Adapting and making a LED driver

[Cosmin3]

Hello,

I'm trying to adapt a LED driver from Mic2282 evaluation board mic2282_eb.pdf
using the calculations shown in mic2282.pdf at "Application information" chapter.
It has 3 100mA white LEDs but it should use a 6 x 6 5/10/15mA white LEDs matrix.
5/10/15 = the user of the driver could set 3 levels of luminosity.
Only one branch should be connected to SNS and R2.
The other 5 branches should be connected to GND with 5 other resistors (same value as R2).
L1 = 15 uH / 3.3 A
C2 it says on schematic 47nF but below the schematic it says 47uF. I used a tantal 47uF.
R2 = 39 ohmi.
C1 100uF a tantal capacitor.
R1 = 20 Ko
C3 = 440 pF

First I tried to make a PCB from the default PCB in mic2282.pdf (slightly modified).
As input I used 2 1.2 V 2000mAh rechargeable Energizer batteries.
As output only a LED branch (6 LEDs x 2.9 V 5 mA).
No light in LEDs but the MIC2282 burned out after a few seconds.

Now I'm using a test PCB (with holes every 2.5 mm).
Same problem but now I'm limiting the input current.
At 1 V input I get 4.5 V output.
If I connect 2 AAA 1,2 V 800 mA Energizer batteries for a few seconds, I get ~ 7.5 V output. Still no light in LEDs (it has to be 17.62 V anyway).
But it's consuming a lot from the batteries so it's definitely not worth using it instead of the current resistors.

What am I doing wrong?

Thank you for your help.

 

[Cosmin3]

I'm not 100% sure but I think the problem may be from L1 (well, at least partially).
Its current value (15 uH) is calculated for 17.62V and ~100 mA output and for the "worst case scenario": 1 V input.
But to charge C1 from 0 to almost 17.62 on a few volts input I think it's not that good. Although it's only 100uF tantal not 10000uF electrolytic...
The circuit is taking 1+ Amps for seconds and overheating until it burns.

LE: Btw, is there any way I can simulate this circuit in a computer application like LTSpice, Multisim or other?

 

[Cosmin3]

To continue my monologue:

I increased L1 from 15uH to 45uH. The output voltage raised by 10..15%, definitely not enough to reach 17.62V.
Well, theoretically I could increase to a few hundred uH and reach 17.62V.

Am I using wrong the formula to calculate inductance from datasheet?

Vout = 6 x Vled + Vsns = 6 x 2.9 + 0.22 = 17.62V
The current for a branch is 0.22V / R2 = 5.641mA (the user might use 2x or 3x)
All LEDs: 6 x 5.641 = 33.846 mA
Iout(max) = 3 x 33.846 ~ 100mA
Vin(min) = 1 V
efficiency = 75%.

Iin(max) = (17.62 x 100) / (1 x .75) = 2349 mA

L = (1 x 0.7) / (2 x 2349 x 20) ~ 7.5 uH.

And I used 15 uH but apparently is still not enough...