Text LED Panel

[Fluence]

Hi,

I always wanted to build a rotating text LED panel, but i've searched on the internet, and it seems the people "avoid" to say how to build one...

So i want to ask if someone would be kind enough to tell me or to point me to a website to make this panel.

I really want to make a "big" or "medium" panel!!!

Thank you for your time.

 

[mneary]

A large panel is just an expansion of a small panel. Many designers handle this by building a lot of small panels and controlling them with a common controller.

Go to sparkfun.com and look for 12ft GPS Wall Clock.

 

[Fluence]

That's 99$!!!

I want to build from the ground the LED text panel, not to buy a couple of pieces and solder them up. D:

Do you have some tutorial or schematic, on how to build a led panel? With about 16 LED height and 128 LED width maybe.

 

[Fluence]

I've found this website: **broken link removed**

But it was made in 2005, so i would want to know if it's still "viable" to build it like it says on the page.

 

[BrownOut]

Dude, check this out! Buck and a half, think I'll order a couple these myself.

 

[Fluence]

Eh, it really is a good price, but...

I wanted to build one myself, not buying one already made. D:

It's just that i can find any "recent" nice tutorial, about doing it properly without having much knowledge of micro-processors.

 

[BrownOut]

Unless you get a real gonga on bulk LED's, you're probably looking at a significant cost. If you can get a good, surplus display module, and find a datasheet, you'd be much better off.

 

[mneary]

I pointed to the sparkfun site not to recommend any product, but so you can follow their design approach. They include their entire design process from beginning to end, with complete schematics. You don't have to buy anything.

If you wanted a schematic for your exact project ready to build, I have misunderstood you.

It's just that i can find any "recent" nice tutorial, about doing it properly without having much knowledge of micro-processors.

I cannot begin to imagine building a "big or medium" display without microprocessors.

 

[Fluence]

Yes i know i have to use microprocessors, sorry for the misunderstanding, what i mean, is that i can't find a schematic for a 16x128 message LED panel, i've found a little one 7x50 but that's too little i wanted to have a bigger one.

 

[mneary]

A large display is just several small displays. The 7x50 display you showed is still a viable design except the AT90s2313 has become obsolete (actually it was "not recommended for new designs" before 2005). That doesn't mean anything, because the code is easy to re-target to a new controller. Anyway for a larger display you would just use a larger micro controller; I would consider the ATMega32 so you have enough memory to store the pattern. Add more 'HC595s to drive more columns, and use the additional I/O ports to drive more row select transistors.

 

[Fluence]

I'm beginning to understand the "process" the bigger display i want, i can just "upgrade" the micro controller and add more HC595.

There's just one big problem, in fact, i'm interested in learning how can i do that with "real schematics", but right now i can't figure out how i can use the ATMega32 and connect it to the rest of the system.

I "think" i've understood what the old micro controller does and where it should connect:
**broken link removed**

But comparing to the ATMega32, i have no idea how to connect it:
**broken link removed**

They seem similar in the connections indeed, but before i begin to solder one ATMega32, i want to make sure im doing the right thing, and the ATMega32 has many other connections that i just can't figure out where i can connect more HC595 to upgrade the LED display. (well, i "could" figure out by "logic", but it's much more probable that i'll just burn the ATMega32 lol...)

 

[mneary]

Are you sure you're up to soldering two thousand LEDs? At one LED per minute, that's over 34 hours.

The additional HC595s just get put on the end of the existing ones. VCC/GND and pins 11, 12, and 13 are common. See how pin 9 out (SDO=serial data out) goes to the next pin 14 input (SDI=serial data in) and so on. The software that sends the columns out to the 595s through Port D (PD3, 4, 5, and 6) just gets a higher loop count. Since the ATMega32 has a hardware SPI interface, you might want to move the HC595s to those pins (PB4, 5, 6, and 7). If you do this you can replace the author's bit-banged SPI software with the much simpler hardware support. PB5, 6, 7, RESET, and XTAL1 also should be brought out to a serial programming connector. I'm assuming a serial programmer like AVR910 or STK500, so check its documentation for the pin out.

The additional rows will be identical to the existing rows on Port B. Actually, since you have 4 ports on the ATMega32, Port B may not be the best choice. Given other possible uses for many of the I/O pins, I would tentatively consider using PA0-7, PC4-7, and PD4-7 for the 16 rows.

The 24C16B is probably an EEPROM, I'm too lazy at this hour to look it up. The ATMega32's internal EEPROM might make this component unnecessary. You would have to check how much the program uses. If you keep the external EEPROM, the ATMega32 has built in I2C, you might want to move those two signals to PC0 and PC1 which are the ATMega32's I2C pins.

The 10MHz crystal is the MCU's timekeeper. It is the basis for scheduling the LED scanning and setting the baud rate of the UART.

I think it's not time to worry about soldering the ATMega32. I would want to have drawn an updated schematic and taken a first swipe at porting the software. Before you can start on the software you would be installing AVR Studio and WinAVR C development environment (all free). AVR Studio has an emulator so you can try out code one step at a time with no smoke no matter how bad the mistake.

I haven't looked into the program in detail, but the section I previewed appears to have its comments in Finnish. Google Translate probably would make sense of this. Another easy yet tedious task.

 

[Fluence]

Yes i'm up to solder 2000 LED. :>

But reading all that info, i got kinda lost, I've downloaded AVR Studio, but i don't know how to use it, anyway i don't even know how to code, i can "edit" code but creating new code... for now i don't have the knowledge to do that.

I've understood how to connect more HC595 but the ATMega32 part i got tottaly lost.

It would be nice, if there's a software to make schematics with the components, and that we could test them on the computer, i have Orcad 9.2, but i don't think i can run a "virtual test" on it.

I think it would be very troublesome for you building a schematic for me, but can you give me some advice and support with the schematic i will "try" to make from time to time? :3

 

[Fluence]

I've already edited the 7x50 schematic to a 16x125 LED panel.

It's in Orcad 9.2 format, can you give a look at it mneary?
**broken link removed**

Thank you for your help! :3

 

[mneary]

I don't have Orcad.

 

[Fluence]

Here's the ATMega32 close up, i know there's wrong connections there, but i just don't know how to connect it:
**broken link removed**

Here's the HC595s connected to the ATMega32 (i think the connections are good):
**broken link removed**

Here's the total view:
**broken link removed**

 

[mneary]

Definite progress. You must have made your own ATMega32 pattern, so it needs a little bit more cleanup. The are duplicate pins (PD4, 5), need Port B for the sake of completeness. Although I suggested pins for the row drivers, do some more research so you're comfortable with the choice. One opportunity lies in PC6 and PC7. If you keep these free then later you can hook up a 32kHz crystal and have a clock. Maybe PB0-3 are a better choice instead of PC4-7.

When you choose the transistors for the rows, remember that each row may have 125 LEDs turned on at about 15mA each, which totals almost 2A. The base drive is only about 20mA so if you use a BJT it will need to have a very high gain. (I'm looking around and I'm having trouble finding a cheap one... where are you located and where do you normally go for parts?) If you can handle a surface mount, the SI4834BDY from "dpi4parts" on ebay would be perfect, at about 27 cents for two in one. In this case it the MOSFET would just drop in and would not require a circuit change.

I've downloaded AVR Studio, but i don't know how to use it, anyway i don't even know how to code, i can "edit" code but creating new code... for now i don't have the knowledge to do that.

. This project comes with code, so you have a head start. Redefining it for ATMega32 and translating comments from Finnish are just bumps. He's done the hard work.

You'll also want WinAVR. I think I got mine on Sourceforge

 

[hdc090360]

A few quick suggestions. What is the fanout of the ATMega32? You have got the pins driving 16 chips in parallel. You probably need to put some buffering in there.

The ATMega32 can be driven at up to 16MHz so you might like to increase your crystal to 16MHz. This will give you more headroom in the code for the expansion you require.

Try darlingtons for the row drivers. They will give you the power and gain that you need.

 

[mneary]

I was thinking about darlingtons for the row drivers, but the difference in saturation voltage when lightly loaded (driving only a few LEDs) and heavily loaded (driving almost all) worried me that there would be a noticeable difference in brightness. It's true that darlingtons are plentiful and cheap, but then I saw the SI4834BDY for only 14 cents per transistor I recommend those instead. The MOSFET at .030 ohms will only drop 0.06V at 2A.

The HC595 is CMOS, so it draws no DC bias. The HC595 has a maximum input capacitance of 10pF, so 16 of them is less than 160pF (plus wiring) which gives a rise time of 35 nanoseconds when using the 200 ohm series resistors in the SPI lines. This still leaves plenty of margin for the SPI bus to run at 5MHz, and maybe 10MHz.

It will need to run faster than 8MHz, because it has to do twice as much work. There's new silicon out in the ATMega family that is faster and cheaper (ATMega324-20PU up to 20 MHz) so when it's time to order parts it deserves a closer look. Since the pins are the same and it's almost code compatible we don't have to explore that in depth yet.

 

[Fluence]

Hmm, you've said that the LED will receive a 15mA current, but i would like them to receive about 20 to 23mA at 3.2v, do you think it's possible?

Maybe i'll have to change the transistors and the HC595 for something more powerfull? Dunno what component is more powerfull but...

I think i'm close to have the schematic finished, but it's better to choose the components i will install now, and the schematic later.