Post by thread:LCD Multiplexing

I am currently working on a redesign of a battery operated product that uses a 28 segment LCD display. Since the production volume is going to be substantial (>10K pieces) I am rationalizing the hardware to its bare bones. I want to eliminate the current 32 segment LCD driver chip (a Motorola part) and do the multiplexing with the PIC itself. One way to reduce complexity is to multiplex the 28 segments as 7 segments X 4 colums. Not being familiar with LCD Multiplexing, I checked out the Intersil databook (yes, taken over by Harris awhile back) to see how their LCD Multiplexors worked. They use something called tri-plexing, which calls for each row and column line to be able to source 4 discrete voltages - not an easy task with a PIC. What I'd like to ask is this: Does an individual LCD segment act like a capacitor in that a stored charge is proportional to contrast, or is it simply that applying a potential difference across the segment causes crystal twist and the effect's duration and intensity is entirely physical? OK, the reason for this complicated question is this: I can setup a 28 segment LCD display as a 7X4 array. All 7+4=11 lines would be tristatable so that only one of 4 column lines would be active at a time, and an AC signal will be applyed across the 'ON' segments by alternatively sourcing and then sinking the appropriate segments. A segment line (1 of 7) would be tristate when the driven segment is 'OFF'. This is to prevent other segments receiving 1/2V while they are connected as 2 segments, in series, with the center connected to a tristated column driver. By only sourcing segments on one pass, and then sinking on the next, we prevent 'bleeding'. However, if an LCD segment needs a sustained charge to stay twisted, then I would be concerned that any leakage off of segement charge would kill the display's contrast. As well, since tristated lines are used, how sensitive would the display be to leakage current overall? I know this is somewhat obscure. Perhaps someone has already tried this or done the research. Thanks. Regards, Dana Frank Raymond spam_OUTdfrTakeThisOuTicom.ca

Dana, Maybe a better solution would be to use a PIC16C923 or PIC16C924. These devices have an on-chip LCD module that is capable of driving: 1 COM x 32 SEGs 2 COMs x 31 SEGs 3 COMs x 30 SEGs 4 COMs x 29 SEGs These are the features of the PIC16C923/924: - 4K x 14 Program Memory - 176 x 8 Data Memory - DC to 8MHz operating speed - Timer0 8-bit timer with 8-bit prescaler - Timer1 16-bit timer/counter - Timer2 8-bit timer/counter with 8-bit period, prescaler, postscaler - One PWM - Synchronous Serial Port with SPI and I2C - 8-bit, 5 channel A/D converter (924 only) - LCD Module - Multiple timing sources - Can drive LCD while in SLEEP - Static, 1/2, 1/3, and 1/4 multiplex modes - Static and 1/3 bias capability - 68-pin PLCC or 64-pin TQFP packages Microchip should be sampling these devics within the next couple of months. Hope this helps. Regards, Rodger Richey Sr. Applications Engineer Microchip Technology Inc. ______________________________ Reply Separator _________________________________ Subject: LCD Multiplexing Author: Dana Frank Raymond <.....dfrKILLspam@spam@ICOM.CA> at Internet_Exchange Date: 6/9/96 10:03 PM I am currently working on a redesign of a battery operated product that uses a 28 segment LCD display. Since the production volume is going to be substantial (>10K pieces) I am rationalizing the hardware to its bare bones. I want to eliminate the current 32 segment LCD driver chip (a Motorola part) and do the multiplexing with the PIC itself. One way to reduce complexity is to multiplex the 28 segments as 7 segments X 4 colums. Not being familiar with LCD Multiplexing, I checked out the Intersil databook (yes, taken over by Harris awhile back) to see how their LCD Multiplexors worked. They use something called tri-plexing, which calls for each row and column line to be able to source 4 discrete voltages - not an easy task with a PIC. What I'd like to ask is this: Does an individual LCD segment act like a capacitor in that a stored charge is proportional to contrast, or is it simply that applying a potential difference across the segment causes crystal twist and the effect's duration and intensity is entirely physical? OK, the reason for this complicated question is this: I can setup a 28 segment LCD display as a 7X4 array. All 7+4=11 lines would be tristatable so that only one of 4 column lines would be active at a time, and an AC signal will be applyed across the 'ON' segments by alternatively sourcing and then sinking the appropriate segments. A segment line (1 of 7) would be tristate when the driven segment is 'OFF'. This is to prevent other segments receiving 1/2V while they are connected as 2 segments, in series, with the center connected to a tristated column driver. By only sourcing segments on one pass, and then sinking on the next, we prevent 'bleeding'. However, if an LCD segment needs a sustained charge to stay twisted, then I would be concerned that any leakage off of segement charge would kill the display's contrast. As well, since tristated lines are used, how sensitive would the display be to leakage current overall? I know this is somewhat obscure. Perhaps someone has already tried this or done the research. Thanks. Regards, Dana Frank Raymond dfrKILLspamicom.ca

On 6/9/96 10:03 PM, Dana Frank Raymond wrote: -snip- >What I'd like to ask is this: Does an individual LCD segment act like a >capacitor in that a stored charge is proportional to contrast, or is it >simply that applying a potential difference across the segment causes >crystal twist and the effect's duration and intensity is entirely physical? It's the potential difference that causes the liquid crystals to "line up" in a way so that an incident plane-polarized ray of light will be twisted about 90 degrees. The LCD segment is often modelled as a capacitor so you are right in both statements. First, light travels through a plane-polarizer and then gets twisted by the "lined up" crystals and finally gets polarized again. The rays are then reflected at the backplate and transmitted back to the user. Rays propagating through the "lined up" crystals will "disappear" and be seen as black and all the other rays comes back polarized but still visible. The "inactive" part of the liquid is twisting the incident light at random so there is a small loss here. This can be seen as darker areas on a common LCD. Note that this is a *very* simplified description of the process and please don't ask me how it actually works, I haven't read all about advanced quantum theory yet (and probably never will). I can recommend the book "Optics" by Eugene Hecht, Addison-Wesley Publ. The book explains almost everything about optic phenomena, lenses, waves, polarization, diffraction and much more. In practise, it's not an easy task to drive a multiplexed LCD in the correct way but the goals are to: * eliminate the DC-component over each and every segment * "light up" the correct segments without disturbing the others the RMS voltage applied determines when the segment is on or off. The first goal is hard to implement in a processor with lots of other things to do, a dedicated LCD-driver is nescessary if the LCD is supposed to live a long and healthy life. The second goal is a matter of clever engineering and experience, the manufacturers often have detailed data sheets on how to drive their LCD:s. A year ago, I used a 7x6 LCD for a simple RPM-counter. The 16C55 did drive the LCD directly and it worked but not very well. There is an app-note for this: AN563. -- Conny PS. I'm cleaning the floor (boring job but nescessary) in my electronic workshop and that's why this letter became long as a s-p-a-m but hopefully some of you could benefit from it. DS.

> A year ago, I used a 7x6 LCD for a simple RPM-counter. The 16C55 > did drive the LCD directly and it worked but not very well. There > is an app-note for this: AN563. Thanks for the recommendation and your thoughts Connie. I believe I have a way of doing it using 11 tristate lines (for a 7X4 LCD) using 4 passes per refresh cycle. The big issue is the effect of leakage current while PIC lines are triatate (could be tricky at high tempertures). I can image some of the difficulties with using a non-interrupt capable PIC for this (16C55). Namely, keeping the timing symetrical to eliminate DC bias. I tend to use the TMR0 interrupt as a foreground task to 'virtualize' the hardware for the application code. This technique has been very sucessful for me. I'm currently completing a design that uses a 220R resistor pack, PIC16C622, and a 74HC374 latch to multiplex 4 8 segment displays and 8 individual leds as an 8X5 LED array. Works great, foreground loading is on the order of 10-20% or so, with a refresh rate of 100Hz. The foreground runs also timers, and scans, debounces, and provides other special functions (like auto-repeat) for 8 switches, etc. I'll check into AN563 as well as the new LCD driver PICs. Thanks again. Regards, Dana Frank Raymond EraseMEdfrspam_OUTTakeThisOuTicom.ca