A tradition in my house at the xmas dinner is the christmas cracker with the paper hats, the corny jokes and of course the tacky toy. This maybe a British thing that is not so common worldwide, for those not familiar they look like this.
ou each pull on one end to create a bang (At best a mild snap) and the winner – whoever gets the bigger part – gets the paper hat, a small toy or trinket and a corny joke to tell.
This year I was somewhat surprised when this fell out my cracker
It’s a fridge magnet with a spring clamp and a calculator. The display is not the greatest but it actually worked. Now these were mid-range/high end crackers, but to get something that I almost would use was a surprise. The thing must cost less than $0.50 as the crackers were less than $2 each! I thought it would make a good opening post and allow me to sharpen my photography, teardown and blogging skills.
Dismantling was pretty straight-forward it used parts screwed together. Four pieces of plastic, held with 10 small screws and a fairly strong spring that makes the unit into a clamp. Onto the electronics – now I was expecting a simple board but you don’t normally find them this simple
Just the 20 keypad buttons on one side and a battery, one surface mount passive device and a single ic packaged in glob-top resin. I probed the surface mount component and it’s a 115 nF capacitor. The LED display is as you would expect a very basic display that was glued to the board with a 27 pin flexible connector which I believe is called a tab mount.
I wasn’t really sure why I would find. What I found is an interesting mix of old and very old technology. It is a tiny piece of silicon just 1.5 mm x 1.5 mm. There are no identifiers other than this generic number. Searching on 65280 of course yielded no information.
The die has 45 bond pads, 42 of which were bonded out which is a lot of pads for such a small die. It has just one layer of Aluminum metal (Which is the very old technology) with two polysilicon layers. The metal tracks are 3 μm minimum size and the minimum polysilicon I measured around 1 μm, so I think it is a 1.25 μm CMOS process which is old technology (Early-mid 1990’s) and large relative to modern technology but smaller than I would have expected for this type of design.
Clearly this is not a MCU, I think it is based on a shift register with a small ROM and a state machine. Deciphering digital logic is not a skill I have but I think this is what we can see in the die image.
The data flow is from left to right, I think at the left edge you have a series of 24 repeating structures that I think are flip-flops forming part of the shift register.
And then there are three arrays to the right, the first is a type of ROM using two polysilicon layers. And to the right of the ROM block there are two structures with identical layout, but with one half the size of the other, these are possibly small SRAM arrays.
Here is a close up of the bottom of the ROM array (~1000x on a 13 inch (1440 x 900) display, your magnification will depend on what size monitor/resolution you are viewing with) . You see from the bottom metal tracks connecting to polysilicon vertical lines, that are crossing (What I think are) horizontal polysilicon lines that are connected 6 at a time to a metal track on the right. There are no contacts visible in the array, I think the way this is programmed is there are contacts (Non visible) or no contact where the polysilicon lines cross.
To the right of the arrays is the state machine logic, and the LED drivers (This being the only device on the board the display drivers are on the die.) I would have expected to see a series of large multi-finger transistors that would drive the LED display, I can only see two that must be doing the display work
Maybe there is a large surplus of these chips that were used for this. I am often surprised by what can be bought for a $1, but it is hard to believe that there is a wafer fab producing 1.25 μm CMOS calculator chips for fridge magnets but who knows? Not to mention the display, board, keypad, plastics, spring and screw hardware costs.