Hi Flashlightguy,
For the LTC3490 circuit, I am currently using the following:
1x LTC3490 in SOIC format (the DFN is just too damned small for me).
1x Toko A920CY 3.3uH shielded inductor.
2x Kemet X5R 4.7uF 0603 capacitors.
The photos in my above links show the second circuit I ever built with this IC. It used the Coilcraft MSS5131 3.3uH inductor and a 2.2uF capacitor of unknown brand.
This circuit is extremely sensitive to lead lengths. Especially the lead lengths from the power source to the IC and the IC to the LED. It is also extremely sensitive to the DCR and peak current capability of the inductor. The datasheet recommends an inductor with a DCR lower than 25mohm and a saturation current in the range of 2A.
The Coilcraft MSS5131 has a DCR of 32mohm and a saturation current of 1.53A, while the Toko part has a DCR of 24mohm and a sat current of 1.89A. The coilcraft circuit lasts 30mins to 50% in my testing, while the toko based circuit lasts 45mins. All other factors were the same. Same battery charged in the same way, using the same RY0J Lux1 for testing. The reduced DCR and higher saturation current of the toko gave me results that were 50% better than the coilcraft.
Note also that I am using 2 capacitors. Even though the datasheet states that an input capacitor isn't required, I added one to reduce the pulse strain on the battery. A tantalum capacitor with its increased ESR would likely hurt efficiency as well.
This circuit has been performing very well for me, but giving me trouble in another area. I have been finding it very difficult to get good efficiency readings from this circuit due to the fact that it is very sensitive to resistance inline with the power source and the LED. Basically, you measure current by measuring the voltage across a resistor in series with the circuit and using ohms law to find the current. The problem is that the introduced resistance decreases the efficiency of the circuit and throws off the measurement.
For example.. Using the same circuit that gave me the 45 minute run time. I placed one 0.1ohm resistor in series with the battery and another in series with the LED. I used djpark's "CPF Logger Lite" to collect efficiency data while also monitoring the light output. My measurements showed ~80% efficiency from the start and fell to ~60% efficiency at the end of the usable battery life. The problem was, with the added inline resistance, the circuit only lasted 30 minutes to 50% brightness. This same circuit, with the current measurement resistors removed, lasted 45 minutes to 50% output.
So the efficency of the circuit is greatly affected by inline resistance, but I am at a loss as to how to determine the circuit efficiency without substantially decreasing the performance of the circuit.
Keep in mind though, that I am able to accurately measure 350mA output to the LED. I am just unable to accurately determine how much power the circuit burns away as heat while powering the LED.
Hope this helps,
pb