Reading LCD control signals

[ataradov]

I'm thinking over one idea, which would require reading control signals of the bare glass LCD panel (COM and SIG lines) by the microcontroller (or some small FPGA).

The LCD is obviously correctly driven by the LCD driver, so it is a multi-level (looks like 4 levels) signal with positive and negative voltages (referenced to the MCU ground).

I have not spend much time thinking about it to know how much of that I actually need to capture to read the state of the segments, but I wonder if anyone has done anything like this before?

Quick google search does not produce good results. People are mostly reverse-engineering   digital interfaces to the controller.

The LCD has about 60 segments, it has 20 pins, and I have not tried to trace the actual signals yet.

Again, just bouncing some ideas here.

[tautech]

Have you seen Dave's LCD vids as they offer some good understanding:
https://www.eevblog.com/2018/02/05/eevblog-1055-how-to-design-a-custom-lcd/
https://www.eevblog.com/2018/04/15/eevblog-1074-custom-lcd-design-part-2/

[ataradov]

No, I understand how LCDs work. I was wondering if somebody had thought through reading those signals back with MCU and minimal external hardware and software overhead.

I thought that would be somewhat common problem if displays need to be retrofitted on old hardware, but quick search does not show anything useful. I'll be doing more research this weekend.

[amyk]

4 levels is a bit trickier but a lot of directly-driven segment LCDs will have voltages either +Vdd or 0V on the electrodes, with the difference between the two being the actual signal. Connecting them to MCU input would be enough in that case.

[ataradov]

Well, unfortunately in this case I have to deal with 4 levels. I will play with the MCU I have selected and see its actual thresholds, I may still be able to read them directly. All I need is to detect strong high and strong low levels.

I may also play with internal pull-ups or pull-downs of the MCU to see if I can bring the levels one way or the other.

And with COM signals, I only need one, since I can derive timings from there. That still leaves 15 segment signals to sample.