neutral loss

There are two methods recognized by the IEEE Green Book for grounding onsite generators. Diagrams can be found in one of the latter chapters. One or the other is generally used. Method 1: Do not open the neutral with the transfer switch. The neutral is hardwired through the switch. By this method the onsite generator is not a separately derived AC system and no additional grounding electrode is required at the generator. After a transfer is initiated the neutral connection to the service is rlied upon. Method 2: The transfer switch does break the neutral connection to the service and a grounding electrode is required at the generator. The generator is considered a separately derived AC system. Method 2 would eliminate the open neutral problem where the open is between the service and the utility transformer Which method are you using?

The loss of a neutral can be determined by measuring both "hot" to neutral voltages. When the neutral is lost, and the load becomes unbalanced, these two voltages will no longer be equal. Now, you have to allow some "slop" as they say. Even with a good neutral connection, the normal load unbalance will cause the voltages to be somewhat different (say 5 volts or so). With a loose neutral, you can see a difference of as much as 240V. Now, I am not sure there is a transfer switch this smart.

Another option is to measure both hot legs and transfer any time one of them goes "low". Again, is there a switch this smart? Don't know. Actually, I do know, but you don't want a $7k switch. In the "cheap" models, I don't know if this option exists.

Charles Perry P.E.

The plot thickens in that we also have 3 single phase Powerware9170 18 kva UPS's at this site. On the day we lost the power company neutral connection (verified) two of the UPS's did not sense the neotral loss as a power failure, one did. Of coures the batteries wewe quickly depleted because the generator never got a start signal from the transfer switch. Powerware says they will have a firmware upgrade to fix the UPS problem, now I need a fix for the transfer switch which is a 225 amp Kohler.

Here are two methods for detecting utility neutral loss. Each has their shortcomings:

1) measure the service neutral voltage imbalance. If the service drop neutral opens, the ground has a high impedance and the load has some imbalance, the neutral voltage will drift. 2) sum the service entrance currents (pass all conductors through a single CT window). If the service neutral opens, the load imbalance current will return to the utility system through the ground. In this case, if the ground impedance is very low, some imbalance current may return through the ground in parallel with a good neutral, so the sensing threshold must be set to account for this.

If you are concerned about a neutral break between the service equipment (where the system neutral is bonded to the grounding system), you must use 'Method 2' (below) and the above detection schemes are no longer applicable.

Is the Powerware capable of 240V input? I would think that a 240V UPS could care less if there is a neutral. I will have to investigate this when I return to the office. We have tested 9170s in ways Powerware could never imagine.

Charles Perry P.E.

The neutral loss sensing that the transfer switch has may only be sensing a low voltage on one hot leg. As Charles pointed out, a broken utility neutral can result in voltage imbalances which cause one hot leg to go high instead of low. If the UPSs and the transfer switch are connected to different hot legs, one may think that the utility power is OK while the other may think it has been lost.

Its more important to get both the transfer switch and the UPS sensing to agree on loss of power conditions.

I don't know whether anyone actually makes such a gadget but an open neutral could be detected by using two current transformers on the neutral between your panel bonding point and the utility.

One transformer would be connected to some signal source and would, when the neutral is "working" cause the test signal current to be injected in the "loop" formed by the neutral being connected to a "pretty good" ground at both ends. The other current transformer would detect this injected current. If the neutral opens, the "test signal" will go away.

Some real world experience would tell you how simple or complex the test signal and the signal detector would have to be.

This is essentially what your BASIC $8 GFCI does to detect FAULTS between neutral and ground.