Hello,
I am working on a project controlling RGB LEDs with TLC5940 constant current driver. And I am observing something strange that I can't figure out why.
I've connected everything pretty much like in the datasheet (see attachment). The controller and the LEDs share the same power supply (5V). I have capacitor between Vcc and GND.
I wrote an application to fade-in/out the red/green/blue of one RBD LED connected to channels OUT0/1/2.
When i power the board from 12V DC power adapter through 7805 everything is working fine, the RGB led is lid as expected.
When i power the board from 5V line of a computer power supply or from the 12V line through 7805 the red/green/blue led are lid at max (no fading). It's like the tlc is not controlled from the controller. I've attached a logic analyzer and I can see that the serial communication is working, PWM is working, but the RGB led is not controlled.
The only difference is the power supply. In the first case it is cheep chinese 12V DC power adapter and in the other computer power supply.
Edit:
I've borrowed a bench power supply from a friend, the same behavior like the computer power supply.
tlc4950 datasheet
add more local decoupling on the power pins ? each IC has 100nF decoupling and the main rail has a few hundred uF.
maybe even a series resistor to the power supply?
Try a series inductor? Using 12Vdc plug packs I normally use a common-mode, parallel cap and then a diff mode inductor to remove any crap from the line,
look at the line on a scope if you can (in both cases), with no long ground probes,
Monitor the led voltages with different power supplies. How hot is the Vreg getting? how much current is flowing with the Vreg and 5V from PC PSU? That should show some tell tails.
Have you disconnected all oscilloscope probes as you powered the board via the PC PSU? If not, it may be a problem with the fact, that PC PSUs are mains earth referenced. Also lab PSUs can be configured to be mains earth referenced. So If you have connected the scopes GND clip to a other potential than GND it will short that paticular node in your circuit to GND. But if that is not the case, then the only problem I can think of, is what Christopher described. That the other PSUs have too much voltage ripple. Although I will be very surprised if that ripple would prevent your circuit from working. Even in the case when you use a 7805 regulator to power the board. You pretty much have to have a ripple of 5Vpp @ 12V and more on the input that this will be a problem with a 7805 regulator.
I've added 100uF to the tlc and controller and added 220uF close after power supply - nothing changed.
Added series inductor (220uH) - nothing changed.
Added series resistor - nothing changed.
It works somehow (unstable, works ones every ~5 power ups) with the 12V line of the PC PSU when i have only only one 100nF across Vcc and GDN and use the 7805.
In the first attachment you'll see CH1=Vcc, CH2=BLANK, during the BLANK signal all the outputs of the tlc are turnd-off, during this period i am sending the new data for the outputs (24 bytes) in the picture you'll see that during the active (high) state of the BLANK, the BLANK has 24 artifacts from the 24 bytes send to the SIN.
In the second attachment you'll see the same BLANK signal, but zoomed to the falling edge. You'll notice that there is a swing of about 900mV on the Vcc. If i add more capacitance (220uF+47uF+47uF) the swing is slightly less (about 800mV).
The pictures are almost identical with the 12V adapter and the PC PSU.