Rolf wrote:
> Here's the rub. I thought, and based on previous mails:
> 1. I could use the CVref to create the necessary (adjustable) voltage to
> set the contrast.

What is CVref?  I don't remember you defining this.

> 2. If I could not do it directly, that I could use an Op-Amp as a Buffer
> / Voltage Follower to accomplish the same.

Probably true.  What is the useful range of Vo according to the data sheet?

> Here's the problem...
> with 1. the resistor network in CVref is to Vdd,

I have no idea what that is supposed to mean.

> and thus, even though
> the voltage may be right, the current can not flow, and the LCD is
> "blank".

OK, you saw a blank LCD.  I'm not convinced your assumed reasons for that
are correct.  I particularly don't get the part about the voltage being
right but the current not flowing.  This leads me to suspect a
misunderstanding of the circuit, and perhaps some of electronics in general.

> with 2. putting the LCD Vo at the output of the OpAmp causes
> reverse currents through the OpAmp (I think), and it is not stable at
> the low voltages required for contrast. In my case, it would be too
> "dim", and then, as it approached "close", the op-amp would jump to the
> Vdd rail, and the display would "blank".

I'm not totally sure what the circuit was, but it sounds like a bad opamp
circuit or the wrong opamp for the job.  It sounds like you need an opamp
with output that is "rail to rail" or at least goes down to the low rail.
Unfortunately there is a lot gray area and sometimes outright misinformation
in spec sheets pertaining to this.  You have to know what kind of fish to
wave and exactly how to wave it.  In general, us a CMOS opamp.  All else
being equal, these are more likely to be able to drive their outputs close
to the low rail without jumps or other unexpected behavior.

> Placing voltage-dropping Diodes after the
> Op-amp to get the right voltage range did not solve the problem, and
> screen is still blank.

That makes sense as the voltage drop on a diode will make the opamp output
rail-to-rail problem worse.

> Bottom line for me.... Controlling *just* the voltage at Vo does not
> control contrast.

Not really.  I think the problem was that you weren't controlling the
voltage as you thought.

> Futher experiments using a voltage-divider after the buffer (and thus
> putting the buffer in to it's linear range) discovered that it did not
> matter what voltage I put at Vo, it was the resistance between Vo and
> ground that changed contrast.

I find this very hard to believe.  Did you measure Vo as you changed the
resistance?  I'm sure it changed.  The difference is that now you had a
circuit that was able to change the voltage, even if it was only loosely
controlled by controlling a resistance.

> Experimenting indicated that it did not matter what the Voltage at CVref
> was (from 0.0V through 3.8V), that it was only the position of the Pot
> that mattered

Again, what is CVref and what does it have to do with this?  I thought the
job was to control Vo.  The load on Vo may be non-linear with voltage, maybe
your driving impedence was wrong, I don't know.  Measure Vo as you change
the pot and you will see it change.

> So, I concluded that the Vo is not Voltage dependant, but resistance
> dependant. I then confirmed it by removing all components other than
> what became the rheostat, and behaviour was consistant with just a
> resistance (to ground) controlled device.

Again, measure the Vo voltage as you are doing this.  You will see it
change.

> yet, to me this is pretty conclusive....

Sorry, but it isn't since you haven't (can't) separate the Vo voltage from
the resistive load applied to Vo.

Another thing to consider.  LCDs are inhernetly switched devices.  The
current coming out of Vo may be very pulsed.  The voltage probably only
matters during one phase of this pulse.  All this means there should
probably be a capacitor on Vo to keep it at a "constant" voltage.  I'm just
guessing so check the data sheet.  Unless it says not to, try putting 100nF
on Vo with respect to ground.


******************************************************************
Embed Inc, Littleton Massachusetts, (978) 742-9014.  #1 PIC
consultant in 2004 program year.  http://www.embedinc.com/products
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist