Problem with Switching Power Supply / LED Driver

[masoandro]

Hi there,
I am trying to drive a LED backlit display using an LED driver which is essentially just a switching power supply. This is the one I'm using: FAN5333B.

The back light takes 120mA into the cathode, and 3.2-3.4 V into the anode. Using the driver datasheet I constructed a power supply to go into the LED back light area. However, it fails when VCC is above 1.6V - it seems to short and my current draw shoots up to the maximum that my power supply will allow! I want it to work work when VCC is ~3.3 but this portion of the board causes problems going into the switching pin of the power supply (the pull is no longer present when the inductor is removed). I've tried removing all of the diodes pictured but the error seems to be in the switching pin.

I am attaching a picture of the schematic I have used as well as the portion of the board in question. Is this an appropriate design for such a power supply? What would cause the power supply to fail at that particular threshold?

I appreciate your insights, I learn a lot from this forum!

[ogden]

What's purpose of D1? Please show LED(s) in your circuit as well - it's not clear how and where you connect them.

[edit]  Typical application of this driver does not suggest FB smoothing. Try removing C18 and see how it goes then. Hopefully LEDs are connected between LED_IN and LEDK? If not, then this is your problem.

The back light takes 120mA into the cathode, and 3.2-3.4 V into the anode.

Just noticed that you most likely use single LED. Particular LED driver is boost converter. LED string voltage + current shunt voltage shall exceed VCC voltage. For single LED you need buck converter having VOUT < VIN.

[masoandro]

Thank you for your reply!

What's purpose of D1? Please show LED(s) in your circuit as well - it's not clear how and where you connect them.

D1 is a regular Zener preventing a surge from the switching voltage to the feedback voltage (although this is very unlikely).

Typical application of this driver does not suggest FB smoothing. Try removing C18 and see how it goes then. Hopefully LEDs are connected between LED_IN and LEDK? If not, then this is your problem.
I will try this and report back.

Just noticed that you most likely use single LED. Particular LED driver is boost converter. LED string voltage + current shunt voltage shall exceed VCC voltage. For single LED you need buck converter having VOUT < VIN.

It is actually part of an LCD screen. I've attached a picture that shows the backlight connections. It is not a single LED so I chose this driver as they seem pretty standard.

[SiliconWizard]

Well, the FAN5333 is specified at a Vin min of 1.8V, so under 1.6V it's probably not even starting. You're stating that it "fails" above 1.6V, but not what happens below that, but I would assume it doesn't start.
So your issue is basically that the current draw is over what you're willing to accept. What's the maximum your power supply will allow? Have you looked at the start-up response graph in the datasheet? It will draw a fair amount of current before getting in steady state.

Your issue may just be that the start-up current draw is over your current limit.

[ogden]

It is actually part of an LCD screen. I've attached a picture that shows the backlight connections.

Instead of showing connector w/o any pin naming, better just tell "LED's are connected to signals LEDK and LED_IN". Is it so?

It is not a single LED so I chose this driver as they seem pretty standard.

How many? What is expected voltage of whole string at nominal current?

[masoandro]

Yes, the LED's are connected to the signals LEDK and LED_IN.

The expected voltage is between 3.2-3.4 supply voltage (Vf) with If=120mA.

After removing the capacitor C18, I still am getting the same result. The switching pin hovers around 1.7 and the feedback voltage Vfb is about 6mV.

Also @SiliconWizard, I checked out the start-up response graph, but it's a little difficult to read. It possibly jumps up to 1.5A as far as I can tell - although this would seem like a problem, I've checked this design with another screen and their power supply is even lower - 320mA.
But you are on to something - when I put 3.3V directly into the board with the switching mechanism on it, I read exactly what I expect: 3.3V out, 320mV Vfb and it stabilizes at about 0.05A. Supplying it through the connecting board, I am only seeing 1.6-1.7V VCC and my current jumps to 0.6A, even though I am also supplying it directly to the board (i.e. it is not going through any other ICs).
I have an extension cable as well that connects it to the main board - the extension cable is 10 feet (3m) in length. Supplying 3.3V to this far end of the extension cable causes the screen to flicker - the power switches between 1.7V and 3.3V. I'm very lost by this behavior, what can be done to smooth the power and get a consistent signal?

[SiliconWizard]

Step-up DC/DC converters have peak currents that can get fairly high. The FAN5333 has an integrated switching current limit that should limit peak current to about 1.5A. It WILL get that high during start-up for a longer time on average that it will in steady-state (that's called inrush current). You have to figure out if your power supply, including all cables, can provide this amount of peak current. Just remember that it doesn't have to do so continuously. That's why you could get away with a 300 mA max continuous PS, as long as it can provide peak currents much higher than this.

You could try adding a 100 µF input capacitor (keep the 10 µF ceramic input capacitor as close to the chip as you can) and see if that improves things. You can test higher or lower cap. values until you get it to behave.

[phenol]

also, make sure that there is no ringing (during power-up when supply voltage is applied) exceeding the max ratings (6v?) on the input of the driver. This is a distinct possibility with ceramic capacitors and long wires.

[german77]

Sorry if I'm mistaken, but the ground path seems to go all over the board or even not connected. I would expect to have all grounds in the schematic connected directly and very close to the same ground specially because the current it's switched other wise you will generate a nice antena.

Here is what the datasheet says, almost at the last page

"PCB Layout Recommendations
The inherently high peak currents and switching frequency of power supplies require careful PCB layout design. Therefore, use wide traces for high current paths and place the input capacitor, the inductor, and the output capacitor as close as possible to the integrated circuit terminals. The FB pin connection should be routed away from the inductor proximity to prevent RF coupling. A PCB with at least one ground plane connected to pin 2 of the IC is recommended. This ground plane acts as an electromagnetic shield to reduce EMI and parasitic coupling between components."

A quick check it's just to solder a jumper wire to the grounds of the input and output capacitor.

[masoandro]

You could try adding a 100 µF input capacitor (keep the 10 µF ceramic input capacitor as close to the chip as you can) and see if that improves things. You can test higher or lower cap. values until you get it to behave.

Sorry it's taken me a while to respond. I was testing out some more things on the board to verify this case. But it seems as if this did the trick. I put on a 100uF low ESR tantalum and the power stabilized right away. It did not do anything for my passive RC filters however, I think the line is simply too long! But the signals are received now, I will add some SPI terminating resistors to be sure.

Also, the ground path is connected both through the incoming wire connectors as well as through the mounting screw holes where it is connected to external housing.

Thank you so much for your input! This project looks great now!

[SiliconWizard]