photos/22368471@N...193126379/

It is the other side of the Lewis dice that is of interest, so I'll have to remove the LCD display, held in place by a metal frame with folded tabs. Even though the bottom of the dice are bare, the top are most likely covered in epoxy, so decapsulation in fuming nitric acid will still be necessary.

[rprosperi]

(12-04-2014 10:55 PM)brouhaha Wrote:  ...even though the bottom of the dice are bare, the top are most likely covered in epoxy, so decapsulation in fuming nitric acid will still be necessary.

I just love a hobby where statements like this can be made, are understood by the readers, and may even bring about other similar suggestions.

You go Eric!!

[brouhaha]

I've taken the 19B apart further, revealing the top side of the PCB iwth the TAB-packaged Lewis chips.



That photo and a few others have been added to the album:

https://www.flickr.com/photos/22368471@N...193126379/

[jebem]

Great work so far, thanks for sharing these excellent photos and information.
This is going to be a truly good reverse engineering work, isn't it?

Many have the background knowledge, but just a few dare to go deep dive into this hobby. I hope to keep learning from you, Eric.
Every time I enjoy your technical presentations (published as video media in YouTube), and of course I can't wait to see the next RPN calculator project alive

As a side note, I can not resist to do a small parody with your sentence:
(...) decapsulation in fuming nitric acid will still be necessary.
I too had my time with plenty of fumes and acid around me as a teenager at school in the hippies golden era. Fortunately I managed to stay away from those harsh environments.
But I couldn't run away from my Master Teacher (Mr. Gomes) in Electricity Workshops classes, where he patiently demonstrated us (the little savages that we where at that time) several techniques to manipulate the materials, including the proper use of some aggressive chemicals.

[brouhaha]

Good news:

G1) It turns out that Travis Goodspeed has already decapped and photomicrographed an HP Lewis chip from a 28S. This is the same chip, with different ROM contents, as is used in the 19B and 19BII. He was kind enough to share the mulitple gigabytes of raw images with me.

Bad news:

B1) It appears that the process geometry is probably too fine to support optical ROM extraction such as Peter Monta did with the HP-35 ROMs.

More good news:

G2) I don't need to finish setting up my decap lab, at least for this project. I may still want to do it for other chips.

G3) I have another plan for how to do it electrically. Extracting the contents of the slave Lewis should be easy, as it has to provide its ROM contents over an external bus to the master Lewis. Having half of the contents isn't useful in and of itself, but once it's done I should be able to cut traces and insert an FPGA-based Saturn ROM emulator between the two Lewis chips. As long as I can indentify some key sequence that results in a transfer of execution to the slave ROM, I should be able to inject code that can dump the master ROM.

As this point I think it will be best to experiment with an HP-28S to prove that this can be done before moving on to the 19B and 19BII; since the HP-28S code is already known, that will make it much easier to experiment with.

If anyone has a spare HP-28S that they're willing to sacrifice in the name of science, please let me know.

[Katie Wasserman]

(12-16-2014 08:18 AM)brouhaha Wrote:  If anyone has a spare HP-28S that they're willing to sacrifice in the name of science, please let me know.

Eric,

PM me with your address and I'll send you a 28S that's in many pieces -- but all there!

[Jlouis]

(12-16-2014 05:03 PM)Katie Wasserman Wrote:  

(12-16-2014 08:18 AM)brouhaha Wrote:  If anyone has a spare HP-28S that they're willing to sacrifice in the name of science, please let me know.

Eric,

PM me with your address and I'll send you a 28S that's in many pieces -- but all there!

I guess I fried my 28s, intalling 3x12v bateries in it... I know, I know, I'm stupid, but the damned batteries are almost exactly like the originals, And fitted in the compartment...

Anyway, if it is really gonne, I can still contribute to science, just give me a couple of days.