UPS Backfeed detection and prevention

[Noway2]

Here is a question that is a varient on a common topic: backfeed protection. In a large, non pluggable, UPS, you are required to detect and stop any backfeed at the input voltage within a one second time frame.

My question is, how does one detect the presence of backfeed?

Measuring voltage alone is insufficient as the backfeed voltage can look like normal source voltage. I am thinking that one would have to also measure the currents, assuming that there are any - and they would likely be low and due to leakage, power supplies, etc, to calculate the instaneous power flow. I say that the current would be low to nil as the input breaker would be open. The problem is that the input terminals can still be electrically live and lethal which is what you want to avoid.

I have seens some representative schematics for backfeed prevention relays and it looks like these relays monitor the current on one phase, the voltage across the other two phases (adding a 30 deg phase shift). The two signals then get multiplied and low pass filtered with the output being the detected signal.

Multiplying the voltage and current gives the power which will be 120Hz sinusois, assuming a 60Hz application. I am guessing that the LPF is to remove the 120Hz and the detector then looks for a DC offset. But this is purely a guess. I am also not clear on the exact purpose of adding a 30 deg phase shift in the calculation and how this affects the detection.

Would anyone be so kind as to help me to figure out a practical means to detect the backfeed and / or to help me better understand how the circuit described above would do so?

 

[sibeen]

A backfeed protection circuit can be done fairly easily by placing a capacitor and a current transformer in the bypass (reserve)line - refer to drawing.

When bypass voltage is available, normal situation, current flows into the capacitor but is not detected by the current transformer.

When the power fails there should be no voltage on the rteserve line and therefore no current flowing into the capacitor.

If there is a problem and the static switch lets voltage from the inverter flow back onto the reserve line then a current will flow into the capacitor and be detected by the current transformer.

 

[MikiBg]

Noway2, can u attach schemes that u have seen?

 

[Noway2]

Here is the circuit I was referencing.

This is from some documentation on ABB's web site.

Sibeen, I think that this approach is ingenious. It addresses one aspect of the backfeed, from the inverter to the bypass input. What about the inverter / DC bus back to the mains input via the rectifier? According to UL, you can't rely on an SCR as a (backfeed) block because a shorted condition is considered normal in their evaluation. Is this something that is normally not worried about? (Note, my own research has shown a lot of emphasis on the inverter backfeeding to the bypass, but not the mains).

 

[MikiBg]

Just thinking in public, never met Noway2's circuit in practice:
Without 30 deg phase shift there wouldn't be DC component on multiplier output, since phi=0 is assumed (Ua and Ia are in phase, pure resistant load). With this shift, there will be DC component proportional to active power and 120(100)Hz component that LPF should kill. Level detector probably should recognize change of DC component direction caused by change of current direction. Because of low voltage detector and pure resistant load assumed (such as short circuit) and the fact that in case of short circuit between CB and GEN there will be a fault current component from active grid with different direction comparing to one when GEN is supplying transformer, the whole thing looks like reverse power (motor regime of generator) and/or short circuit protection to me. Somebody will sure know better.
However, it has nothing to do with backfeed protection of UPS itself. The interesting thing is that sibeen's circuit needs exactly current flow direction detection to become BFP triping circuit. When UPS is in electronic bypass mode and STS is supplying the load (not desirable, but that is the role of STS), current will flow through CT but in other direction than when BF occurs.
As far as I know, basic backfeed protection should disconnect mains2 (bypass) input from inverter output in case of STS fault. Don't know how backfeed contactors (optional in UPS units I worked with) are actuated, maybe "watch dog" of STS can do the job.

 

[MikiBg]

Correction:
Resistive nature of impedance at the point of fault and of the electric arc doesn't explain phi=0 assumed. At least not to me