Table LED Matrix, Weeks of debugging, column problems

[mekohler]

Im concered that the LED clusters are not as bright as when I tested them individually, and they are barely shining through the diffusive glass I have over them.

Kevin suggested to change the program to not run pulses. Is anyone familiar enough with this to help me?

Here are some measurements:

Voltage:

1) From PSU Ground to a single LED Cluster (6 LED's in parallel with a resistor) COLUMN input: 5.25 V

2) From PSU Ground to single LED Cluster (6 LED's in parallel with a resistor) LED Cluster ROW input: 3.85 V

3) From ROW input to COLUMN input of a single cluster: .86V (I dont think this means anything, since this is an incorrect way to measure V)

Current:

4) Removing one LED cluster in the all lit column and connecting column output of LED below and column input of LED above to the meter: 2.53 using 20mA setting/resolution

5) Removing one LED cluster in the all lit row and connecting row ouput of LED to the left and row input of LED to the right to the meter: 12.17 using 20mA setting/resolution

[mekohler]

Update: I still have row 1 and column 1 on for testing. I removed one of the clusters in the first column and tried lowering the resistor values of 300 to 150, 47, and no resistors. The brightness obviously goes up as I go down in R, but even at no resistors the cluster is not as bright as an extra "test" cluster detached from the table and connected to a 9V battery.

I used this extra cluster connected to a 9V battery for my intial table construction, as it was bright enough for the glass, and I assumed it would be getting 12V, so I knew for sure it would be sufficient.

Any ideas?

[nomad]

the plain fact is it is a matrix. in order to control each led without having a separate output for each led it needs to switch from row to row or col to col. in other words scan the matrix. if your seeing hy flash, then it is not scanning fast enough. by adjusting the scan rate and pulse width you can control the brightness. as loannis said it's doing what it's supposed to be doing. with a scope you can actually see the pulses and determine if they are too short or too far apart or anything else that is happening with them. if you just want them all on or off take out the pic and insert a switch.

[nomad]

you will never get the matrix as bright as your test cluster. to switch say 5 rows only one is on at a time. so each row is on only 20 percent of the time max.

[nomad]

i'm not a C programmer but it looks like the line that says 4ms delay is your delay between rows. you could try making 259 a smaller number. i think that would get rid of your flicker but may make it even dimmer.

[mekohler]

i'm not a C programmer but it looks like the line that says 4ms delay is your delay between rows. you could try making 259 a smaller number. i think that would get rid of your flicker but may make it even dimmer.

There is no visible flicker in person. It only showed on the camera because of the fast framerate it captures at. The only problem is brightness.

Update: I tried increasing the delay...that helped brightness a little but made the flicker noticeable to the human eye

[Ioannis]

Yes, that's the price one has to pay, I 'm afraid. You cannot scan the matrix without a little flicker.

On the other hand, you may increase the current that LED's draw. Do you have the data sheets about that LEDs? Are they capable to withstand more than 20mA?

On a recent project I used Kingbright LED 5mm @ 20mA. They were really bright! They are 10000mcad and may replace car lamps (that's what I intend to do soon).

Ioannis

[T.Jackson]

i'm not a C programmer but it looks like the line that says 4ms delay is your delay between rows. you could try making 259 a smaller number. i think that would get rid of your flicker but may make it even dimmer.

The faster the better is what I concluded with all the research I did in this area. Forget about any sort of delay on top of pulse width control. Just scan the thing like a bat out of hell.