I'm working on a small USB powered product. The situation is that the product uses a captive 6' USB cable with 28awg power conductors. Unfortunately, the design was originally implemented and the mechanical design executed without a complete analysis of the product power requirements. The resulting mechanical design has restricted the choices of cable diameter and PCB connector such that it will be very difficult to go to a larger wire gauge without major rework.
When the product is operating, it draws right at 500ma max in the USB cable, as measured on the bench. About 100ma of that is going to some LEDs that are not powered by the SMPS. Dynamic current changes through the SMPS are about 100ma, as various loads are enabled and disabled so the SMPS is supplying 300ma to 400ma. The frequency of current change is about 16ms with roughly a 50% duty cycle.
USB minimum voltage at the supply at the USB A connector is 4.75v per spec
Voltage drop in the cable is about 10% of input voltage, reducing the voltage at the input of the regulator to around 4.25vdc worst case. Calculated cable impedance is 0.87ohms.
I ran a Webench simulation using the following parameters:
supply voltage: 4.75
supply resistance: 0.87 ohms
output voltage: 3.3v
output current: 500ma
after some experimentation, it appears that an input cap of 47uf, an inductor value of 3.3uh, and an output cap of 10uf, should work well.
I experimented with reducing the supply voltage down to 4.4vdc with the input voltage to the inductor remaining high enough that the SMPS can regulate to 3.3v steady state.
I also tried some transient behavior experiments, first switching the load from 300ma to 500ma, and then 500ma to 750ma with the supply voltage at 4.5vdc to check for bad output behavior, but find that the regulator seems to be able to maintain regulation with these transients, even with the high cable impedance.
Does Webench use the worst case process and temperature corner for the SMPS component during simulation?
Does it also use the worst case combination of component tolerances?
Is this design going to be robust over temperature and process, or should I bite the bullet and increase the wire gauge? I'm concerned about dynamic response with the high cable impedance.