enter image description here

All of the LCD drivers I have seen use square waves to generate the RMS voltage across the segments. Since a segment acts as a capacitor, this seems to needlessly use coulombs to constantly charge and reverse the polarity by shunting to the supply rails.

Could it be more power efficient to generate the same RMS voltage using an LC oscillator where the glass is the C (or at least part of it)? There would still be R losses, but these could be significantly less than the losses of the digital drive.

Is there a reason why this is not done? Or is it done and I just haven't seen it?

NOTE: This would only apply to static segments and not multiplexed displays. I also know that the power requirements for the existing digital drives are already very small (nA's), but static LCDs are often used in extremely power and space constrained applications so any reduction of power could be worth it.

  • \$\begingroup\$ I've wondered this myself, though your circuit would not be suitable as it would create a DC bias on the LCD. LCDs behave primarily capacitively, so it would seem like this approach should improve efficiency. I suspect, though, that LCDs probably have a sufficient series resistance that the actual improvement in efficiency would be modest. \$\endgroup\$
    – supercat
    Feb 8, 2017 at 16:39
  • \$\begingroup\$ @supercat Ah, yes- good point on the DC bias. I think this could be overcome using a bi-polar drive? Alternate which direction you push on alternate pulses so the bias on the damping sine waves cancel out? You would still get power savings by not having to dump all that charge on every cycle, yes? \$\endgroup\$
    – bigjosh
    Feb 8, 2017 at 17:00
  • 3
    \$\begingroup\$ No. You are still transferring the same amount of charge to the glass, and using saturated switches (square waves) is the most efficient way to do that. This is especially important when you're talking about hundreds or thousands of individual row and column drivers. Besides, no matter how you implement them, inductors take up a lot of room and add cost. It doesn't matter whether you're talking about a simple wristwatch or a 4K monitor, space and budget are overriding concerns. \$\endgroup\$
    – Dave Tweed
    Feb 8, 2017 at 17:09
  • \$\begingroup\$ perhaps the best approach reduces the square-wave toggle rate on LCD until the subjective eye response sees flicker: in the 1 - 10 toggles-per-second ballpark? \$\endgroup\$
    – glen_geek
    Feb 8, 2017 at 17:32
  • \$\begingroup\$ @DaveTweed It seams like if I drive the LC with pulses of alternating polarity and always wait for the previous ring to die off before sending the next pulse, then though symmetry the DC bias is zero. Am I missing something? \$\endgroup\$
    – bigjosh
    Feb 8, 2017 at 21:57

1 Answer 1


A series capacitor would get rid of the DC bias issue.

In theory (ignoring inductor resistance and hysteresis losses in the LC fluid etc.) resonating the capacitance with an inductor would work (only taking energy when the state of the segment changes) but I doubt it makes sense practically. Even if a segment was as high as 100pF it would take thousands of H to resonate that at a reasonable frequency.

Simply switching the segments means a loss proportional to \$fCV^2\$ so reducing f, decreasing C (smaller segments) and reducing the voltage will minimize power consumption.

  • \$\begingroup\$ All great points. How about adding additional capacitance in parallel with the segment to decrease the resonant frequency to match the segment's desired refresh rate? At very least it could mean longer intervals between having to generate the drive signal, and segments would have a nice fade-to-black effect when they turned off (were not re-pulsed). Possible? Practical? Problems? Thanks! \$\endgroup\$
    – bigjosh
    May 24, 2017 at 21:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.