Validating Power Select - LTC4417

This page will outline my validation and results and troubleshooting for the power select board.

• The first test is kinda a smoke test → I will have 5V initially on the Aux input from a power supply, and the eload connected to the output in CC mode currently drawing 500mA (to start off) → I will then increase the input voltage to see when the output is connected

• I may have figured out a logical flaw → for the LTC4417 to be enabled, there must be a 3V3 output from the controller board. However a standalone controller board needs 12V input to have the 3V3 reference, except that 12V in the schematics seems to come from the output of power select, which only has a 12V reference when the LTC is enabled. Therefore, it should never enable standalone

• I connected the controllerboard to my laptop for the 3V3 reference, and started the power supply at 5v, with the eload showing that it has around 70mV. I increased it in 1V increments until it reached 10V output, in which the eload suddenly changed to 9.998Voc, meaning the LTC was regulating the supply! Note that at this point, I took out programmer from my laptop, and I still see the 3V3 rail which further verifies my assumption is probably right. see the following table for some points

• important note from the above table, no matter the valid input reference level, the output would only be enabled if the eload was set at less than 1.1A. not sure why → will ask josh

• before i left for the day, I tried to see if the priority system would work → for some reason the supply on the DCDC input did not ever seem to allow output no matter the condition. I switched the supply as well and it still wouldnt allow output.

• turns out for the 1.1A limit i forgot that i set the power supply limit to 1A LOL so yeah of course when I increased that, the output was enabled for higher currents

• back to the enabling input problem, I notice that the valid1 pin is always low and the other valid pins are always high no matter the configuration of the input supplies → a low valid pin means the presence of a supply should be detected and allowed to turn on output in the case of priority

  • turns out there is a short to ground on the VALID1 pin, so im gonna check all the solder

• after probing around the board and fixing a few solder joints, it seemed that the entire current sense resistor for DCDC input was not soldered down. One thing to note for next rev is that the pads for the resistor should be bigger so its easier to solder down, as the resistor pads almost perfectly fit the footprint pads.

• finally the input from dcdc is going to the LTC4417, however the output is still not being enabled (there is only input on the dcdc connector)

• I noticed in this case the VALID2 pin is low and the rest are high, meaning that DCDC is enabled, however there is still no ouput.

• I then noticed when i connect DCDC and aux, the output is still 0, but when i connect only aux, the output is non zero, which means that the ltc is prioritizing correct and therefore the issue is turning on the fet. I noticed that with Aux input, TP1= 10.6V, TP2=0.4V, TP3=7.29V and with DCDC connected, TP1= 8.2V, TP2=0.0V, TP3=0V. will match them to their gates to see what is wrong next time.

• Next day of troubleshooting and im resoldering all the fets to see if thats the problem, because aside from the power supply mux fets seemingly turning on for only aux input, there doesnt seem to be an LTC problem from the above numbers. I am finding that soldering down the fets is incredibly difficult because of how large the associated poly pours and copper is. For next rev, we might want to minimize the poly pours by recording how much they heat up from this testing and adjusting accordingly.

• YES it finally works turns out it was a fet soldering problem, look at the above^

• i tested priority between dcdc and aux, and dcdc always takes priority which is good. Something else that josh wanted to test was the switchover time from output turning on to input being supplied. I used the scope to plot both input and output and compare timing. See below for a difference of about 300ms from aux on to output

Open

Open

below is the switchover time from DCDC to output. The small difference from aux to DCDC seems negligible.

• weird update, all of a sudden, my controller board stopped blinking its led (meaning something was wrong) and started becoming extremely hot when plugged into my laptop. I probed across 3V3 and gnd and see that now its shorted on the controller board. something must have happend during my testing to somehow short 3V3 to gnd with my controller attached, thus damaging the CB and most likely the MCU. In addition, for some reason the LTC seems to not work properly because it is able to turn on the output without a 3V3 reference, which should not happen. After waiting a couple of minutes and trying again, it seems that this is not the case, and the bulk capacitor was probably the reason I was still seeing output on the LTC as it was holding the 12V for a bit longer.

• More testing on switchover (with a new controller board that I haven't destroyed of course), I will check if the supply will switchover in the cases outlined in josh’s test plan → that is if the primary source goes OV or UV, the ltc will switch to the other supply. the following table are my results

• from the above table, the values for UV seem a bit off, otherwise though it was successful.

• adding on to the switchover times recorded, another important one josh noted is the time from swithching from aux to DCDC, and the time in which there is no 12V reference on the output. Looking at the scope shot below, there is not really a moment where the 12V output is stopped, which means that the caps are large enough to keep it going through the switchover time (purple is output voltage, blue is aux voltage and yellow is DCDC voltage with the rise being when i turned DCDC on and it took priority over aux)

• Note, that all the tests done today were done with a 1A load. DCDC would only ever supply 17A max, and even with out power budget we dont expect to fly higher than about half that. To confirm the board is able to handle high loads, I will be running the aux supply at 10A for about 10min to see if there is any weird heating or other faults. Note: I have placed a 20A fuse on the aux input so it should be sufficient for this testing. Also note that the voltage form the supply is 13.5V. Also note that while doing this, I found that loading 10A has the output voltage at 12.7V! Thats a huge drop from 13.5V, 800mV to be exact. I probed the voltage difference from the input annode to output annode, and only notices 350mV drop, so the rest ~450mV has to come from the wires I’m using (18 and 16 AWG).

• After about 10min, i used a temp probe to measure some components facing that high current. The following table summarizes that:

• It seems that there are no huge temp or voltage drop problems board side, so it seems good!

• The last thing i wanted to check was how long before the output discharges with the cap. the output was captured on this scope