So many chips worth looking at from this 1999 Palm V teardown, to avoid making it too large I have split it into two entries in part 2 I am looking at the display backlight driver, the LCD driver and the DC/DC regulator.
Even for 1999 the 160 x 160 pixel LCD was not that advanced, but the Palms were (If I remember correctly) less than $500 each. The form factor (That it could be comfortably held with one hand) made the 3.2″ display a good compromise at the time.
This is the display board attached to the back of the display containing three chips. An Epson SCI7500 a dedicated LCD display controller. The display driver chips are flip mounted on to the display tabs. They are the long skinny rectangles along the top and side.
The screen is backlit (An upgrade from the original Pilots). It is interesting how this is done. These days most LCDs are backlit with a row of white LEDs emitting into a diffuser. Back in 1999 white LEDs were still a laboratory novelty (They had only been invented in the mid 90’s.) So the common solution used in backlit displays was an electroluminescence back light.
Glued to the back of the display is a plastic layer coated with white phosphors on one side and conductive ink on the back.
Two terminals connect to the phosphor film and to get it to luminesce it is driven with a ~160V 350Hz ac signal by the Sipex SP4422 electroluminescence lamp driver (which is a DC to AC converter.)
The die is a rather grubby looking affair (And it’s not from my depot!) It is a really old single metal Bipolar process. It’s a small die measuring 1.35mm x 1.25mm with just 8 bond pads. Die marks MS532, SIPEX and 1995.
The metal tracks are 4 μm minimum (So something that could of been produced as early as 1980 and pretty old technology by 1999.)
A really nicely written (In my opinion) datasheet is obtainable for the part that describes how it works. Here is the block schematic
Reviewing the datasheet together with the die photo I can piece together how this thing works. We can now see that the round wire wound coil on the board is 5mH/18Ω connected between the battery and pin 3 and is used to store the energy that is discharged into the lamp (The phosphors). A capacitor between Cap1/Cap2 sets the frequency of an oscillator circuit around 90kHz (The unmarked area around Cap1/Cap2 pins), Then a series of 8 flip flops divided down the oscillator frequency into 8 frequencies (45kHz, 22.5kHz, 11.2kHz, 5.6kHz, 2.8kHz, 1.4kHz, 703Hz, and 352Hz.) The last 350Hz is used to drive the coil, and the 3rd (11.2KHz) is used to drive the lamps – such that the ratio fcoil/flamp is always 32. The coil is discharged through two SCR’s (Semiconductor Controlled Recitifier) one for positive and one for the negative half of the sine wave. The lamps are pulsed with 16 pulses per half cycle, thus deriving a ~160V 350V signal from a 3V battery input. The datasheet does a really good job of describing the circuit.LCD Driver
The drivers are flip mounted as bare die directly onto the connector, which at the display end has a 0.22mm pitch connection to the LCD. There is one for horizontal and one for vertical edge. This is the vertical driver I have depotted.
Its a pretty large chunk of silicon measuring 15.6mm x 2.5mm. And proved pretty difficult to image, even using a 5x objective, I had to stitch 42 frames together (And given the repetitive nature of the image it was not easy, I was wishing I had an automated stage on my microscope?!).
The only die marks are a D175HDHB and no logo’s or any identifying marks to indicate who made this.
I was pretty surprised to see this thing is wire bonded to the connector. That is 203 bond wires (To save you counting) between die and connecter. I can’t comprehend how they flip mount the die onto the connector and bond it.
Now I expected this to be an array of 160 large transistors each powering the ability to change a liquid crystal pixel. Looking at the circuit there is a whole lot more going on. There is an array of 160 circuit blocks (One for each row) – here is a stitch of just one.
You can see a couple of large transistors here.
And another pair of reasonably sized transistors here.
But then there is a whole lot of circuit for each row.
It is fabricated on a old single metal (Aluminum) 3μm CMOS process. Searching on the part number unsurprisingly returns nothing. And I can find very little explaining how LCD driver works and why the circuit appears so complicated.
Measuring 1.50 x 1.16mm it is another single (Aluminum) metal Bipolar device. The metal tracks are 5μm on 10μm pitch (Again old technology for 1999.)
Searching for Linear Technology LTDT did not yield anything but after finding a 1610 die mark and searching for LT1610 a datasheet was found, confirming the LTDT is the LT1610 a 1.7MHz, Single Cell Micro-power DC/DC Converter.
The die marks and the LT logo and 1997. And two sets of initials (In different locations?) SP and MA.
This single transistor has a very odd layout. The collector ring appears split, with two base connections around a very irregular emitter metal contact.
There is still more silicon I could look at in the Palm V but I think I will finish here.