TL;DR: The precharge sense resistors need to be 2kohms or less total for it to work with our motor controllers. If both motor controllers are in parallel, and we only use one motor controller interface (which is the plan) then I would even suggest that we target less than 1kohm of precharge resistance instead. Note: James said we plan to use 300 ohms for the next revision anyways, so we should be good.
Note that I am defining EPR as Equivalent Parallel Resistance even though this is not a standard term but is the short way of me saying the impedance that results in leakage current.
Precharge had been verified to work with a film capacitor pack with equivalent capacitance of about 520uF (4 paralleled 130uF film caps), but had run into issues executing when connected to the motor controllers. What would happen is that the precharge circuit would not conclude with the latch out closing the relay and ending the precharge sequence. It however, worked flawlessly with the capacitor pack, so some debug was done to see what was going on.
...
Basically, the biggest difference between our cap pack and the motor controller caps is that there is extremely low impedance on those caps in the motor controllers. At first it didn’t really make sense how this might impact the precharge state, but I then outlined it in a few diagrams. Since the motor controllers use about 48 paralleled 4.7uF (after further discussion, this is actually 48 paralleled 10uF caps in series with another 48 parlleled 10uF caps) ceramic caps (see Diagram 2), they have super low ESR(EPR) and a capacitance of about 230uF, compared to the film capacitor pack I was testing with. The connection of the precharge circuit and the bulk capacitance is shown in Diagram 1 part A. I then found that if we have the ESR(EPR) of the cap at a certain value (20kohm for example) we would get a certain voltage that the caps could charge to (see Part B). If we decrease that ESR(EPR) of the bulk caps because of their extremely low impedance, we find that using 10kohm for example decreases the max voltage that the caps can charge to (y variable in the diagram) - See part C. So to counteract this low ESR(EPR), we can decrease the value of the precharge resistors - such as make them to 1kohm instead of 3kohm as done in Part D - and find that the max voltage the capacitors can charge to is higher to compensate. The reason this is important is because the precharge circuit will measure the in+ node (the 50V node in the diagram) with the in- node (the y node in the diagram) to make sure that precharge is within a range of being over. This resistor divider being formed with the bulk capacitance ESR(EPR) to ground is probably the reason for failures thus far, and therefore we must decrease precharge resistance to 2kohms when testing with one motor controller. When we parallel two motor controllers on the output of precharge, I would suggest limiting precharge resistance to 1kohm to compensate (due to the paralleled esr ESR(EPR) which should be half the resistance total).
...
Diagram 1: Difference in low ESR(EPR) and precharge resistance (note all esr ESR(EPR) resistance values are not accurate to actual values - just for visualization of calculations)
...
Diagram 2. Bulk Caps on WS22 (48 paralleled 4.7uF caps10uF caps) - Measured 250uF with LCR meter on HV terminals, resulting in the cap config approx. (https://product.tdk.com/en/search/capacitor/ceramic/mlcc/info?part_no=CKG57NX7R2A106M500JH ?)