Multiple LCDs

[dcajacob]

Hi, I'm interested in using 1000s of LCDs to display in-store prices. I have several questions:

1) Can I use a single computer, hooked to a switching board to communicate with each LCD board in turn to update the information on the LCDs?

2) As long as power is being driven to the board, will the current displayed text remain on the screen. i.e. if I switch to a different LCD (per question 1), will the text remain on the last LCD, assuming I have a constant power source to all the screens?

Thanks.

Crystalfontz LCD Blog

Looking for additional LCD resources? Check out our LCD blog for the latest developments in LCD technology.

 

[CF Tech]

The big issues with automated price tags would be:

1) cost

2) communications

3) power

For cost reasons, you would almost certainly be looking at a small "segment" LCD, even dumb old character LCDs are too expensive when you multiply them by the number of products in a store.

For communications you could look at a custom multi-drop or pass-through wired method, or perhaps some kind of wireless setup (wireless would probably not meet the cost requirements). The wired methods would require some kind of a track that has connections in it.

The number of items in a typical supermarket is huge, so addressing becomes an issue. Speed can also be an issue during mass updates.

For power, a long life battery (possibly high cost) or a powered (possibly high complexity) track come to mind.

If you only want to build a "proof of concept" prototype, you could build a small module containing a CFAX12864AP-NFH and a low-cost flash based micro that could do the communications and store the display contents. Cool thing is it could animate small ads too

 

[dcajacob]

Thanks for the reply.

I realize the cost would be enormous, but assume that isn't a factor. I'd probably be using 2-line character lcds by the way, at least initially, so call it $4/unit for 1000+ units.

Now, the power issue. If a unit like this takes 5vDC to tun on and run, then all I need is a power rail at 5VDC to supply an entire shelf of price tags. Assuming the rail is designed to provide parallel volatge levels. Current goes way up, but we're only talking milliAmps per unit. So maybe 125 W power for 100 tags drawing 250 mA (very conservative I think) at 5V.

But really, my question remains: for a generic lcd, can I connect it to a computer, let's say and drive a message on to the display, then remove the data driver cable (without removing the power) and have the message stay on the screen?

 

[cosmicvoid]

The display content will remain on screen as long as power is maintained. But if you actually "disconnect" the data and control signals, you must take care that they are held stable (by pulldown resistors perhaps), so that transient glitches on the inputs don't cause false writes.

 

[X7JAY7X]

If cost isnt an issue, I think a better idea would be a Microcontroller to control each LCD. This adds two advantages, one if power was to fail, when powered back up it will redisplay the prices. Two, you could address each dispaly more easily, using serial communications. For example, you could send an update code, then the code of the corresponding lcd, then a price code(or something along those lines). This would also allow you to scroll data across, maybe an advertisement for the product? A microcontroller has enough memory to hold a price and to display a string of text.

 

[CF Tech]

I agree with cosmic here, without a micro, how are you going to address each display? You pretty much need to be able to update them from a central location.

 

[dcajacob]

I would be controlling them from a central location. A computer in the back room is connected to an electronic switching board and that board is then connected to 1000+ lcd price tags. So, to update a tag, the computer commands the switching board to connect it to the relevant tag and then serial communications update the tag and the switcher moves on to the next tag. All tags are continuously powered, whether connected to the computer through the switching borad or not. Is this feasible? I think using a micro-controller for each lcd is proibitively expensive...

 

[CF Tech]

So there will be one cable per LCD? Basically a star topology with 1000 points?

You could probably do all the micro tasks at < $1 per micro. Maybe less.

 

[cosmicvoid]

I think the $3~$4 micro attached to each display would be far cheaper than a switcher that would switch 5 (at a minimum: 4 data bus bits, and "E" strobe) signal lines noiselessly. Just imagine all the crosstalk from 1000+ 7-wire (5 logic + 2 power) cables fanning out from the back room, assuming you're silly enough to try to run logic level signals for tens or hundreds of feet. The cost of the wire alone would be outrageous, probably a lot more than the micro. And consider how fat the wire bundles would be, snaking along the shelves; an installation nightmare.

Much more reliable (and switcher-less) to have a 4-wire (rs-485 + power) or 3-wire (rs-232 + power) daisy chained to groups of displays. Do it in groups of 50 or 100 on each multidrop chain, both for signal loading considerations, and troubleshooting simplification.

At first, when you talked about "disconnecting", I had envisioned power only being daisy-chained to all the displays, and then to do the update you would walk around to the ones that needed changing, and "plug in" with a portable micro. After all, how many would need updating on any given day?

Edit: Ooops, I see you've now said "serial communication", so it's a 3 or 4 wire cable. But that implies a display with serial comm ability... more costly (?) than a generic character display + micro.

 

[CF Tech]

I think you could have a "C" shaped rail, power and ground tucked into the upper an lower groves (somewhat protected) with the communications line(s) in the back of the rail. Sections of the rail could connect like the old electric race tracks, and just be linked to the central station from one end. Imagine a Ethernet based box at the end of each double row of shelves, linked through standard network cabling to the master controller. Each of these end boxes would have several short cables--one to each track on that shelf group--that communicate the lower speed multi-drop protocol to the tracks.

 

[CF Tech]

. . . that implies a display with serial comm ability... more costly (?) than a generic character display + micro.

That is exactly what a serial display is . . . a standard display with a micro and communications interface.

 

[dcajacob]

CF Tech, I like where you're going with this...

Questions:

A serial LCD already has a microcontroller built in?

If so, can I reprogram the controller so that it responds only when a certain cookie is passed? For example, I have 100 lcds wired up in series on my rail, commanded at the end by one master line to the computer. The computer sends a cookie that dientifies the lcd i want and then sends the updates for the screen; all other lcds on the rail ignore this because the cookie doesn't aply to them. Does that make sense?

 

[X7JAY7X]

That is a nice idea CF with the rail, however, if you want to add one you would have to slide them all down, if I understood that correctly. A "snap in" idea would be better.

Why not make a serial LCD? It is probably cheaper than buying one. A character parallel lcd and a pic microcontroller is all you need.

 

[dcajacob]

You want to know the sad part? I'm a rocket scientist, no joke. My degree is in astronautical engineering. I just have very little practical experience with electronics. I'm learning though...

 

[cosmicvoid]

From what I can tell by looking at the 632 data sheet, there is no display addressing scheme in the CFA serial displays. So you'd have to use your own micro with an addressing method in the software, to allow multidropping. Although, at 1000+ units, CFA may be willing to make a custom version for you. But still, you could do it cheaper yourself, even if you hired me to do the code for the micro .

 

[dcajacob]

So, how much am I looking at for each microcontroller? Would a PIC controller be down near the cheapest? I could probably use one PIC for several LCDs depending on the avaialble data channels, right?

 

[X7JAY7X]

No one pic per lcd, a 16F628 is $2.75 from digikey. This would probably be the best option becuase it has built in USART.

You can get real cheap non-lit displays from ebay, last I saw $1.50 each (2x16 character displays)

You would have to designate each lcd a "serial number" to distinguish itself among the others. OR you could program in a routine to change this serial number by programming it with a computer. It wouldnt be hard at all.

I have been doing alot of work with PIC Microcontroller LCD programming lately, and a bit of Windows based serial programming.

 

[CF Tech]

I was thinking of a snap-in or push & turn arrangement.

You would definitely need some kind of small molded plastic enclosure to hold the display and mate with the rail.

For the lowest cost, you would use a "segment" display connected to an "LCD" microcontroller (the MSP-430 has some variants) then there would only be one chip per module.

For the highest functionality, you would use a small graphics display (COG maybe) with a separate micro.

No disrespect intended, but if you are going to build this as a product, you do not want to go with e-bay or surplus places. You will end up using all they have, then be in a situation of re-design or buying "real" versions of the display. You want to spec in something that is a "real" part (even if it is not from us) unless it is just for a demo or proof of concept.

I am serious that the micro should cost < $1 . . . they are out there

 

[X7JAY7X]

CF, I was just saying for testing purposes or what not. But that would probably be the cheapest place to find LCDs is ebay/surplus.

What micro cost < $1?

And where?