Ipod Video LCD hacking

More LCD hacking here

Looking for a small, cheap LCD for a possibly large-scale project, I decided that the 2.5" QVGA TFT in the Ipod Video may be worth a look, as cheap clones of it are available from Hong Kong via Ebay..
Not finding any info on this unit on the net, I decided to investigate myself, & bought a faulty ipod on Ebay to dissect. The hard disk was dead, but I could get it into service test mode, which was all I needed to fiddle with the display to see how it was driven.

First problem was that the connector is a 0.3mm pitch FFC, so somewhat fiddly to connect to......  but I'm not easily defeated by fiddliness, and have a really good magnifier..
The FFC connector is staggered, with pins on alternate sides, so the wires soldered on to it are 0.6mm apart. I also needed to keep them very tidy & flat to stop them fouling the flex cable entry and locking lever.

I'd been reading a few controller datasheets, so had a good idea of what I was looking for. After identifying the power and ground pins, some probing with the scope pretty soon found what looked like the control lines - WR,RD,CS,RS. This left exactly 16 more signals - the data bus.

A major stroke of luck was that the test mode had some screen tests which displayed a graduated scale from black to each primary colour, so the RGB values counted upwards through the frame. This made identifying which pins mapped to R,G,B pretty trivial :  

I assumed that the bus was mapped from D0..15 along the pins, and this appears to be confirmed by looking at the commands sent at the start of a frame, which contained values corresponding to the 320x240 resolution ( see later)

Took a while to find the connector manufacturer - it's a 32 way 0.3mm pitch ZIF FFC - and unusually has an even number of pins and reverse-side latch. Couldn't find it from HRS,JAE, JST or Molex. Then I noticed the connector on the hard disk had a similar style of latch, so looked up the HD on the Toshiba site & found a datasheet listing the connector manufacturer. It's a DDK FF12 series, part number FF12-32A-R11A. I've also seen an Omron part used on another ipod PCB, and it looks like the part number should be XF2B-3245-31A, but the datasheet for this series doesn't list a 32 way option.

I enquired with a DDK distributor about availablility, and found that the minimum order is not just one reel, which I sort of expected, but three (7500 connectors), at GBP£0.71 each.
I think I'll have to see if it's feasible to use a more readily available 33 pin connector.....

Next step is to try to identify the controller. One way will be to look at the commands and try to match with possible chips for which I have data. But will post a few requests on the net to see if anyone already knows.... It should be possible to figure out how to drive it without the data, by logging the commands, but there may be some more interesting features that can be used once the controller is identified.

If you know for sure what controller this LCD uses, please let me know!

The register structure looks similar to the Toshiba JBT6K71, but registers are used that do not exist on that part, so best guess so far is it's another one in the Toshiba range.

possibles ( not checked yet) : JBT6K98,

Eliminated so far : JBT6K71, R61503, LDS285,S6D0154,ILI9320,S1D19122,JBT6K74A,JBT6K85,

Initialisation sequence 

Start of initialisation - Burst at end of trace is main initialisation sequence shown below. The six pulses inbetween are writes of 0000 with RS low.

Start of main initialisation

b

Sequence on powerdown (Power to LCD is removed on switchoff, so this command may not be to put display in powerdown mode!)

Sequence when writing new frame of image data

Access timing - command

Access timing - Data - rate is 5.5MHz. Full frame write takes 14.2mS

Driving it youself.

This display should be drivable from most microcontrollers ( *I've not yet verfied that it can do 3.3v I/O. If not, you'll need  alot of resistors, or a level shifter), however the 16 bit data bus will need rather alot of  I/O pins. If you're running from something like a NXP LPC2xxx ARM processor, this may not be a big deal, and you may want the speed of the 16 bit bus.

However there are a couple of simple ways to cut down the number of   I/O pins needed, at the cost of extra components, and speed. They also assume you're using the display in write-only mode, and so would not be suitable if, for example, you wanted to read out the display RAM contents.  A write-only interface could leave RD tied high ( assuming the initialisaiton sequence doesn't require the read - seems unlikely). You might also be able to tie CS and WR together.

1) 8 bit latch. Use an 8 bit I/O port to drive the bottom 8 data lines. Connect an 8 bit latch like a 74HC574 with its inputs to the 8 bit port, and the outputs to the top 8 bits  of the display's data bus. Connect the latch clock to another I/O pin. To write a 16 bit value, output the MSbyte to the port, wiggle the clock pin, then output the LSbyte and wiggle the LCD's CS/WR pins.
You should be able to share the latch clock with WR to avoid needing an extra IO pin - the HC574  will latch data on the rising edge, so the LCD module will ignore the latch pulse when CS is high, and the latch data will have been read by the LCD when WR goes high after the write with CS low. 

2) Shift register. Use two 8-bit shift registers, e.g. 74HC164, driven from a SPI port ( could use normal IO but a hardware SPI port will usually be faster) - clock 16 bits of data into the registers to set up the required value on the display databus, then twiddle the LCD CS/WR lines to do the write. As above, if you're short of pins, you could share a pin between the shift clock and WR, and use the data out to drive the RS line, so you'd only need 4 lines in total ( clock+WR, data+RS, CS, Reset)