Does anyone have 15C LE dissection pics?



#34

I'm curious if anyone has done this yet. HP seems to have taken a liking to screws over heat stakes in recent years. If that is indeed the case with the 15C LE, I will take a stab at improving a couple of minor issues involving the keys. The first step will be that of taking the wiggle out of f and g. Then I will move on to filling them, probably with some polyurethane reactive hot melt to minimize downtime.


#35

I've opened the back of one. It's heat-staked, there's foam tape behind the LCD to keep it from bouncing around in the case, and the Atmel CPU is in an LQFP package instead of being a raw die underneath epoxy like my 12C+ is. I have a picture but the quality's not that great, I'll see if I can clean it up a bit.


#36

It's the same as the 12C+. I've posted these pictures before:


#37

Nice pictures of the 12C, but if Lincoln R. is correct, the 15C LE does NOT look the same. That seems rather surprising since there isn't any need for the hardware to differ.. I haven't opened mine, though, so I can't confirm.


#38

I believe the 15c uses a packaged IC. That is the only difference IIRC.

TW

Edited: 21 Sept 2011, 1:28 a.m.


#39

Does the AE 12C have bad keys, too?

#40

Indeed, the PCB on the 15C LE doesn't have the potted chip on board, it's got the real packaged Atmel chip in all it's glory showing.

15C PCB

and the back of the case

If I'm reading this correctly the date code on this chip is from September 2009, ancient!

Why the switch away from the potted chip on board? Is this the same in the 12C 30th AE and newer 12C+ calculators too?

Edited: 21 Sept 2011, 3:34 a.m.


#41

Katie:

Nice work!

This is indeed good news in enabling any potential hardware modifications or studies.

TomC

#42

Are there instructions to open up the 12C+/15C LE posted somewhere?

Edited: 21 Sept 2011, 12:02 p.m.


#43

There are 6 screws to remove, one under each foot and two inside the battery compartment (one of these is smaller than the other 5). Once you do this it prys apart quite easily.


#44

The only mechanical constraints appear to be along the top and bottom edges of the case.

ALSO BE CAREFUL: Try to have the calc face down as you pry it apart as there are three springs in plastic collars in the keyboard half of the calc which can fall out. (These are used to electrically connect the keyboard to the metal plate on the battery compartment half of the calc - for ESD protection.)

TomC

#45

Spec details on Atmel AT91SAM7L128

A member of the Atmel SAM7L series of microcontrollers based on the 32-bit ARM7TDMI processor. It operates at a maximum speed of 36MHz and features 128KB of flash and 6KB of SRAM. The peripheral set includes a 40-segment LCD controller, two USARTs, UART, SPI, TWI (I2C), three 16-bit timers, RTC and four 10-bit ADC. The SAM7L128 operates in single-supply mode down to 1.8V and consumes 0.5mA/MHz typical in active mode when executing code out of its flash memory, and a mere 100nA in power-down mode. The SAM7L128 is supplied in a 128-pin LQFP and 144-pin LFBGA packages.

It is the SAME Chip on HP20b Calculator:
HP20b Calculator using ATMEL 91SAM7L128 Chip

Another good article about the 15c and ATMEL Processor
EDA360 15C and ATMEL AT91SAM7L128 (BTW, I think that's where the "inmate" discussion started in the forums.)

Edited: 21 Sept 2011, 1:26 p.m.

#46

Katie,

is that lipstick on the Atmel chip? LOL!


#47

Yes, but not mine. Probably whoever did the assembly in China was kissing it goodbye.

If it were an HP-16C+, that might be another story.

#48

I haven't checked, but if there is a bigger memory chip in the same package, it could be replaced fairly easily.

Dave.


#49

Quote:
if there is a bigger memory chip in the same package

There isn't one with a larger non-volatile RAM at least last time I checked.

#50

I remember seeing an AT91SAM7L256, I think that is double the memory. I will confirm and check on the packaging.


#51

There's a AT91SAM7S256 with lots of RAM, but without the LCD driver and requires higher voltage and more power.


#52

Flash isn't the big problem. RAM is.

The 34S could use more RAM. Lots more.

I know how to squeeze more into flash. It will require recoding many of the mathematics routines but it is possible.


Oh and the display needs to be a lot better. 400 segments isn't enough to do the hardware justice.


- Pauli


#53

There are a few spare i/o pins that are made available on the 20b and 30b PCBs. Have you experimented with adding a small serial non-volatile RAM chip? Such RAM wouldn't be in the calculator's memory space but you could write routines to page it in as needed or it might be fast enough to use for all memory access.


#54

We've not experimented but this possibility has been mentioned :-)

- Pauli

#55

Quote:
Indeed, the PCB on the 15C LE doesn't have the potted chip on board, it's got the real packaged Atmel chip in all it's glory showing.

15C PCB


Very nice indeed.

Quote:
Why the switch away from the potted chip on board? Is this the same in the 12C 30th AE and newer 12C+ calculators too?

The change to (and developemnt cost of) a new board layout with
an lqfp package is puzzling. Maybe a design/layout modification
was pending for whatever reason and the projected volumes didn't
justify the setup cost for COB a this stage.

Frankly I'm still not sure how a bare sam7l die ended up in the
12c+ version. There isn't a motivation due to high die pin count.
Perhaps it was an early attempt to discourage reverse engineering
over a concern of using an off the shelf commercial SoC. Hard
to see an economic motivation but Atmel doesn't appear to openly
sell the part in bare die form so that's inconclusive.


#56

Cost. In the quantities that HP makes the 12C, they are able to buy the bare die, attach and bond it, and encapsulate it, for less than what it costs to buy the packaged part.

The reason Atmel doesn't "openly sell" the part in bare die form is that they don't want to sell it that way to anyone that isn't buying a large volume and prepared to deal with the complexity of chip-on-board assembly. If you design a high-volume product, you can negotiate a bare die contract with Atmel too.


#57

Packaged chips go through extensive relaibility and retention tests including at high temperatures. Sometime the defects are not detectible until a high temperature test is performed. However bare die only goes through a simple probbing test. Testing becomes much more complex and difficult once the bare SoC chip is sureface mounted on a board, and some tests such as high temperature testing are not possible.

Therefore I would argue that 15C could be more reliable than 12C at least as far as the reliability of the SoC is concerned.

#58

Quote:
Cost. In the quantities that HP makes the 12C, they are able to buy the bare die, attach and bond it, and encapsulate it, for less than what it costs to buy the packaged part.

Possibly, but I believe the answer is only known to HP.
There is also the need for precision die handling, die adhesion
to pcb, wire welding, and encapsulation which add to the
manufacturing process cost traded off against the uneventful
alternative of reflowing a 128 pin lqfp on a board to be
printed and reflow processed anyway.

Quote:
The reason Atmel doesn't "openly sell" the part in bare die form..

Apparently they do but through specialized die distributors
and wafer processing third parties rather than directly.


#59

Quote:
There is also the need for precision die handling, die adhesion to pcb, wire welding, and encapsulation which add to the manufacturing process cost

I stated that, less precisely. However, they don't add very much to the manufacturing cost, because the ODM in China is already equipped to do all of that, and does it routinely for many products.

This is why if you open up just almost any sub-$50 electronic item, you'll find that it uses CoB assembly. SMT assembly in low-cost items is the exception.

This is what makes it particularly bizarre that the 15C LE has moved to SMT assembly.


#60

Quote:
This is why if you open up just almost any sub-$50 electronic item, you'll find that it uses CoB assembly.

An interesting (to me) example of that is the Casio fx-115/991ES. It has a 204-connection chip die under the black epoxy blob, most of which go to the 31x96 dot-matrix LCD with 18 annunciators (2994 elements). It sells for less than $13 at some US major retailers.


#61

Quote:
An interesting (to me) example of that is the Casio fx-115/991ES. It has a 204-connection chip die under the black epoxy blob, most of which go to the 31x96 dot-matrix LCD with 18 annunciators (2994 elements). It sells for less than $13 at some US major retailers.

Most contemporary graphic lcd controllers of such density are
as well only available in die form due to the sheer
number of pins present on the device. But obviously a valid
argument as the process must be economical at that price
point.

#62

Quote:
This is what makes it particularly bizarre that the 15C LE has moved to SMT assembly.

After a chat with an Atmel die reseller today it isn't quite
as intuitive as I'd expected. The pricing is all over the
map and in some cases singulated
die will have an OEM cost greater than
its packaged counterpart due to "market forces". Even if you're
interested in buying full wafers from a semi vendor you may
be stuck going through a reseller due to a vendor's
traceability and
liability concerns of the die ending up in unapproved hands,
or being packaged for resale via unapproved parties.
Other lesser logistical issues exist such as vendors
only shipping full wafers so if your production
pipe isn't continuous you may need to warehouse unbonded
die in a nitrogen bath to avoid atmospheric contamination
of the die's metalization.

No idea if any of the above applies to the 15c le's manufacturing
disposition. However this tends to confirm the suspicion a few
other variables exist in the equation concerning the economics of
bare vs. package die use.

#63

Quote:


The change to (and developemnt cost of) a new board layout with
an lqfp package is puzzling. Maybe a design/layout modification
was pending for whatever reason and the projected volumes didn't
justify the setup cost for COB a this stage.


speculation follows...

The lqfp might be the new design/layout.

Has anyone opened a recent 12c (30th AE or similar vintage) to see if it is still using the COB?

Perhaps the bare chip contract ended, and rather than extend it the board was redesigned for the lqfp.

Maybe the lqfp price was near enough the COB not to matter?

Maybe the 12c volume has declined so that the COB was no longer justified/cost-effective?

#64

This opens up opportunities to connect the missing JTAG connector and crystal for repurposing.


#65

Yes it does if you're good at doing some very precise soldering, it's a 0.5mm lead pitch.


#66

Quote:
Yes it does if you're good at doing some very precise soldering, it's a 0.5mm lead pitch.

0.5mm pitch gull wing packages aren't too formidable.
You can have at
it relatively comfortably using a 5x jeweler's loupe and fine point
soldering tip. Assuming the rework will be done in place with
the pcb mounted, one caution would be to keep the
typical flux drenching in check to avoid contamination via creepage
through vias into the key matrix. Then again IIRC
the vias may be small enough to have been closed by the solder
mask. If so this may prevent the flux from migrating.

Alternatively (judging from the autopsy above) the 15c le as
well uses hollow studs securing the pcb to the frame. I'd
assume as for the 12c+ the ID of the stud bore is just under
1.2mm. So after cutting the heat staked heads the pcb could
be reattached with 1.2~1.4mm self threading, flange head
screws -- a major simplification relative to
the trauma of pcb replacement for legacy voyagers.


Possibly Related Threads...
Thread Author Replies Views Last Post
  OT--Nat'l Semiconductor NS-7100 pics anywhere? Matt Agajanian 25 5,210 12-15-2013, 08:39 AM
Last Post: gene wright
  OT - A lucky find - Busicom Handy-LE LE-120A Cristian Arezzini 2 530 09-26-2013, 04:43 AM
Last Post: Cristian Arezzini
  Low power warning for HP-15C LE and batteries Nick_S 1 479 09-16-2013, 09:34 AM
Last Post: Hans Brueggemann
  HP 15C-LE replacement still available? Borja 16 1,665 08-22-2013, 11:16 AM
Last Post: Michael de Estrada
  JTAG on HP-12C and HP-15C LE Ingo 5 889 07-01-2013, 06:37 PM
Last Post: Paul Dale
  HP 15C LE extremums Richard Berler 29 2,915 05-20-2013, 03:26 PM
Last Post: Dieter
  New 15C LE bug? Paul Dale 3 663 02-05-2013, 09:27 PM
Last Post: Mike Morrow
  HP 15C LE, Program Display Format Control Uli 4 753 01-20-2013, 01:22 AM
Last Post: Ethan Conner
  So any pics / videos from HHC 2012 bluesun08 6 858 10-01-2012, 04:49 PM
Last Post: aurelio
  HP-15C LE, wow! New owner euphoria. Sasu Mattila 16 1,500 09-07-2012, 06:55 PM
Last Post: lars Bergström

Forum Jump: