I got some of the TLC5940 SR's. When I run the library provided, I am seeing color variations. I then tested a single chip at a time on a breadboard and find that some of the legs have different brightness levels when running the simplest of tests (all on - testing the outputs on a single light from each pin). (in some cases .. only one pin is off)
I would think with no dot correction set, they would match at max, but they do not. Different chips have different levels on different pins .. and it is very obvious with just one color and mixing shows it worse.
(it is not the LEDs .. I have tried many .. and even diff types)
I tried to use dot correction but no luck.
The chips were purchased form a major provider and in tubes, so I'd think they would be fine.
I am getting way better color matching using software PWM on 595s .. just seems wrong.
I changed the input from 5v to 3.3v input for the chips and that helped when I had three daisy chained, just so you know I tried that too.
Any ideas / anyone run into this before? Am I expecting too much from the chip .. to have the same brightness levels on the same LED on all pins .. out of the box with no dot correction?
When I have used them I have not noticed this effect. The dot correction is meant to compensate for different efficiencies of different LEDs. If you are seeing the same LED show different brightnesses in different channels it suggests that something is not right. Like the software feeding numbers into the shift register.
Try and look at the waveform on the LEDs cathode and see if they are the same when you set them to be the same.
Thanks for the reply. I would think if it was software, the diff ones would not have different legs being off and I am using the TLC library out of the box .. just turning them all on to 4095.
The TLCs are different from each other but then consistent from there. I am starting to wonder if I got a bad batch or something. A few seem OK, but most of them have 1 or 2 off pins and some have more "bad" pin outputs.
I am not sure what looking at the waveform means so i guess I am again .. in over my head .. it seemed so straight forward too.
I realize I did not supply enough details for a detailed response - mainly wanted to see if others to get consistent values. So thanks for the response / verification of that.
As nice as this chip is, I hate to just drop it, but may have to for RGB.
If I find the issue, I'll post it under "inconsistent color in the TLC5940" .. because I can find nothing under that search now .. making me think it may just be bad chips.
The waveforms will look a bit of a mess but basically you should be able to make out a consistent pulse width. This should be the same for all pins if the software is set to be the same for all pins.
Wow - what a fast reply Mike - thanks for this and many replies from you.
I do not have any capacitors in the layout .. just hooked it up like the sample shown in the tutorial (which may be too simple for perfect current).
As for power in general ...
In the simple one LED test with one SR hooked up .. I am pulling the power from the USB on the arduino but using outside 5v for more LEDs , it fails in both cases. I was seeing worse color issue with I had three of them daisy chained with 5V to the chip .. oddly going to 3.3v on the chip made it better (did that after reading some posts here).
Also .. I am feeding 5V directly to the LEDs and have a 2.2K resistor to the reference pin, which should be limiting the current to just under 20ma. But I may be missing something there.
I'll read up on supply decoupling and waveforms - thanks for the pointers Mike.
Great write up, learned some good stuff. Still same issue though.
I am seeing very consistent off values making me think there are some dot correction values in eeprom (on these new chips?).
I can not make the dot correction show any differences .. so I am still working that out. I tried the library code for dot correction and a few other slices of code with no luck .. YET :).
I'll update again on new findings when I get back on it ..just didn't want to "drop off" with no more response after other put in time to reply.
If it is any consolation I can't get the dot correction to do anything. I think it might be that the chip needs a higher voltage to program the eeprom.
making me think there are some dot correction values in eeprom (on these new chips?).
This is entirely possible, it could be something left over from the manufacturing test of the chip and the wipe clean state got missed.
I have been playing with the TLC5940 and noticed this as well. I have an 8x8 multiplexed RGB matrix with the 2 x TLC5940 on the negative leads, and a 595 shift register controlling the positive.
I'm not entirely sure what causes the dimness on specific pins of the TLC5940, however I was able to correct it by using 2n3904 NPN Transistors on the positive from the shift register. I believe that with the switching, I was not getting consistent current draw. I also made sure I was using a consistent voltage source as I ran into a number of issues attempting to just use USB power.
This significantly corrected the issue for me, I do not know if this would work in your case if you are individually adressing the LEDs.