Solar PV question

An interesting problem has occured on some recently installed PV installations. On the most project there are four-240 volt single phase grid-tied inverters connected to a 240V/3P/4W Delta electrical system. L1 & L2 of inverter 1 (or I1) are connected across phases A-B respectively, I2 across C(L1)-A(L2), I3 across B(L1)-C(L2)and I4 across C(L2)-A(L1). The neutral is tapped on the utility transformer at the center of A-C. When connected like this, inverter 1 and only inverter 1 will not connect to the grid(this means any inverter in position 1 has the same failure), the failure shows up as a XFMR Disturbance, meaning a utility transformer problem. But, if the line leads for are reversed on I1, meaning phase A to L2, B to L1. The inverter starts right up every time. This same problem occurs only and on every 240V/3P/4W Delta project. I believe the problem is something inherent to the Delta system itself and not the cause of system loads, but I need some help explaining why this works. In other words, why would simply reversing the polarity on I1 result in stable operation for I1 in this setup?

Here's a guess (all voltages in RMS):

Are these inverters really designed to float above neutral (ground)? Its possible that they have some internal line to ground voltage sensing. Normally, a single inverter connected across a 120-240 single phase transformer (connected to the two hot lines) will see 120V to ground on both L1 and L2.

The way you are connecting these three units, Inverter 1 sees 120V to ground on L1 and 208V to ground on L2. Inverter 3 sees 120V to ground on L2 and 208V to ground on L1. If the (hypothetical) ground voltage sensing monitored L2 in each inverters and 208V exceeds its trip level, then Inverter 1 will trip unless you switch it (as you described). Moving Inv1-L2 to phase A would cure the problem, since now L2 sees 120V to ground. Contact the manufacturer and see if this is in fact the case.

There is a possibility that running the I1 and I3 inverters floating above ground may place additional stress on some components and shorten their life. The manufacturer will be able to tell you if floating them (as you are doing) is acceptable. If so, there should be nothing wrong with switching the I1 connections around as you have described......

...until they make a design change and connect the ground voltage sensing circuit to L1. :-|