Sharp Memory LCD

[andyturk]

I'm trying to get a Sharp Memory LCD working, but haven't had any luck yet. I know some of you folks have used these things, maybe you could take a look to see if I've missed something obvious. The display never shows anything--no flicker, no pixels, no signs of life at all.

Here's the board:


And the back, showing the connector for the display. The bodge wire is to tie EXTCOMIN to ground (which was floating previously). Pin one is on the left as you see it here:


Here's what gets sent out over SPI. The logic analyzer is connected to the 3V side of the bus transceiver:


The SPI clock is super-slow: only about 8kHz. Note that the lcd wants the bytes to arrive with LSB first, but this screen shot is showing the hex values as MSB first. The first byte going out is a command with the first two bits (as sent over SPI) high. That says it's a write command with VCOM set high. Next is one byte of row address (1-96) followed by 12 bytes of data (every other pixel) and a two-byte trailer.

The rest of the wiring is as follows:


The mcu is running at 1.8V and U1 is a SN74AVCH4T254 to translate the signals to 3V for the lcd module. Note that EXTCOMINT is floating in the schematic, but tied to ground currently (didn't seem to matter either way). Also, the signal labels between the bus transceiver and the mcu are wrong, but the connection are (I believe) correct.

Any suggestions?

[c4757p]

Well, I've got no clue, but... just saying, I love the purple board labeled "Azul".

[c4757p]

Wait, isn't your CS inverted? They're usually active low.

[andyturk]

Wait, isn't your CS inverted? They're usually active low.

Yeah, they usually are. But the Sharp seems to want it to be active high. At least, that's how I'm interpreting page 9 of the app note.

"Azul" is because of the bluetooth module on the same board.

[Bored@Work]

Didn't Mike play with one in his videos?

[andyturk]

Didn't Mike play with one in his videos?

Yes, he did. I think he said they were "easy" to interface too. 

[mikeselectricstuff]

Remember you need to periodically flip the COM line, either directly or via the bit in the SPI command (depending on the setting of EXTMODE) . I think it will still display if you don't, but may damage it if left in one state long-term due to electrolysis.

This can be as slow as 1Hz - I'm not sure why you would want to do it faster - it doesn't appear to make much difference and draws more power, but slower does make it start to flicker visibly.

ISTR you also need to sen exactly the right number of clocks and dummy bytes as specified in the datasheet. And remember that CS is active high.

The other thing to watch is that you make sure the SPI port has finished sending bytes before de-asserting CS

[andyturk]

Remember you need to periodically flip the COM line, either directly or via the bit in the SPI command (depending on the setting of EXTMODE) . I think it will still display if you don't, but may damage it if left in one state long-term due to electrolysis.

This can be as slow as 1Hz - I'm not sure why you would want to do it faster - it doesn't appear to make much difference and draws more power, but slower does make it start to flicker visibly.

ISTR you also need to sen exactly the right number of clocks and dummy bytes as specified in the datasheet. And remember that CS is active high.

The other thing to watch is that you make sure the SPI port has finished sending bytes before de-asserting CS

Jeez, I hope I didn't hurt it in my early attempts (which probably didn't have COM toggling correctly). It's not like I left it powered up in that state for very long though. Maybe a minute or two each time.

I'm attempting to do the COM toggling in software. The EXTMODE pin is tied to ground (along with EXTCOMIN). My test code has a loop that flips the VCOM bit (that's or'd into each command byte) once per screen sent (96 rows). So 96 rows will go out with VCOM=0, and then 96 with VCOM=1. When the SPI clock is running higher, it does about 10 screens per second.

CS is active high, and there's plenty of dead time between when the CLK is quiet and CS finally goes low. Could it be too much time?

It's got to be something simple, but I have no idea what.

[nctnico]

How is the clock polarity? Does the display sample the data on the rising or falling edge? You need to setup the SPI interface so it drives the data before the clock edge the display uses to sample the data.

[amyk]

From the app note it says the driver area should not be exposed to light. Try shielding it?

Jeez, I hope I didn't hurt it in my early attempts (which probably didn't have COM toggling correctly). It's not like I left it powered up in that state for very long though. Maybe a minute or two each time.

When that app note says you shouldn't exceed 1 second and you leave it for a minute...

[andyturk]

Pin 5 on the flex connector wasn't actually soldered to its pad. Pin 5 is the DISP line that actually enables the display. I bet the controller was running the whole time. Without a doubt, that little 10-pin connector is the most challenging thing I've had to hand-solder yet. 0.5mm flatpacks are easy-peasy in comparison.

Now that I can see the little bugger, it's pretty interesting. The "circuitry" on the edge of the display visibly pulsates like tiny blood vessels. Surprisingly, it seems to work even at 2 volts.

[c4757p]

Without a doubt, that little 10-pin connector is the most challenging thing I've had to hand-solder yet. 0.5mm flatpacks are easy-peasy in comparison.

Flux. Flux, flux, flux, flux, flux. When you think you have enough, add about ten times more. (And 0.5mm QFPs are easy-peasy)

The "circuitry" on the edge of the display visibly pulsates like tiny blood vessels.

 

[andyturk]

Flux. Flux, flux, flux, flux, flux. When you think you have enough, add about ten times more. (And 0.5mm QFPs are easy-peasy)

I used solder paste to begin with (which has flux mixed in). I probably didn't get enough under that pin because there was a gap between it and the board. I only saw it with a 40x loupe held at just the right angle. Looking straight down onto the board, it appeared normal.

They are organic molecules, you know. "It's Alive! Muahahaha!"

[mikeselectricstuff]

Flux. Flux, flux, flux, flux, flux. When you think you have enough, add about ten times more. (And 0.5mm QFPs are easy-peasy)

While I'd normally agree, you need to be careful with FFC connectors as you don't want the contacts clogged with flux. Use plenty of cleaner (e.g. Fluxclene) afterwards.
Remember that flux only works once, so clean it off & put more on after a failed soldering attempt.

[andyturk]