Issue 1 - unstable when Vtrip < 0.2V -------------------------------------------------- The data sheets specify Vtrip min 0.2V. We initially had 10k to GND on this pin, i.e. 0.1V. That made the controller extremely unstable, particularly in FCCM mode, and the output voltage jumped from the target voltage to 9V. One thing is that OCL would not work with Vtrip < 0.2V, but it is totally unacceptable that the output can run completely out of control like this just because a trip voltage is set too low. The datasheets do not have any warnings about such a behavior. When we changed Rtrip to >20k Ohm, i.e. Vtrip > 0.20V (we are now using 27k as minimum on all designs), the controller became stable. This issue was seen on all designs we have (5 different Vout). There are all kind of over/under voltage/current protections, so that a too low Vtrip can result in a Vout completely out of control, is totally unacceptable.
Issue 2 - FB rises to 0.85V --------------------------- We have two different PCB designs. One design has 4 x 6A POLs, the other design has 4 x 6A POLs + 1 x 20A POL. We have had two samples of each board running in our lab for two weeks with no issues after issue #1 was solved. But then, one day turning on the board with the 20A POL, it was dead. We found that the 0.85V 20A POL was totally wrong. The device it powered was killed by a severe overvoltage of 1.8V from the 0.85V POL. The device now tries to draw > 25A on 0.85V. Normal load is 12A. So there is a severe damage inside the chip. We see that the 20A POL controller is operating totally wrong as follows: - when turning on (using I2C) the Vout ramps up and the FB follows. However, the controller continues to ramp up even though the FB pin rises above 0.6V. It actually rises all the way to 1.2V. And Vout rises accordingly to 1.8V instead of the 0.85V target. At this point the controller turns off because of OCL. Hiccup mode is disabled. Switching frequency = 1MHz. Soft-start time = 1ms. We have scope pictures, but I can't see any way to attach then to this post. - if we now turn the controller off and on again with I2C, then it doesn't go into OCL and seems to ramp up correctly. However, the FB stabilizes at 0.85V!! And Vout rises accordingly to 1.25V instead of the target of 0.85V. See attached picture TPS53819A-FBovershoot2nd. If we power off the main supply, the whole process can be repeated again. So it is repeatable and consistent.
Issue 3 - DRVL off / DRVH on with 0ns dead time ------------------------------------------------------------ The 3.3V 6A POL is the only POL design that shows no dead time from DRVL off to DRVH on. According to the datasheets this is typically 10ns, so I expected minimum to be more than 0. There is no overlapping, so no danger, but also no margins.