F5 lcd sobstitution

[Daxxin]

hi , since my f5 custom multiplexed lcd is broken and not worth to buy replace part i am try to sobstitute with gloss lcd the 4 areas

the serial data coming from the pic16f886 is decoded by the tristate 74hc595 , i was thinking to use the ser data from the pic
to graphic lcd but too much complicated so my question is what the logic IC to decode at 7 segments lcd?

[Skippy]

I don't think you need a 7 segment decoder, the HC' logic IC's drive the LCD directly, there is no additional logic inside the LCD itself. So you need to try and work out the internal connections of the LCD. For example, if you short an LCD pin to ground, does an entire section of the display stop working, or does it only affect odd random segments all over the screen? The LCD is basically a big X-Y grid, each segment is one place where a row and column 'cross over'.
One thing you might try (hard work) is set the machine running so it picks up lots of EMI, and the 88 display is continuously updating. Then connect a second LCD in parallel with the original, and trial and error see if you can determine the 88 lines. Any basic junk LCD could be used for this, but it's better to use something you have a known datasheet / pinout info for.
Don't forget, LCD drives are constantly being inverted (several hundred times a second) to stop damaging the crystals.
If you can look at the LCD pins with an oscilloscope, you may see repetition of patterns, these may help you work out what are row and what are column signals.

[Daxxin]

thanks for the reply , understand your explanation , first big problem ..the lcd display is almost all damaged so far can do test, if you look at the f5 display its splitted in
4 separated sector since onboard there is 4 hc595 i m supposing every 595 driving 1 sector and this logic IC's getting data directly from the pic , i guess the format of that data is 8-bit serial format
compatible to the logic ic's the output of 595 is paralled i think BCD ..maybe i need an IC that have decimal output to drive required numbers of segment

[Skippy]

No, you don't understand my explanation. I suggest you locate a datasheet for a SIMPLE LCD display, like an 88:88 clock display or similar, and read it. You will see that all 4 of the top segments are joined together, all 7 of the first digit's segments are joined together. Hard to explain in words. But roughly half of those edge pins drive 'rows', half drive columns, there is no obvious relationship between a pin and what it does on the screen. The screen is like a little circuit board, with tracks running over it.
I suggest you remove the smashed screen, in case it is shorting out the IC's that drive it.
You don't need any extra IC's.

[Skippy]

Here is an example display. The F5 display is just a 'bigger' version of this. Understand how the segments are driven by MULTIPLEX drive.
As your original display is junk, it just might be possible to remove the polarising plastic stuck on the front (heat from a hair-dryer may help), and the rear, and you may be able to see the almost transparent tracks deposited on the glass, and possibly see where some of them go. Just an idea, you have nothing much to lose, just be careful with the broken glass.
http://uk.farnell.com/varitronix/vi3...git/dp/1183159

[sinclairuser]

is the f5 display truely custom?, i find that hard to believe, if you dig around long enough you would be amazed what turns up.

[Skippy]

Getting custom LCD's made is not too expensive. It's not even the same LCD as the F4 and F2 models ( the latter two are very similar, you would think they could use the same display...).
Is it this detector?:
http://www.ebay.co.uk/itm/FISHER-F5-...orig_cvip=true

[Skippy]

Here is a good close-up of the display. Notice that there are six 7-segment digits, an 8 segment disc/notch indicator, two 4-segment bar-graphs. As a designer, I would choose to have all the 'a' segments of the 6 digits commoned up, all the 'b' segments etc. And similarly with the bargraphs. I reckon there is probably 8 connections for the 'anode', and about 10 - 12 connections for the 'cathode' (6 digits, 4 bargraphs + ?). Obviously I'm using LED terminology here, LCD's don't have anodes/cathodes, I hope you understand my meaning.

[Daxxin]

from lcd i count in total 30 pins , 27 goes to the 4 hc595 Q outputs , the 3 remaining go to resistor partitor then to a pic micro ..l ithink

[Skippy]

Another possible arrangement is for there to be 16 'anode' connections, driving each '88' as a pair, and each 5-digit wide segment on the bargraph. This would probably use up 6 - 7 'cathode' connections, one for each group of 88, one for the bargraph, one for the notch, one for the other smaller bargraphs. That would account for 23 connections. Again this is just an idea.
This would be an ideal use for a 'solderless breadboard' prototyping board, where the wires can be easily swapped around, solving the puzzle by skilled guesswork.
The way the 4 serial->parallel shift registers are connected should give some clues, too. Chances are they are arranged as two 16-bit shifters, if you can work out the pairing, you may get some ideas about the LCD pinout.
One other thing that may be useful to know - it is likely the F5 has an LCD test mode, which puts all the segments to black. I know the F2 has one, I think you hold in a certain button before you power on the machine. If you can find this feature on the F5, it could be worthwhile.

[sinclairuser]

if you can work out the the way its connected, maybe you can build an alternative display, using 3 sets of 88 led displays, and rows of low power leds to replace the indicators, a lot of work and as skippy said custom displays are available, perhaps fisher will sell you a display.
but a custom old school led numerical, say two large green ones in the centre, two smaller red ones for gbal, and two the same size but yellow for phase, with lines of leds to make the other indicators its a hell of a lot of work, but fitted behind that fisher faceplate it would look cool.

[Daxxin]

At this point i believe its very complicated , have detach the glass and some traces are overlapped , so far cant trace , maybe its better to use graphic display decoding the data?

[Skippy]

You don't seem to have realised - you can't decode the data from the PIC unless you know what that data does. So first, you HAVE to work out what the LCD connections and internals are. There are about 120 segments in the display, (though you can dispense with info like ground phase, notch settings, too)So if you don't want to reproduce the basics with an 88:88 display, you could read the serial data from the PIC, by working out the shift register connections. 120 bits of data is just 16 bytes, so your own micro will have no problem keeping a record of what should be displayed on screen. Then you can display this on a 128 x 64 dot matrix panel.
Do you have access to an oscilloscope? This would help you work out the 'anode' and 'cathode' connections. The polarity of these signals needs inverting 20 -100 times per second, so I expect the shift registers will be clocked with a rapid burst of 16-bits, then nothing for 20 millisecs, then a new burst of inverted data is shifted, repeating continuously.
It not a Rubiks cube, you can work it out.

[sinclairuser]

even to use something like the vdi project and use a different kind of display, you need to know the pins/connections if it was me i'd do something unique, but you cant even think about a replacement till you know how it works, it may be relatively simple or a complete nightmare no way to know yet.

[Skippy]

Because of the limitations of what tracking you can deposit on a glass panel (eg. no crossovers) it could be there are some of the connection pins joined together on the PCB, for example a pin on the left side might join one on the right side, because it's too hard to make the connection internally. It's likely the left end pins join to parts of the display at the left end, etc. Kind of obvious statement, and there could be surprises.