Over the past year I have pulled apart several old phones. I thought it would be interesting to look at a selection of the LCD display drivers and how they have progressed.I am going to look at three LCD displays, the first from the Nortel Meridian office phone from 1994 with a 2-line 24 character LCD. Secondly from a Panasonic cordless phone from 2005 also with a 2-line backlit LCD. And the last is the display from a Blackberry 8830 from 2007. This one is more advanced with a 2.5″ (Diagonal) 320 x 240 pixel colour display.
Nortel Meridian LCD Display
Made circa 1994/1995 the display uses a conventional daughter board PCB connected to the LCD panel with a 94 pin flex connector. On the front of the board the LCD display driver is split into two chips. Conventionally bonded and placed on the board, however the die are not in normal plastic packages. Instead they have a thin rough plastic coating. They also do not have any package marks, which is typical of display drivers. I don’t know why they do not label/brand the parts, perhaps it is related to the supply chain of the display module. Maybe the LCD module makers control the IP and make it harder for new entrants in the space. In nearly all the other chip applications package branding is the norm.
A single sided board with nothing of interest on the back
Meridian Large Die
The only die mark 3C7985-8733 is an internal code number and there are no manufacturers logo. Thus the die will have to remain anonymous forever.
The part is made in a 3μm CMOS process, which is consistent with the Nortel and Motorola parts I previously looked at in this phone. Here is a section of CMOS logic. What is neat is that you can see the polysilicon layer clearly has two colors. The green and brown sections are differentiating the PMOS and NMOS gate regions. The different doping the polysilicon receives during the process sequences for NMOS and PMOS transistors results in the different appearance.Also in this image the metal layer definition is not the greatest, notice how the track in the middle of the image is notched where it runs parallel to the polysilicon edge. Several contacts also seem to have only partial metal coverage on one side.
There are two RAM areas at the top of the die, which I think is 6T SRAM, here is the RAM at high magnificationAnd the memory array at the bottom of the chip is I believe a ROM array programmed with metal links (Notice the gaps in metal in places)
Meridian Small Die
A lot simpler than the larger die, the bulk of the die consists of 40 repeated elements. These each contain a very small logic area with large driver transistors.
Panasonic Cordless Phone Display
Front and back of the display module
By 2005 the technology for small LCD displays had advanced. Mounting the display driver chips directly onto the LCD glass using a a new bonding technology “Chip on Glass”.
I peeled away the soft black coating to reveal the display driver chip hereThe COG technology is quite fascinating, here is a white paper from NXP describing the technology. The die has gold bumps bonded to Indium Tin Oxide (ITO) tracks on the glass by means of an Anisotropic Conductive Film (ACF)
From NXP: COG for LCD modules white paper
This Anisotropic Conducting Film is interesting stuff. A solution of conductive beads cured at low temperature. Anisotropic as it conducts only vertically ie. not horizontally between pads! The surface tensions of the glue on the conductive beads is apparently the key to this magic.
I released the die from the glass with some heat, and cleaned it up in hot sulphuric acid. The die is now classic rectangle of a display driver measuring 7 mm x 1.6 mm (11.2 mm2). You can clearly see the gold electroplated pads
KS0093 Die Photoclick on image for higher resolution version
The die marks are KS0093 and searching for that I was surprised to find a datasheet. The Samsung KS0093 “26 COM/80 SEG Driver and Controller for STN LCD”. There are no Samsung logos on the die, but it is clearly our chip matching the size and description perfectly.
With 180 pins it is quite a complex chip. This is the block diagram from the datasheetThis is made on a much more sophisticated process, its using Copper metallization and I think is a 3 or 4 metal 180nm process. Here you can see the electroplated Gold contacts on a 90μm pitchAnd here is a higher mag image of a logic area (80x objective focus stacked), you can barely make out the transistor layers
Blackberry 8830 LCD Display
From 2007 this is only a couple of years younger than the Panasonic display, but being one of the first Smartphones this display is really in a different league with 65000 colours, and 320 x 240 pixels (It really is quite remarkable how rapidly the resolution of smartphone screens progressed in the past 10 years).Peeling off the frame reveals the die that is also using COG bonding.
It’s a really long die measuring a whopping 23.5 mm x 1.5 mm (35.25 mm2)It was difficult to remove the driver chip from the glass, I tried lots of hot air with no success and it eventually took two hot acid baths to finally separate the die from the glass. With the huge aspect ratio getting a reasonable looking die photo image proved a challenge. The only way I could make the die photo practical to view was to split it into three parts
Made about two years after the Samsung KS0093. Fabricated on a more advanced process, likely 6 levels of Copper 130 nm CMOS (Or even 90 nm).
Finally we have an LCD driver with a manufacturer marking/logo. Searching for “Renesas R63401” also yielded a datasheet which indicates this is customized driver for Sharp who manufacture the LCD module.
Nothing worth exploring in the optical microscope for a part that is this advanced.