LM3414 LED Driver Issues

[boye]

Hi.
I'm designing a controller for a 100W RGBWW LED using LM3414 constant-current LED driver chips. I've encountered an issue where the Blue and Warm White channels aren't functioning properly.

My Setup

• 100W RGBWW LED with the following specifications:
  • Red: 20-24V @ 900mA
  • Green: 30-34V @ 900mA
  • Blue: 30-34V @ 900mA
  • Warm White: 30-34V @ 900mA
• Using LM3414 LED driver for each channel
• PWM control for brightness adjustment

Problem While the Red and Green channels work as expected, the Blue and Warm White channels output extremely low voltages. Here are the voltage measurements at different PWM duty cycles:
Red Channel (Working):

• 25% - 0.77V
• 50% - 1.56V
• 75% - 2.35V
• 100% - 3.15V

Green Channel (Working):

• 25% - 0.67V
• 50% - 1.36V
• 75% - 2.04V
• 100% - 2.74V

Blue Channel (Not Working):

• 25% - 42mV
• 50% - 82mV
• 75% - 122mV
• 100% - 161mV

Warm White Channel (Not Working):

• 25% - 33mV
• 50% - 69mV
• 75% - 104mV
• 100% - 138mV

Circuit
Schematics:

PCB frontside:

PCB backside:

Any insights why I am seeing these problems, and general feedback on the PCB. It is my first PCB design, so I highly suspect there is room for improvement.
Also is the LM3414 the correct chip for this job?

 

[betwixt]

The PCB layout actually looks quite good but the location of the diodes isn't clear and the connectors are not properly numbered

One thing that doesn't look right, and it may just be a feature of the PCB package you used is the pads on the connectors are gray which I assume means they are not covered in solder mask but the underneath of the ICs is red and that is the ground connection that must be soldered down.
You could consider flood filling the bottom (blue) layer as ground as well to reduce ground resistance. The two working channels are closest to the power input, the other two have longer paths to it.

As an experiment, try adding a wire link between CN1 ground and the ESP power ground to see if it makes a difference.

Brian.

 

[KlausST]

Red Channel (Working):

• 25% - 0.77V

Please tell us where you measured these values

Klaus

 

[boye]

Thanks Brain. I will try that.
Klaus I have measure the voltage using the test pads.
So a multimeter connected to GND and PWM-R, then GND and PWM-G and so forth.
I have also tried inputting a pwm signal directly to the test pads. It works for RED and GREEN, but not BLUE and WW

 

[KlausST]

Klaus I have measure the voltage using the test pads.

I can´t find test pads on the schematic.
And on the PCB ther are 20+ tast pads ... but I dont want to spend the time to find out where they go schematic wise. ..

But if you measure on a PWM, let´s say 0V/3.3V then the displayed voltage will depend on which method your DVM does the calculation.
A averaging meter will show: 3.3V * duty_cycle
A true RMS meter will show: 3.3V * sqrt(duty cycle)
some RMS meters have a high pass function and thus will display even different values. (I don´t want to spend the time for the calculations)

Klaus

 

[boye]

I can´t find test pads on the schematic.
And on the PCB ther are 20+ tast pads ... but I dont want to spend the time to find out where they go schematic wise. ..

But if you measure on a PWM, let´s say 0V/3.3V then the displayed voltage will depend on which method your DVM does the calculation.
A averaging meter will show: 3.3V * duty_cycle
A true RMS meter will show: 3.3V * sqrt(duty cycle)
some RMS meters have a high pass function and thus will display even different values. (I don´t want to spend the time for the calculations)

Klaus

So fair.
Thanks for the feedback. Yeah I put test pads for everything just in case. Their is a test pads for each power line and input/output pins for the chips. I know it's a bit confusing to look at. Especially when it mirrored.

Another thing. Do I really need the buck converter? The G, B and WW channel needs 30-34V and the R channel needs 20-24V. However the chips are constant current, so will it be fine without?

I did used an averaging meter. I will try to hook up the PCB to a logic analyzer at work on Tuesday. I guess that will give some more information.

Also is there any information that would be helpful to make it easier to give feedback? Both for this specific question, but also in general, so next time I have a question, I can make it better?

Best,
Christoffer