I have implemented what should be a reasonably close replica of the reference design for this chip.
Input voltage is nominally 12V and output 5V. The circuit works apparently OK until output current reaches close to 700 mA at which time the output voltage collapses. Removing the load will then restore operation.
The layout looks like this:
It is a 4 layer board with ground fill on all layers (gray in the image, unfilled for clarity). The colored outlines are other unfilled copper areas. Orange Vin UVLO and yellow FB tracks are on the bottom side.
The related schematic looks like this:
Output caps are 47uF 6.3V tantalums. The upper feedback resistor R22 is in fact 12k, not 10k as shown. Also, R21 is 100k instead of 560k.
I have tried the following:
-Moving the compensation nodes has no effect (C26 -> 470pF; R27 -> 22k) no visible effect. The circuit appears stable before and after and the current limit is unchanged.
-Increase the inductor 2.2uH -> 4.7uH; no effect on current limit.
-Increase the boot cap from 220nF to 470nF; no effect as makes sense (an earlier test circuit failed to work with the recommended 100nF cap but did work with a 220nF)
- Introduce a modest lowpass to the FB by paralleling R26 with a 470pF cap; no effect
- Increase input voltage from 20VDC to 24VDC; THE CIRCUIT NOW WORKS CLOSE TO 800mA but then crashes again!? WTF?
At all times the scope trace from the switching node was very clean. Output voltage has modest spiking and some ringing that attenuates quickly - could be partially from the scope ground lead.
At this time i can't come up with any other moving parts that could be tweaked. Is there a fundamental problem with the layout because the schematic is practically identical to the reference? Any advice is appreciated.
