Non-uniform Contrast when Direct Driving LCD with PIC24FJ128GC006

[NivagSwerdna]

I am trying to drive a piece of LCD glass from a weighing scale (with an aim of completely replacing the internal electronics).  The exact physical characteristics of the glass are unknown but measurement of the original PCB showed it was four way multiplexed with 1/3 bias.

Anyway.... I'm using a PIC24FJ128GC006 and it works pretty well except when a lot of segments are ON there seems to be a bit of 'fade out'.  I am driving the display with a 60Hz frame-rate (Type A) which is what I saw in the original.

According to the family PIC24F LCD documentation (http://ww1.microchip.com/downloads/en/DeviceDoc/39740a.pdf) there is a recommended value of 0.47uF for VLCAP and VLBIAS1,2,3.  (But on page 20 it seems to refer to 0.047uF)

I have max contrast on using the internal ladders and bias levels on the commons look spot on.  I'm using Power Level B 100%.

Currently I am powering from an ICD3.

Should I be increasing/decreasing VLCAP et al?  Is the glass too big to use the internal regulator?

Thanks in advance

(It looks more faded to the human eye in the middle bottom than is obvious in the camera shot)

[NivagSwerdna]

According to the family PIC24F LCD documentation (http://ww1.microchip.com/downloads/en/DeviceDoc/39740a.pdf) there is a recommended value of 0.47uF for VLCAP and VLBIAS1,2,3.  (But on page 20 it seems to refer to 0.047uF)

Could this be a bug in the documentation?  I came across http://ww1.microchip.com/downloads/en/AppNotes/01428a.pdf and on p14 it has a diagram with 0.047uF.

Maybe the 0.47uF was bogus?

[StillTrying]

"I am trying to drive a piece of LCD glass from a weighing scale"

Is the LCD well used/old. The faintest C segment in the middle would have been the most used - on for over 90% of the time that the display was on, and the first '18' segments being the least used, perhaps the C segments characteristics have changed with long use, - which doesn't help much.

[NivagSwerdna]

Is the LCD well used/old.

It's circa 2013 but has had almost zero use since I took it apart straight away! 
I think those segments are OK when it is not driving all of the segments.  I need to do some more experimentation.
I find the LCD documentation very inconsistent with the use of .047 and .47uF examples... I might try .047uF on LCDBIAS1,2,3 and see if it makes a difference.
The application note uses .047, the family manual uses .470 but a demo board schematic I found (http://ww1.microchip.com/downloads/en/DeviceDoc/52026a.pdf) shows .470uF
Maybe I just have a dodgy Zebra. 

[NivagSwerdna]

I sorted it.  It was a mechanical/contamination thing on my zebra strip.  Once I realized the dim segments shared a segment line it had to be supply to that segment.  Took out my rough and ready 3d printed holder and shaved one of the pillars to level it slightly, removed and cleaned the zebra strip, reassembled and looks much more uniform now.

Zebra strips are clever but a certain degree of magic is required to make the glass and the pcb contacts line up. 

Thanks for the interest.