Go Down

Topic: Problems with multiple 5940's (Read 2090 times) previous topic - next topic

denniscj7

I have two 5940's, each with 16 leds. I can drive each 5940 separately -  with the Knight Rider example and they work perfectly. When I tie the two together, one works pretty well  and the other appears to have all the leds on - dimmed - and is fluctuating the brightness up and down - but not in the expected Knight rider sequence. Are there some special considerations when stringing 5940's together other than just connecting the 5 wires between the two as in the documentation? BTW I am using 480 ohm resistors -not 2k - in order ot increase the current. The leds are very high mcd. I am powering the 5940 and leds with separate power supply set at 7.0 volts.

Msquare

I think you have wired the yellow wire wrong, when you put two 5940s on. It should go to pin 12 instead.  :smiley-eek:

If that answer makes no sense, then give us a bit more info: The schematic and/or a good picture of your wiring. Whichever one looks on it, you are not doing "something" right. And the friendly people here can not guess without more info ;)

Grumpy_Mike

Quote
Are there some special considerations when stringing 5940's together other than just connecting the 5 wires between the two as in the documentation?

Decoupling capacitors?
http://www.thebox.myzen.co.uk/Tutorial/De-coupling.html\

Also
Have you got a capacitor on the clock signal? If so remove it at once as this will damage the arduino.

bios31

i tried to drive 18 tlc's and it isnt working!! it worked with 6, 12 but never with 18. changed this uint thing in the tlc config file.. but nothing worked with it.

now half of our tlcs r broken. mostly the sout of the tlc died.. no idea why. how is that possible?

Grumpy_Mike

Sadly when you scale things up it seldom woks. This applies to a lot of things, it's why home baking is better than shop bought and why you can make a paper duck out of a sheet of writing paper but you can't make one 10 foot long without changing something.

It boils down to signal integrity and power supply. Did you use a scope and see the signals as you added more?
The TLC9540 is not well suited for long strings due to the high frequency signals you have to feed it. These signals have trouble traveling these sort of distances intact. You have what is known as a transmission line problem when the line is longer than the signal can travel during one quarter of a cycle.

Quote
mostly the sout of the tlc died

That is interesting I wonder if the reflections on the line bounced back and damaged the output.

bios31

uhh ;D that has to be the problem!! but how can i solve it? shorter data lines? ive got 3 pcb's with 6 tlcs on it to drive totally 96 rgb leds. every pcb is connected together with an 12cm long cable. would it solve the problem if i use a thicker wire?

thx for your help!

Grumpy_Mike

If only it was as simple as ticker wire.
You really should get an oscilloscope on it and look at the signal, failing that then this is what I would do.

What I would do is to get some non inverting buffers like a 74LS07, these are open collector so they need a pull up resistor at the far end. Then I would send the signal to each PCB through an buffer for each of your three boards. Use a star arrangement rather than chain them. Then take the sout and feed it to the sin on the next PCB again through a buffer. Also I would wire the power to the boards in a star as well. Finally I would connect the boards with twisted cable like cat5, twist one signal and one ground together.

denniscj7

I finally got to the bottom of my 5940 problems. I was breadboarding 5 of them and getting all sorts of strange results. It looked like cross-talk in that when the program was controlling leds attached to one 5940, I would see flickering - random - on the leds attached to the other 5940s. I'm a retired electrical engineer and it occurred to me that the 140khz pwm signal was radiating all over the place. I proceeded to move from the breadboard to a pcb - used wire wrap and it worked great. Now that the 5 lines interconnecting the 5940s were very short and in the same plane as the pcb, the performance/behavior was rock solid. No problems at all.  I guess it's easy to forget that 140khz is in the AM radio band, and we know those signals radiate very well.

I hope this helps others.

bios31

im gonna cry  =(

now i did what u told me on here.. but its still not working. i used cat 7 network cable this time and tried to make the line as short as possible. i tried it with 2 boards (6+6 tlc's) but it wont work. one board is working (6 tlc's) if i connect the other board with my cable, it starts flickering like crazy and usually the first tlc on the second board gets really hot. i also tried it with this buffer but nothing..

these tlc's die that fast, that im so angry and  im so close to threw this thing out of the window!! -.- (the whole desk is full of "broken" tlcs....)

i wonder how you all did that..

whats else should i try?

Grumpy_Mike

Quote
i also tried it with this buffer but nothing.

How did you buffer it? Have you a photo or schematic?

bios31

hey thx for the reply!

i used an non inverting hex buffer the CD4050BE ---> http://pdf1.alldatasheet.com/datasheet-pdf/view/26878/TI/CD4050BE.html

wired it up like in the datasheet with vcc 5V from the sam power supply.

when i get my internet back, i will upload some pics. now im using an open wlan connection  :smiley-red:

Grumpy_Mike

That's not a very good buffer for this purposes. It will only drive 2 TTL loads, that is known as the fanout. You need something with a bit more drive like a 74LS07. These are open collector as well so you need an external pull up resistor.
It would be good to see a schematic of how you wired the buffers and the TLC5940s along with the decoupling you used. A photo rounds it all off.

bios31

today i will pick up six 74LS07 ;D and today my internet and phone should work again, so i will post some pics and tell you if it worked with this buffer. 

but.. where should i place the buffer? after the connection cable to the boards, or before the signals go through the cable?

lars_schreiner

Place the buffers right after your 5940s.

Any new results so far ?

I tried to drive one TLC5940 through 10m of CAT5 cable, which did not work. I am thinking about using another chip with internal PWM frequency source and then using some fast RS485 drivers for the signal lines. The differential signal would suite the CAT cables..

cheers,

Lars

bios31

#14
Sep 19, 2011, 05:57 pm Last Edit: Sep 19, 2011, 06:03 pm by bios31 Reason: 1
hola!

im sorry that im answering that late ;( friend was sick (we r doing it together) and we had to do other stuff.

now we tried it with the buffers and it seems to work.. its flickering but i think when we make a pcb for the buffer and do everything clean it will work (i hope so Oo). now we wired it hanging on the cable and isolated with glue ;D

here some pics of the whole thing...









dont worry about the wiring ;D it looks like that, because we took it many times out...

Go Up