We have one of these on a custom board we designed. Behavior is found to be erratic.
History :
We populated a complete board (ARM Cortex CPU and various other peripheral ICs) and it appeared to work correctly first time, albeit with rather more ripple than wanted. This was put down to using incorrect components that we had in stock at the time, for example 4.7uH inductor, 3 x 10uF Cout etc. But the device faithfully operated across the entire Vin range 2.5 to 12V and didn't seem that sensitive to the incorrect values.
The correct parts were duly ordered but in the meantime something happened to the one working prototype. We're not sure if it was the person using it did something or whether it happened on it's own... It appears that Vout rose above 3.3v at some point and destroyed the CPU. At the time we were not too concerned by that with such out of spec parts (L and Cout). In particular the inductor had a low Isat and speculation was that at some point it may have gone into saturation and destroyed the Vreg, exposing the load to Vin.
A new prototype was assembled with the following values but now operation is completely erratic.
Test loads 12K and 24R resistive (ie have only populated power supply components, not rest of PCB).
Vout : ~3V3 (R1=1M, R2=180K)
L : 1.2uH ~2A Isat
Cout : 3 x 22uF 6V3 MLCC
Cin : 2 x 10uF 16V
Other : Vaux cap 0.1uF, Cap parallel R2 10pF
The PCB layout is similar to the EVM. Differences are : PS tied permanently to GND via a 8mil trace. EN Tied permanently to Vin right at the EN/Vin pins. PG : NC.
The datasheet says connect GND to PGND at only one place which we've done (ie by GND pin to pad under device), but interestingly the evaluation PCB layout shows a via connecting GND at R2 to the ground plane - we added it (by drilling and soldering PCB) but it made absolutely no difference...
The required current will be in the order of 200mA max when PCB fully populated.
What we are seeing is the device will not start up until approx 6V and then operates correctly up to 15V (which was as high as we tested). Once started it will operate down to approx 4V, but then one needs to go to 6V to restart. Adding or removing Cout alters the behaviour - 5 or more 22uF on Cout stop operating altogether. According to our capacitance meter the 22uF are actually more like 14uF so for one think you might think 5 would be better (5 x 14 = 70 which is in the order of 66 specified), also MLCC capacitance drops in proportion to applied DC voltage so I wouldn't really expect 5 x 22uF Cout to be too high... The first prototype operated over the full Vin range, and certainly lower than 4V, lower than 3V in fact.
When the device is operating you can see switching wave form at L1. When Vin is below 6V and there is no output, there is also no switching wave form whatsoever at L1, therefore it appears to me that there is some static condition the device doesn't like.
We could try cutting and strapping PS to Vin instead of GND - one question - I assume this Vreg is intended to power up in power save mode? It certainly always did with the first prototype...
The TPS63060 is so simple in terms of number of external components and connections, it's difficult to know what else to try really.
If anyone has any insight or suggestions I'm all ears!
One thing I do notice is that Vaux tracks with Vin up to approx 11V where it stays if Vin rises higher. The datasheet specifies a much lower value for Vaux in abs max. Do not know if that is indicative of partial device failure.
We'd really like to use this part in our design as it is very low part count and allows large freedom of power source - at least on paper ;)
PCB layout : http://www.icemans.co.uk/tps63060.bmp
Simon.
