Autotransformer Adventures

[Flimflam]

Hey Guys!  Thanks for letting me post my question here.

I've got a 240V inverter installed in my RV with an autotransformer ("AT") as the sole load on its AC output side.  The AT is center-tapped and thus produces split-phase 120V to supply the house loads. 

The inverter is connected to a DC battery that is fully isolated from everything else in the circuit.

The inverter also accepts 240VAC single-phase input, plus the EGC/safety ground.  I am feeding it the two hots (L1 and L2) from my standard 240V, 50A residential split-phase outlet, plus the green EGC.  The white neutral conductor coming from the grid is capped off and unused.

Both the inverter and the AT have chassis ground terminals which I have bonded to the steel frame rail of the RV.  These terminals are also connected inside each of the inverter and AT to the input and output EGC posts.

When the inverter senses power on its AC input, it passes it through to the output side (and stops inverting off the battery).

The AT's output side includes L1, L2, and the newly-derived neutral.  The AT includes a relay that can be engaged to bond the derived neutral to the EGC/ground.  This relay is turned on whenever the inverter is not connected to the grid.  (Then, the RV's frame rail becomes the safety ground.  The whole RV is floating with respect to earth ground when running on battery.)

The AT output leads (L1, L2, N, EGC) are wired to a traditional distribution panel from which all the loads are connected in the traditional fashion.

Problem:

When the grid is attached to the AC input described above, everything "works," but my clamp ammeter measures some current (7 or 8A) flowing out of the inverter and back out the EGC to the grid, even when there is almost no load on the system (<100W).  The reverse current does seem to increase slightly when a larger 120V load is placed on the system.

My inverter's computer monitor gets confused by this reverse current and adds a net -1kW to the grid demand readout at all times.  In other words, if the real load is 7kW, it reports 6kW.  If the real load is 200W, it reports -800W.  Roughly speaking.

I am 99% certain that this current is at a very low potential, so there is very little power here. 

Additional info:

1.  My inverter manufacturer says that I should simply disconnect the EGC from the shore.  To their credit, if I do this, the computer reverts to perfectly accurate power readings in every direction.  I am aware of the safety consequences of not having the EGC from shore power in place, and I believe that I can eliminate them by placing an industrial-sized RCD at the shore connection point.

2.  If I unplug from the grid, the inverter switches to battery.  In this mode, everything works flawlessly as well, and there is no weird offset.


I am trying to understand why this reverse current is developing on the EGC in the first place, first and foremost so that I can convince myself that I understand this circuit.  I'm also open to suggestions about how to eliminate the need to leave the shore EGC open and thus, not need a big GFI.

I can attach a small, crude diagram if that would be helpful.

Many thanks for reading!

[Electro Detective]

Pictures, video and diagrams all help   

Welcome to EEVblog btw   

[Flimflam]

Here's a crude block diagram showing the basic connectivity:

[Jwillis]

Is it possible that something else is leaking current creating a ground loop.

[Flimflam]

Is it possible that something else is leaking current creating a ground loop.

Maybe.  But, it seems even if I remove all of the external loads, I get (some of) the spurious return flow. 

[BNElecEng]

Can you perform a quick test by removing the Live and Neutral connections between the inverter and the autotransformer? This would take the transformer and the loads out of the equation. I'm curious to know if the leakage current still flows under this condition.

I've had issues in the past with inverters that had capacitors between their neutral and earth paths as part of an EMI filter. Since you are using L1 as the input for the inverter Neutral, the potential difference between the Neutral (now L1) and Earth connections could cause leakage current to flow.

[Flimflam]

Can you perform a quick test by removing the Live and Neutral connections between the inverter and the autotransformer? This would take the transformer and the loads out of the equation. I'm curious to know if the leakage current still flows under this condition.

The leakage current does NOT flow under something similar to this condition, however I only know that because I tested the inverter on the bench in my garage first and did not see the backflow.  In that setting, however, I also did not have chassis ground connected to anything.  I will repeat this test in the RV, with everything grounded to the frame, and see what we see -- good idea!

I've had issues in the past with inverters that had capacitors between their neutral and earth paths as part of an EMI filter. Since you are using L1 as the input for the inverter Neutral, the potential difference between the Neutral (now L1) and Earth connections could cause leakage current to flow.

It's an interesting idea.  One thing to note is that this inverter was designed and built in the EU, and my understanding is that in many countries over there, the N and PE are usually referenced/equipotential.  If that understanding is true, there would be almost no potential difference in their most-common deployment.