In the WEEE-k#1 follow up - testing the displays

[tron9000]

So I managed to get those displays tested and no luck on one of them and the other: the jury is still out to weather I have the right data on it, or if I'm barking up the wrong tree - or even barking the right questions.

[Muttley Snickers]

With the first LCD in the same orientation run your finger down the right hand side edge of the display, if you feel a convex arch or bump then this edge is the indicator for pins one and forty, you will also see a difference when looking straight at the screen, I think it may be upside down.

The decimal points indicate that the orientation is correct but these displays are available in quite a number of segment configurations, minus symols, clock colons and so on. Flip the display, whack 470 ohm resistors on 1 and 40 back to ground and start probing the other pins with another 470 ohm in series with 3 volts, each individual segment should fire up so write them down.

If it doesn't fire up it could mean it's a common anode display and not a common cathode as pointed out in the PM I sent you, just reverse it if thats the case.

On your next video I expect to see the word nIcE on the display.......


Best Regards

Muttley

[tron9000]

Just read my messages - keep forgetting to do that! Another inbox to look at

Yes - Doh! that could be why pin 1 & 40 are not connected and seems a sensible.

you want n1cE? - you got it!

[Ian.M]

Bare glass LCDs don't really like DC bias - they will survive a short test but its not good for them.  Also the link between supposedly common pins is almost certainly Indium tin oxide which may well not be conductive enough for your multimeter's continuity range to register.

The best approach is to AC couple adjustable amplitude antiphase squarewaves of about 30 to 90 Hz to the segment and common or digit contacts you want to probe.  Start by providing a leakage path to ground for all pins e.g a 1M resistor per pin,  connect two pins to the antiphase signals and increase the amplitude from zero till one or more segments are visible.   If you see two segments, you've found two segment pins but haven't found the common or digit pin.  If you see lots of segments, you've found two separate common/digit pins. In either case turn down the amplitude till the segments are faint then move one probe about until a single segment appears with reasonable contrast.  Once you've got one segment, you should adjust for best contrast then start mapping out the display. 

Here's a fairly crude circuit using a quad OPAMP to generate adjustable antiphase drive waveforms for testing only. RV1 is a 10K pot for the amplitude and R12,R13 are actually the individual pin bias resistors, not part of the probing signal generator.


If you've got a square wave output signal generator that can go up to about 5V peak to peak, you can skip all that, and simply use 1K resistors and 470nF caps in series with each probe, and return all the bias resistors to the mid-point of a potential divider using 1K resistors across the signal generator output.

Application circuits:
For a  non-multiplexed bare glass LCD you can use a job lot of XOR gates to drive the segments and a 50Hz squarewave signal to drive the XORs and the common.  If the segment is in phase with the common, its off, and if antiphase (i.e. other input of XOR is '1'), its on.  The signals can be generated in a MCU on an I/O port rather than by XORs.  Care should be taken to keep the duty cycle at exactly 50:50 to minimise any residual DC bias.

Multiplexed bare glass LCDs are a *LOT* trickier to drive with multi-level stepped waveforms required, and although you *can* get there with divider chains, CMOS switches and OPAMP buffers, it makes far more sense to use a proper LCD driver chip or one integrated into a MCU.

Edit: Corrected the bias resistor designators in the description + some clarification.

[tron9000]

I think I have read about this method a long, long time ago, you just reminded me.

So if I've got your attached circuit correct:

• U1A is basically creating a virtual earth point for the oscillator...
• ...U1B, which from my calculations is generating an approximately 30Hz square wave
• U1C is buffering (0^o phase) whilst U1D is inverting the signal(180^o phase) - hence anitphased signals
• R8A & R8B is the amplitude control, right? Which you've just used for SPICE simulation using the .step param

so R12 & R13 are the leakage path resistors right? cos I don't think I have 40 1M resistors! lol!

I could build that, I have most of the parts (LT1014). Quite a few passives but all 80% are same values. Cheers Ian.M

[Ian.M]

Yes you've got the right idea.

The 1M 'leak to ground' resistors are non-critical. They are only there to minimise ghosting, and provide a discharge path when you lift a probe   One approach is to stick the display in static dissipative foam, with a foil ground contact on the back between the two rows of pins.  If a pin to ground reads anything in the 100K to 10M range you should be good to go.   Lower value leak to ground resistors need bigger coupling caps so the applied squarewave doesn't droop excessively.

Most of the resistors round the OPAMPs are non-critical.  Just use all the same value between 10K and 100K.  The timing components C1,R6 should be chosen to keep he frequency between 30 and 90 Hz and yes, RV1 (R8A,R8B) is a simple 10K pot.  If using lower value feedback resistors round U1D, reduce the total pot resistance accordingly.

I've attached the LTSPICE simulation if you want to play with it.  It uses a .ic command as its somewhat slow to settle on an operating point otherwise.
You may wish to use .param RV=.5 instead of the .step command for a fixed ratio.
It would be a good idea to add 1 (1 ohm) inside the {} for R8A,R8B to avoid errors if RV goes all the way to 0 or 1.

[tron9000]

Thanks Ian.M, will have a play later on tonight.

On another note: anyone have any comments about the video - they are rough and the sounds levels are pants, but was the content ok? easy to understand? did I explain anything badly?

[tron9000]

double post apologies, thousand pardons!

Ian.M & Muttley - found an app note.

Ian.M, page 2 describes what your getting at quite well - http://www.atmel.com/Images/doc2569.pdf

[tron9000]

sorry, just thought I'd post up some pics of getting that circuit working.



Some minor amendments to component values, but nothing too far from originals, frequency comes out at 47Hz and on a 9V PP3 battery max output is 8V_pkpk

So Hopefully another success video in pipeline...that's if the LCD's not knackered! in which case it'll be a failure vid!

[Muttley Snickers]

I found a video showing a similar display in action and a picture that seems appropriate based on your experience with this display.

[tron9000]

  ha!

[Ian.M]

It looks like there is some asymmetry in the duty cycle.   The series caps on the output will prevent any net DC bias, and its going to be good enough for LCD pin identification, but I certainly wouldn't base a practical LCD driver on this quad OPAMP circuit and a job lot of analog switches!

8V pk-pk should be plenty of drive even if you've got two segments in series because of a misidentified common pin.  Once you've found a common, keep the voltage down.

http://www.goldenviewdisplay.com/driving_LCD_display.html is worth a read + click through the links at the bottom.

[tron9000]

Ta da!


Yeh there is ghosting of the DP's but I wound the wick down after realising to about 2.4V_pkpk

Pinout is EXACTLY per the data you sent me muttley!

Video on the way.

Thanks guys

[Muttley Snickers]

nIcE.......

Many thanks to Ian.M as well for a stack of valuable information that I was not aware of.......

[tron9000]

Indeed, that websites a good 'un Ian.M