LCD Snooper: Best way to sample multi-level LCD signals?

NivagSwerdna · Joined: 29 Jan 2013 Posts: 8

I am trying to work out how to take a weighing scale and decode the LCD signals (a google shows this has been done before) and to take the signal and send it wirelessly.

However, I thought I would make the project more interesting by not only interpreting the signals and sending them but also driving the display myself to show an enhanced display.

The weighing scale has an LCD display with 16 connections. I have scoped these and found the backplanes (4) are multilevel 1/4duty 1/3bias.

Existing designs I have seen for this problem have not had to cope with 1/3bias.

I thought I would target 16F917 as it has the necessary LCD drive logic (to re-use the existing panel), an AtoD and enough pins to drive the wireless module and a demultiplexer to select across the 16 pins of the existing LCD signals.

The existing LCD has a 60Hz refresh rate.

I was considering sampling each LCD line and then differencing the segment signals with the backplanes to produce an indication of segment on or off.

16xSample Rate conversions sounds a bit excessive... Any other ideas?

Basically I need a reverse 1/4duty 1/3bias LCD driver...

TIA
Nivag

temtronic · Posted: Tue Jan 29, 2013 9:30 am

Obvious starting point is getting the LCD module datasheet from the LCD mfr....as it'll have ALL the details with respect to timing and pin usage.

You could 'cheat' and locate the testpoint that the scale mfr has on the pcb for calibration...feed that into the PIC ADC and then send the data wirelessly.

hth
jay

NivagSwerdna · Joined: 29 Jan 2013 Posts: 8

It's not a "LCD module"... it's the raw display, i.e. the bit of glass via a zebra strip connector. The driver is on a custom chip that is inaccessible but the tracks to the zebra strip are accessible.

I'm not aware of a test point... but that would be a nice alternative.

You do raise an interesting point though... maybe there is such a point since otherwise how would the scales be calibrated...

Gabriel · Joined: 03 Aug 2009 Posts: 1067 Location: Panama

test points are usually small round exposed copper pads... they look like vias but without the hole in the middle... these are used with pogo pins that press on the board once the board is assembled.

they are often not clustered together so you might have 2 test points for the LCD or chip_X in diferent locations of the board, in non logical places...

you may be able to track them using continuity measurements...or just following visually the track if its only a 2 layer board.

G.
_________________
CCS PCM 5.078 & CCS PCH 5.093

temtronic · Posted: Tue Jan 29, 2013 2:08 pm

Even raw 'displays' should have a part number. Classic 3 1/2digit DVMs using the Intersil 7109 dsadc chips come to mind....

Unless you can get a number for the LCD 'driver interface IC' you're going to have 'fun' trying to figure out 16 channels of data ---> display segments !!

As you've said it has been done before (BTDT 10-12 years ago) but you must have a known starting point.

hth
jay

asmboy · Joined: 20 Nov 2007 Posts: 2128 Location: albany ny

do you actually know how the scale measures ??

LVDT?

strain gauge ??

SPRING/GEAR/with :
potentiometer??
eddy current sensor??
capacitive sensor ?

even if you have to "linearize" the raw data , i think you would be
better off getting at the raw data from the scale mechanism than
the UBER difficult way you imagine......

NivagSwerdna · Joined: 29 Jan 2013 Posts: 8

The scale has four load sensors... presumably in a whetstone bridge arrangement. I could try using those directly with an instrumentation amplifier etc.

I quite liked my approach but I'm starting to have second thoughts.

The display is definitely custom made. It has 4 digits, lb/kg/st a fraction display (5 segs?) and a divisor digit and line.

I put the lines individually into my scope (actually found a backplane and used that for relative timings) and I found 4 backplanes and 12 segment driving lines. I thought I would work out the segment lines by tickling a combination of one backplane line and one segment line; there are only 4x12 combinations and I'm hoping to find some systematic arrangement pretty quickly. (Haven't done this yet as I want to build a nice oscillator with little DC component first).

There is only 1 black blob and a couple of surface mount Rs on the current board.

NivagSwerdna · Joined: 29 Jan 2013 Posts: 8

Front

Back

Zebra Strip on left to connect to LCD glass.

Single CR2032 battery and 4 load sensors

NivagSwerdna · Joined: 29 Jan 2013 Posts: 8

NivagSwerdna · Joined: 29 Jan 2013 Posts: 8

On the underside of the board you can see the wires from the four load sensors each with two resistive strips.

Presumably white is common and red is one sensor and blue is the opposing sensor. The load sensors are located at the four corners of the load plate.

NivagSwerdna · Joined: 29 Jan 2013 Posts: 8

I tried all the spare pins and no signals of interest.

I looked at the load cells... there are four, one in each corner. The load cells each come with three wires... white, blue and red. For some reason I am guessing they have white common and a resistor on two surfaces, one for tension and one compression... it's a bit of a guess. Measurements suggest each R is 1k.

The configuration seems to be a whetstone bridge of some flavour with 3v76 applied across it. The two outputs then feed, connected via a small C into a 18k and then 68k resistor combo.. looks like the input to an amp?

Maybe I should give up on the LCD signals and try to measure the outputs of the bridge... not sure if it produces 2 results you add or if you are interested in the difference between them?

I tried a bit of compression but the signal is too small to measure without some serious gain.

NivagSwerdna · Joined: 29 Jan 2013 Posts: 8

After due consideration I've decided, as recommended, not to decode the LCD levels but instead replace the controller in its entirety and use the output of the load cells via an Instrument Amplifier etc.

Thanks for all the contributions

temtronic · Posted: Thu Jan 31, 2013 4:10 pm

It sounds like the best (or easiest) solution. I'd have 2 or 3 known weights for calibration/testing based on what the scale displays now (reference).
When you add your Instrumentation amplifier circuit it might affect the original circuitry. Just be sure to document which wire went where !!! Those pictures are great, just add a few notes and label which set or r-w-b wires go to which sensor.
Nothing is more frustating that finding out you swapped a + for a - input.....sigh...
good luck
hth
jay