The Rogue HP-12c

[AnnoyedOne]

(09-13-2024 09:30 PM)dm319 Wrote:  Being a cheap bastard is better than having a really bad sense humour.

Unfortunately I cannot read the Bloomberg article (no account).

I knew about Mike Lynch's demise from his "yacht" sinking. I didn't know (or forgot) that HP was trying to get US$4 billion from him (or his now estate).

That's a boatload (pun intended) of calculators! Even Brazillian ones at the bottom of the ocean

A1 (aka "that cheap bastard")

PS: Did Lynch use a "Rogue HP-12C" while at Autonomy? Does that explain the "over-valuation" HP is mad about?

[dm319]

(09-14-2024 12:26 PM)AnnoyedOne Wrote:  PS: Did Lynch use a "Rogue HP-12C" while at Autonomy? Does that explain the "over-valuation" HP is mad about?

This thing goes deeper than we ever imagined.

[dm319]

Thought I would do a little video to demonstrate a few things:

1. The comma bug
2. IRR example error
3. Digit precision
4. Divide by zero issue
5. A few TVM differences
   

See here.

A couple of things I haven't done:
- I can't compare on video to an HP-12c. Maybe in retrospect I should have explained what you would have got on the HP-12c, but it is pretty similar behaviour to the HP-12c platinum but returns 10 digits rather than a maximum of 12. I also didn't show you the full 16 digit precision (just 15).

I couple of thoughts - some effort has clearly gone into making this device match an HP-12c - the way the stack lifts for example, and returning only an integer N.

[AnnoyedOne]

(09-15-2024 02:54 PM)dm319 Wrote:  Thought I would do a little video to demonstrate...

I watched it. Not sure that counts (except for Ewetoob hits).

Comma bug ==> no "user error". Real.

And we now know your real name. And have heard your Brit accent

I keep my online identity deliberately secret. Decades ago I used my real first name and location on a forum site. Another user tracked down a photo of me (and someone else) that I'd taken with my camera (self-timer). Uploaded to the internet without my knowledge/consent. Not to any social-media site either. A small rarely viewed one.

A1

PS: GCHQ/NSA/etc already know enough about me.

PPS: I created this online pseudonym about 25 years ago. I'm not really "annoyed" but had already created it.

[dm319]

(09-15-2024 03:23 PM)AnnoyedOne Wrote:  I watched it. Not sure that counts (except for Ewetoob hits).

My commiserations for sitting through it! I notice it's on 3 views, so that's you, me and one other (..looking around...).

(09-15-2024 03:23 PM)AnnoyedOne Wrote:  PPS: I created this online pseudonym about 25 years ago. I'm not really "annoyed" but had already created it.

I am glad you have mellowed over years. Yes, the gap between by real name and pseudonym has been breached many times already.

[AnnoyedOne]

(09-15-2024 03:41 PM)dm319 Wrote:  My commiserations for sitting through it!

None needed. Now I've seen your 12CP as well.

(09-15-2024 03:41 PM)dm319 Wrote:  I notice it's on 3 views...

My main browser is very much "secured". Youtube (i.e. Google) recently changed something so now I often have use another (not as secured) one to watch videos. Google still can't learn much about me though.

(09-15-2024 03:41 PM)dm319 Wrote:  Yes, the gap between by real name and pseudonym has been breached many times already.

For better or worse I know how such things can be used against you.

A1

[Surfin]

(09-13-2024 06:00 PM)dm319 Wrote:  

(09-13-2024 03:01 PM)AnnoyedOne Wrote:  You probably already know this but you're the only one in these forums (to my knowledge) that has a "Rogue 12C".

I feel very special. There has already been a fair
significant cost, so maybe with my next pay check... Of
course if anyone in the UK fancies lending it to me, and I
can send it back. My main worry is damaging the device.

Long time lurking here, but now had to register just to say I´m from Brazil and I also have one of this 'Rogue': C1GP320 serial number with a screw on battery door, same date firmware as yours.
Also the "comma bug", I have some other 12C´s and they all alternate, but not this one.

I also watched the video, but already on 509 views.
Seems like people are interested!

Initially was thinking about flashing the 2015 firmware to it, and place the internals on a '30th anniversary' that has a not so good keyboard, but now I think I will keep it as it is.

[AnnoyedOne]

(10-04-2024 09:31 PM)Surfin Wrote:  Initially was thinking about flashing the 2015 firmware to it...

If you do please save the current firmware first! Some of us would like to see it.

A1

[Divasson]

Beware because there is no guarantee that the Rogue 12c has the latest chipset - it may be the old chipset (2007-2014) and then the old cable needs to be used.

In Moravia's instructions of cable use, it makes an exception with a serial number range including your unit:

QUOTE:

The cable does NOT work on previous HP calculators (HP15c LE, HP12c
prior to 2015, and a new version with batch code “C1GP000”
, HP20b, HP30b), even if they have the same POGO connector. These other models
follow a serial RS232 protocol, have different functions assigned to each pin
and different voltage levels. It also cannot be converted to be used with these
calculators because it is missing the IC that handles the voltage levels and the
protocol. Operating this cable with HP15c LE, HP12c prior to 2015, HP20b
or HP30b may destroy the calculator!

UNQUOTE

[Surfin]

(10-05-2024 01:15 PM)AnnoyedOne Wrote:  If you do please save the current firmware first! Some of us would like to see it.

A1

For sure!!

(10-05-2024 01:42 PM)Divasson Wrote:  Beware because there is no guarantee that the Rogue 12c has the latest chipset - it may be the old chipset (2007-2014) and then the old cable needs to be used.

In Moravia's instructions of cable use, it makes an exception with a serial number range including your unit:

QUOTE:

The cable does NOT work on previous HP calculators (HP15c LE, HP12c
prior to 2015, and a new version with batch code “C1GP000”
, HP20b, HP30b), even if they have the same POGO connector. These other models
follow a serial RS232 protocol, have different functions assigned to each pin
and different voltage levels. It also cannot be converted to be used with these
calculators because it is missing the IC that handles the voltage levels and the
protocol. Operating this cable with HP15c LE, HP12c prior to 2015, HP20b
or HP30b may destroy the calculator!

UNQUOTE

That´s something interesting!!
So I´ll have to open it to check if the PCB has the place for the USB connector...

[Surfin]

So I had to open it!
Also opened a 'Rev 2' unit to compare!

The 'Rogue':
https://i.imgur.com/ZGO75FO.jpeg

And the 'Rev 2':
https://i.imgur.com/1deKiPu.jpeg

And indeed, the pinout is different, while the 'Rev 2' have mention to 'USB' on pins, the 'Rogue' one does not.
Later I will try to open a '30th anniversary' to check if it´s the same pinout as the 'Rogue' unit.

PS: neither of them had the springs that other 12C units I opened had...

[Surfin]

And here is the 30th anniversary:

https://i.imgur.com/Z4kkgBB.jpeg

Dirty pcb and case, no indication on pins...


Another thing: seems the battery pinout is different on the 'Rogue' only.

[dm319]

Oh how exciting to see another one in the wild!!

Yes I ordered the USB cable before I was made aware that it wouldn't work on this device. If anyone wants to lend me or Surfin one that would be very appreciated and I will send it back I promise. I have no idea what I'm looking at in the internals other than it looks different to either of those devices. But maybe someone can tell us what we're looking at here. I have a 12c platinum made in China and maybe it looks similar internally? Is it safe to open without damaging it?

I've been busy trying to work out whether the device is internally binary or binary-coded decimal, but I'm not getting that far, other than if it really is binary it is doing a great job of hiding it. Though I am still suspicious it is 16 digits precision which is what you would display if you used a binary double. If anyone has any good tests I am all ears. My next brainwave was to use a loop of some sort to magnify a small error. I'm leaning towards this being 16 digits of BCD, but I would like to know for sure.

If we were able to extract the ROM would we be able to decompile it? Albert Chan has pointed out privately to me that the result for puzzle 3 is simply the algebraic equation for FV performed as if you were keying it in on the device.

D

[Surfin]

(10-05-2024 10:52 PM)dm319 Wrote:  Oh how exciting to see another one in the wild!!

Yes I ordered the USB cable before I was made aware that it wouldn't work on this device.

I found this other topic explaining how to make an Adapter to make the USB cable work on a serial unit.

The problem is: I know the '30th anniversary' is a Serial unit, but since there are no markings on it to compare to the Rogue, on a first glance I can´t confirm if the 'Rogue' is also a Serial unit, the only certain thing is: it won´t work with the USB cable (made for the 'Rev 2' pinout).

My Platinum is the 1 battery variant, so no connector.

[AnnoyedOne]

(10-05-2024 05:29 PM)Surfin Wrote:  So I had to open it!
Also opened a 'Rev 2' unit to compare!

Thanks!

Interesting. The "rogue" unit #2 processor is PCB mounted and has no markings at all. It could be anything but is probably a Atmel ATSAM4LC2CA. And a 32kHz crystal as well.

The "Rev 2" HP-12C has a Atmel ATSAM4LC2CA (like the 2015+ HP-12C+, Moravia HP-12C+ and HP-15C CE) processor but no 32kHz crystal.

The "30th Anniversary Edition" HP-12C also uses a PCB mounted processor (also no markings but probably a obsolete Atmel AT91SAM7L128). No 32kHz crystal either.

So the "rogue" has different hardware and firmware. This is new information! We are making progress here!

BTW the HP-12C Platinum/Prestige internals are different again.

FYI if the "rogue" POGO USB connections are different per the PCB markings that is significant (non-HP and non-Moravia?). The labels C22/C23 suggest capacitor connections yet I don't see them. That said COM22 and COM23 test pads are present. The plot thickens!

Also the (red/black/white/orange) wires to the PCB seem roughly stripped/soldered to me. There's also flux around the 32kHz crystal (Y1). "Professional" manufacturing should be "cleaner" IMHO. All in all "fishy" to me.

A1

[AnnoyedOne]

(10-05-2024 11:40 PM)Surfin Wrote:  The problem is: I know the '30th anniversary' is a Serial unit...

Correct. Newer stuff is USB.

(10-05-2024 11:40 PM)Surfin Wrote:  ....the only certain thing is: it won´t work with the USB cable (made for the 'Rev 2' pinout).

It may be possible to alter a standard HP-!5C CE cable to make it work. Or not. Unless someone actually tries it that is unknown.

HP has/is reportedly reflashing these units with 2015 firmware so my guess is that it'll work.

A1

[AnnoyedOne]

(10-05-2024 10:52 PM)dm319 Wrote:  Oh how exciting to see another one in the wild!!

Agreed. Now we just need to "suck" the current firmware.

(10-05-2024 10:52 PM)dm319 Wrote:  Yes I ordered the USB cable before I was made aware that it wouldn't work on this device.

Can you say who told you that? I presume that the manufacturer and later HP knew but until Surfin posted photo's of the internals yesterday that wasn't public knowledge.

I'm sure that HP wants this to "go away" and that they consider the issue "resolved" (returned units are being reflashed with 2015 firmware).

A1

[dm319]

(10-06-2024 02:50 PM)AnnoyedOne Wrote:  Can you say who told you that?

It was actually official from Moravia, soon after I got hold of the device:

(10-05-2024 01:42 PM)Divasson Wrote:  The cable does NOT work on previous HP calculators (HP15c LE, HP12c
prior to 2015, and a new version with batch code “C1GP000”
, HP20b, HP30b)

(10-06-2024 02:50 PM)AnnoyedOne Wrote:  I'm sure that HP wants this to "go away" and that they consider the issue "resolved" (returned units are being reflashed with 2015 firmware).

Agreed!

We really need more units, and more brains to figure out what's going on.

Albert has been helping me, he found me two numbers which equate to the same 64 bin binary value, which is very exciting, but the Rogue HP-12c seems to do a lot of weird rounding, so I haven't been able to completely convince myself one way or another. I'll post my results here.

[AnnoyedOne]

(10-06-2024 06:35 PM)dm319 Wrote:  It was actually official from Moravia..

So, they're aware of these "rogue" HP-12C's. Perhaps from HP HQ. I wonder if they made them.

(10-06-2024 06:35 PM)dm319 Wrote:  

(10-06-2024 02:50 PM)AnnoyedOne Wrote:  I'm sure that HP wants this to "go away" and that they consider the issue "resolved" (returned units are being reflashed with 2015 firmware).

Agreed!

Too late! We have two of them And photo's! Also known to many members here.

A1

[dm319]

Ok summary of attempts to discriminate BCD versus binary.

Firstly, why do I think there is a chance it could be binary? The device has been returning up to 16 digits of precision on some calculations. 16 is an interesting number because it's the decimal precision most people equate 64bit binary floating point with, and if you were calculating in binary using this type, then rounding to 16 digits would be a fairly sensible thing to do. Secondly, there are vastly many more maths libraries working on 64bit binary, though there is 64bit decimal (Decimal64).

Albert has been helping me a lot with this, so these are mainly his ideas!

1
3
/
3
*
should give you 0.9999999999999 etc in decimal
whereas in binary the rounding should equate to 1

On the DM42 which doesn't do any silly rounding, we get -1e-34.
On my HP-12c platinum, again which is sensible, we get -1e-12
But my TI-83+ (which I am sure is decimal) gives zero.
And my Casio fx-85GT CW (which I am also sure is decimal) gives zero.
And my Casio fx-115W also returns zero.
because of sneaky rounding.

You can coax it out of the TI-83+ by subtracting, 0.9, then 0.09, then 0.009 then 0.0009 and finally it will give you 9.999999999e-5. The 85GT refuses to be coaxed. Also the 115 refuses to be coaxed - so maybe this is a mechanism Casio introduced a while ago.

The rogue HP-12c also returns 0, but can be coaxed out using repeated 0.9/0.09 subtractions. Unfortunately doesn't confirm if binary or BCD as Albert pointed out that 0.9 isn't represented precisely in binary.