LCD Driver PCF8566 (NXP)

[MrEmbedded]

Hi there guys,
I've recently come across a PCF8655 LCD driver IC that I am trying to control my LCD with. It is not a cascaded application and the PCB-setup is a direct copy of Figure 4 on pg. 7 in the datasheet provided (

http://www.nxp.com/documents/data_sheet/PCF8566.pdf

)

VDD: 3.3V
VLCD: 0V (also tried VLCD 3.3V).

In the datasheet I read that on reset are all backplanes set to VDD. Should I see them oscillate here? (Since it is multiplexed 1:4 with 1/3 bias). I've used a multimeter and there is no voltage difference between any backplanes or voltage difference between a backplane and a segment output. Does this indicate that there is some kind of error on the PCB?

I am trying to control the PCF8566 with an Atmel AVR32 uC (uses TWI, similar to I2C). I do get ACK back to my microcontroller and when I change device select bits I do not get ACK, this makes me convinced that the communication works fine. However, do I need to always transmit device select? Since my LCD is 1:4 multiplexed with 1/3 bias the normal mode (power on reset) settings should work fine. So I only need to transmit data to the RAM and I should see some voltage differences between backplanes and segments?

Is there any sample code I could take a look at? I does not neccessary have to be code written to an Atmel Device.

Many thanks!

Regards,
MrEmbedded

 

[zappedagain]

VLCD = 0V - is that floating or tied to ground (VSS)? You'll need a current path so floating doesn't work. Read through section 7.3.

"I do get ACK... I do not get ACK... convinced that the communication works fine" Huh??? - It looks like you'll need to read up on I2C buses too. You must always select a device to communicate with in in I2C.

I expect you are going to need an oscilloscope to continue debugging this; a multimeter won't give you the clues you need.

Welcome to electronics!

Z

 

[MrEmbedded]

Thanks for the welcome and the reply

VLCD is tied to VSS, from page 28 in the datasheet: VLCD = VDD-2,5 to VDD - 6,0 V. Since my VDD is 3.3 V, i thought that VLCD could be 0 (VDD-3,3V). Do you think that is a good choice?

I do transmit bitstreams with Device-ID all the time now

Best regards,
MrEmbedded

 

[zappedagain]

VLCD=0V seems reasonable from the datasheet; you would go to higher voltages to dim the display. Section 7.2 shows that you can let the PCF8566 define VLCD also.

Check Figure 22 and Figure 23; do you have enough bias current with these series resistances?

Are you sure your commands are getting through? After a quick look through the datasheet I can't tell if they want the MSB sent first or the LSB; that could be an issue.

Setting VLCD with the PCB8566 might be a simple test to make sure your commands are getting through.

Z

 

[MrEmbedded]

In Section 7.2 isn't like this that they mean that you can mix with the contrast:

http://forums.parallax.com/attachment.php?attachmentid=51768&d=1201638099

<-- picture of a schedule with adjustable contrast.

The currents can be a problem, but later on. I actually haven't even attached my LCD to the PCB yet. I am just looking at the output pins with an oscilloscope. The output pins are stuck. I think that is the weirdest thing. As I mentioned before: According to the datasheet Power On Reset should put the driver in Multiplexed 1:4 with 1/3 bias. I thought this meant that regardless of initialization adjustments, I should be able to see some kind of oscillations between the backplanes?

I am more towards that it is some type of Electrical problem that I am having, i.e. wrong VLCD. All my outputs for segments and backplanes are stuck at VDD (without any oscillations).

Thanks once again for taking time zappedagain !

Regards,
MrEmbedded

 

[zappedagain]

you can add a pot to set the contrast externally. You don't need to though as the part can set the contrast internally; you'll get much better adjustment range with the pot.

I don't have the datasheet in front of me anymore but I thought the part didn't do anything until you enabled it, even if it has default values.

Try adding a load (10K resistor) between a backplane and a segment output to see what happens with some current flow. I agree, the segment outputs at VDD still don't make much sense.

Z