This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LM3405A Problems

Other Parts Discussed in Thread: LM3405A

Hi,

I have some trouble getting the LM3405A to work properly - I'm using the circuit and recommended parts from page 18 on the data sheet. Vin=5V driving a single LED @ 1A I do not get 1A current instead I get

R: 0.61A
G: 0.32A
B: 0.4A
W: 0.38A 

The output current is dependent on the forward voltage of the LED and not even close to 1A. I tried changing various components e.g. L1 to 6.8uH, D2 to BAT54, I also verified that R1 is indeed 0.2R. Increasing Vin to about 6.6V brings the current up to the correct range. I'm attaching the PCB I'm using - any feedback is greatly appreciated.

  • Is EN/DIM floating? The layout looks ok although I always try to tie the GND pin of the device more closely to the CIN and catch diode grounds. With vias so small I would probably at least double up on them to tie both points to the ground plane better. Of course you can solder small wires through the vias (or completely fill with solder) to solidify that and see if it helps. Otherwise could you get some scope shots of the SW pin and LED current when you are at 5V and not getting the current you expect? What are the forward voltage drops of the LEDs you are using?
  • The EN/DIM is tied to 5V - Using PWM input works as well just not at the correct Imax. I'm using a Cree XML RGBW (single color/per driver). The Vtyp are R:2.2V/G:3.3V/B:3,1V/W:3.1 V. This is already the second board I made following the PCB recommendations in the data sheet. I first one showed the same problem but with currents even smaller. Let me try to make scope shots tonight ...

    Thanks. 

  • That shows the signal at the SW pin - btw/ shorting the LED drives the current to exact 1A. Also, it seems extra grounding though the vias does not make a difference. The signal the the LED is basically a flat line - at least with my scope (50MHz) I barely see any ripple.

  • You are in maximum duty cycle so it isn't able to deliver full current until the output voltage gets low enough. But you shouldn't be with those LED voltages and 5V in. You are just using 1 LED on the output of each? Do you have long leads or an ammeter at the input that could cause significant voltage drop between the 5V supply and the PCB? Have you measured at the PCB input to make sure it is 5V? You should only be around maybe 70% duty cycle for the green LED so something must be causing a voltage drop at the input.
  • I remeasured - the voltage at the PCB was indeed a little lower - so I adjusted the power supply that I have now exact 5.0 at the PCB Vin pin.  Instead of using an DMM anywhere in the LED path I used now the power supply current reading which is @0.68A - frequency at the SW pin did not change. Cranking up the voltage at the PCB to 5.8 V will give me 0.93A but >5.8 it breaks down. Voltage at the L- pin on the PCB is 0.233 and at the L+ pin 4.41V. At the green LED it's 3.96/0.78. The leads are about 39 inches long as seen on the image. Measurements are done with a Fluke 73III.

  • So you have a multimeter in series with the LEDs as well? Your LED+ voltage is still pretty high, there has to be some voltage drop created by the multimeter and the long lead lengths, probably mostly the multimeter. This is a very low headroom design and adding any voltage to the load (or dropping voltage at the input) will force you into maximum duty cycle. Can you remove the fluke and use a current probe instead so you are not adding voltage at the load?
  • No, no more DMM in the LED path. I just used the power supply to read the current

  • Let me solder in the short leads from the LED directly onto the PCB's L+/L- pads. These are the red/black 9 inch long cable you see in the image. I could also shorten these little but not a lot as I need a minimum cable length in my final design.
  • We are getting closer ;-) I shortened the cable to 6" and soldered it directly into the PCB. I see now 0.87A@5V and interestingly 0.97A@4.5V - Let me use the scope tonight an see where we are with respect to the duty cycle in both cases.
  • The duty cycle @5V is now 592ns - but still not at 70%. But I guess it comes all down a very short distance to the LED and minimizing any kind of voltage drop. Increasing Vin might be the best option. Thanks a lot for all your help in debugging this.
  • No problem. Like I said, you are kind of on the edge with that application for a switcher running at 1.6MHz, so any extra drops are difficult to deal with. A higher input voltage is a good idea, but if you really need the 5V you can consider a lower switching frequency buck converter. There are plenty that can easily do this, but of course the solution size would grow with a lower switching frequency.