I've been comparing the TPS2113 and TPS2115 datasheets trying to figure out how they differ, and after a long time I finally figured out that the reason I was having difficulty understanding the difference is because there is a mistake in the TPS2115 datasheet.
As stated in the EVM datasheet, there are two modes which these controllers could operate. The TPS2113 supports both choosing the greater of IN1 and IN2, or makes IN1 the default UNLESS it's voltage drops too far, in which case it switches to IN2. The TPS2115 however does NOT support this second mode.
Yet the datasheet for the TPS2115 states on page 16 that:
(I would link it here, but your forum appears to be broken and inlines a blank image when I try to paste the link or add the link to selected text.)
"Some applications have two energy sources, one of which should be used in preference to another. Figure 14
shows a circuit that will connect IN1 to OUT until the voltage at IN1 falls below a user-specified value. Once the
voltage on IN1 falls below this value, the TPS2114A/5A will select the higher of the two supplies. This usually
means that the TPS2114A/5A will swap to IN2."
This is clearly incorrect. You could argue that setting the device to check if IN > IN2 is "a user specified" threshold, but that's simply misleading and this description is extremely confusing when it could be very simple. Instead, the text appears to be copied verbatim from the TPS2113 datasheet.
Also your EVM module datasheet has a mistake where it states:
2.4.1 Mode 1 Automatic Switching
OUT connects to the higher of IN1 and IN1 in Mode 1 automatic switching.
Refer to Table 2-1 to configure the EVM for Mode 1 automatic switching.
Obviously this should say the higher of IN1 and IN2, not IN1 and IN1.
Also, while I have your attention, I intend to use the TPS2115 in an implementation of USB OTG, where I will need to short IN1 and IN2 when the device is intended to be used in host mode. Do you see any issues with doing this, and do you have any suggestions as to how best accomplish this?
The enable logic needs to be low and compatible with 3.3V logic, so I was considering just sticking a couple P-FETs back to back. But if my calculations are correct there are only a small range of Vth which will work as desired with this range of voltages and 3.3V at the gate. Perhaps you could offer a better solution or a chip that would perform as desired without additional components? This is a space and price constrained design, so I'd like to avoid adding buffers or additional diodes if necessary.
