80x16 RG Led Matrix Design

[Pepper]

I am trying to design and make a 80x16 Red/Green Led Matrix "Scrolling Sign". I have 20 8x8 RG led modules to use. I am somewhat new to Electronic Circuit Design so I'm going to need some help on how to do pretty much everything. I would also like to include an ethernet interface to be able to update the sign, but I don't want things to get too complicated.

I've searched the internet for hours for some sort of tutorial but couldn't find anything using RG leds or anything on the scale of 80x16. Any help would be appreciated, I'm even willing to give out monetary compensation for significant help.

 

[3v0]

I'm even willing to give out monetary compensation for significant help.

I am somewhat new to Electronic Circuit Design so I'm going to need some help on how to do pretty much everything. I would also like to include an ethernet interface to be able to update the sign, but I don't want things to get too complicated.

It will be complicated from a newbie's point of view.
Have you read the "getting started" stick thread under this section?

This is going to take a microprocessor and you are going to need to learn to program it or hire someone to do it for you.

Can you provide a link to info on the LED modules you want to use.

3v0

 

[Pepper]

Im not worried about the programming side of things, I've been programming in various languages for years. I just don't know what type of microprocessor to use and the circuit design etc.

**broken link removed**

 

[cadstarsucks]

Im not worried about the programming side of things, I've been programming in various languages for years. I just don't know what type of microprocessor to use and the circuit design etc.

**broken link removed**

You can expect to require a over 320 bytes of RAM just to save the image and a byte transfer rate of double that to get to half intensity (if you are taking half your time to transfer you can only be lighting the display half the time) times 100 (refreshing the entire display 100 times a second).

You need ten registers, byte serial, and a 1-16 decoder for the columns. You send the data out to the registers, the turn on a column and wait. Turn off the column and repeat.

Realistically if you can not expect a technique to put out 128KB/S (320*100*4) ditch it. You will probably want two display buffers so you can do the manipulation on one while an interrupt routine is displaying the other.

D.

 

[picasm]

You could build a simplified version of the rgb matrix controller made by SparkFun electronics:
https://www.sparkfun.com/commerce/present.php?p=Ethernet_LED_Matrix
It uses an AVR micro but you could use a PIC if you prefer.
If you use a PIC18F range, then you have more options for adding usb support and more memory etc.
Microchip also supply free C code and application notes for building your own ethernet interface if you are keen not to buy a ready-made one.

 

[3v0]

Im not worried about the programming side of things, I've been programming in various languages for years. I just don't know what type of microprocessor to use and the circuit design etc.

**broken link removed**

You experience was not evident from your first post. Sorry

This is mostly rambling but you may find it helpful as a starting place for your thinking. You did not provide any guidance on budget. I am thinking the following could be built for $200 to $300 including connectors. I suggest using PICs here because they are what I know. If you want to use another processor others can help you with that. This design idea is modular/expandable. You could build a less expensive one using fewer processors to control larger chunks (mabye all) of the display.

As pointed out by cadstarsucks the first thing you need to figure out is the timing. I THINK the following will work.

I would use an identical controller for each 8x8(16) matrix. You will need 24 processor pins to control each display plus 2 pins for serial IO. To get this you need to use a 40 pin processor which gives you several spares. (you could go with a 28 pin chip and use an external shift register, not worth it IMHO)

This pin count assumes that each LED is either on or off. The beauty of this is that you can build one controller for one display and test it. When you get it working you build up the other 19.

Light the LEDS by setting the 16 bits that control color (top) then activate the row by activating that rows bit(side). This cycles through the display in half the time.

You could do a proof of concept on a solderless breadboard. Hook the PIC directly to the display with current limiting resistors on the 8 side. The key here is that you should only turn on one LED at a time on the 16 side. The real thing would use transistors that handled more current so you could light all LEDs in a row.​

Your sign is a 2x10 matrix of these displays. Think of it as two seperate displays. The top 10 only need to store the top half of the image. Similar for the bottom 10. This cuts storage by 50%. So instead of 320 bytes you only need to store 160. This would work. You could move to the 18F series to get more ROM (and 48MHZ) if you need to. The PICs have a self write ability. You could use 2 banks of flash to store 2 images if you wanted. I am not sure exactly how that is done.

The 40 pin PIC16F877A has 368 byte of ROM. These are 20Mhz processors and cost about $5 each. Put transistors on each output pin. You can get transistor arrays in sip form.

Use a master controller to manage each matrix controller. It will program the images, set the initial position within the image. You still have to figure how to keep the displays in sync. Maybe use a dedicated pin on the master as a step line that is sent to each controller and serviced by an interrupt.

You can buy an Ethernet to serial converter for about $25 that will make that part simple. I can provide the link if needed.

3v0

 

[cadstarsucks]

As pointed out by cadstarsucks the first thing you need to figure out is the timing. I THINK the following will work.

I would use an identical controller for each 8x8(16) matrix. You will need 24 processor pins to control each display plus 2 pins for serial IO. To get this you need to use a 40 pin processor which gives you several spares. (you could go with a 28 pin chip and use an external shift register, not worth it IMHO)

3v0

There is the advantage that you mention further down, but if the part you recommend has the horsepower to drive the whole display, I seems to me wasteful to use 20 $5 chips when one will do with the addition of 12 $0.10 chips that saves you an additional 80 transistors...

While I might be a bit biased, I would still use the $2.20 ARM part. Particularly if Pepper is going to have to learn the dev package anyhow.

Dan

 

[Pepper]

Being experienced in programming and having OCD I tend to obsess with creating extremely efficient code, even though I have no idea what I'm talking about it seems that using 20 microcontrollers is a bit overkill and expensive. I feel that cadstarsucks' idea seems more practical. Iv'e read a few articles on PIC vs ARM vs AVR and Im leaning towards using AVR or ARM architecture. To me ARM seems like putting a 502 in a Civic, but thats just my observation. Thanks for the responses so far im getitng a better idea on how to complete this project.

 

[3v0]

There is the advantage that you mention further down, but if the part you recommend has the horsepower to drive the whole display, I seems to me wasteful to use 20 $5 chips when one will do with the addition of 12 $0.10 chips that saves you an additional 80 transistors...

While I might be a bit biased, I would still use the $2.20 ARM part. Particularly if Pepper is going to have to learn the dev package anyhow.

Dan

Not a PIC vrs ARM issue.

I am a bit worried about display brightness if you use a single controller to run the entire shooting match. We have no idea under what lighting condition the display will be used. It seems poster never give enough info. This could be the datasheet.

If one started adding shift register chips and maybe latches the board space could start to add up. I am not sure the extra $100 is such a bad deal from the hardware viewpoint.

If Pepper mocks the thing up as I suggested, and adds a few transistors to drive it he can get a better idea of what he needs and wants.


3v0

 

[cadstarsucks]

Being experienced in programming and having OCD I tend to obsess with creating extremely efficient code, even though I have no idea what I'm talking about it seems that using 20 microcontrollers is a bit overkill and expensive. I feel that cadstarsucks' idea seems more practical. Iv'e read a few articles on PIC vs ARM vs AVR and Im leaning towards using AVR or ARM architecture. To me ARM seems like putting a 502 in a Civic, but thats just my observation. Thanks for the responses so far im getitng a better idea on how to complete this project.

Hey, if the 1.5Ls cost as much as a 502, more Civics would have 502s!

Dan

 

[cadstarsucks]

Not a PIC vrs ARM issue.

I am a bit worried about display brightness if you use a single controller to run the entire shooting match. We have no idea under what lighting condition the display will be used. It seems poster never give enough info. This could be the datasheet.

If one started adding shift register chips and maybe latches the board space could start to add up. I am not sure the extra $100 is such a bad deal from the hardware viewpoint.

If Pepper mocks the thing up as I suggested, and adds a few transistors to drive it he can get a better idea of what he needs and wants.

3v0

3v0,
The brightness has nothing to do with the number of processors used. It has to do with peak current and multiplex ratio. A 16:1 (16 high side transistors) mux ratio will look the same whether driven by one micro or twenty. If you want an 8:1 mux you simply double the number of row drivers to make up the difference.

I am not sure what you meant by shift registers...you connect ABT family register BYTE serial. That is to say DB0-R1D0, R1Q0-R2D0, R2Q0-R3D0, .... Effectively creating 8 parallel shift registers. The ABT family will maintain logic levels at 62mA to GND and can safely deliver 128mA so it eliminates the low side drivers as well. On the high side you are still stuck with discretes as the peak currents can easily reach 5A.

I have found that 16:1 is a practical limit and that sending out the data byte serial is the most efficient way to get it out of an 8 bitter. Of course on the ARM you could make it a 16 or 32 bit bus, but that gets cumbersome in hardware.

And before certain other denizens challenge me on this as well, I HAVE designed displays that contained over 2000 LEDs in a matrix for production!

Sorry about the pricing above on the logic. It is probably more like $0.80, I forgot that I am used to the volume pricing that we get on them.

Dan

 

[3v0]

I agree with most everything you said.

No problem about the pricing. I sort of figured that was the case.

3v0

 

[Gayan Soyza]

I am trying to design and make a 80x16 Red/Green Led Matrix "Scrolling Sign". I have 20 8x8 RG led modules to use. I am somewhat new to Electronic Circuit Design so I'm going to need some help on how to do pretty much everything. I would also like to include an ethernet interface to be able to update the sign, but I don't want things to get too complicated.

I've searched the internet for hours for some sort of tutorial but couldn't find anything using RG leds or anything on the scale of 80x16. Any help would be appreciated, I'm even willing to give out monetary compensation for significant help.

Probably you need a microcontroller for this project. You cannot do it without getting complicated it’s a complicated project for a beginner. Before going to two or three colour matrixes have you tried it with single colour ones? How to control LED’s in a matrix, how to shift columns, rows, frames etc….at first forget the length of the Display (columns).Reduce to 5 or 10 columns & start learning. Like 5x7, 8x8, 10x7 etc…Once you come up to a standard you can extend upto a larger level. Its easy to control two or three colour matrixes but it needs a messy wiring.

Here is a link which has the basics of moving signs. Study it. I also learned many things from it.

**broken link removed**

 

[cadstarsucks]

Probably you need a microcontroller for this project. You cannot do it without getting complicated it’s a complicated project for a beginner. Before going to two or three colour matrixes have you tried it with single colour ones? How to control LED’s in a matrix, how to shift columns, rows, frames etc….at first forget the length of the Display (columns).Reduce to 5 or 10 columns & start learning. Like 5x7, 8x8, 10x7 etc…Once you come up to a standard you can extend upto a larger level. Its easy to control two or three colour matrixes but it needs a messy wiring.

Here is a link which has the basics of moving signs. Study it. I also learned many things from it.

**broken link removed**

I beg to differ. The wiring is not bad at all as it is primarily internal to the modules.

Dan

 

[cadstarsucks]

I agree with most everything you said.

No problem about the pricing. I sort of figured that was the case.

3v0

Here is a clip from an active design that might make more sense...

Dan