elcb and borehole pump

There have been reports of RCDs tripping on large startup currents (often largish motors or welders). The problem seems to be due to RCDs only being accurate within a certain range of currents - as I understand it the effect is down to things like manufacturing tolerances so that the L and N coils don't precisely cancel even when exactly the same current is flowing in and out. Say the two current carrying coils on the torroid are only accurate within 0.01% of each other, so with say 63A flowing the residual current might be out by 6.3mA but as that would still be below the RCDs tripping threshold it still wouldn't trip. But if the load current was say four times that, even if only for a short time, the error might climb above 25mA and be enough to trip the RCD.


If would imagine that a borehole pump might indeed  have a difficult time starting, if it needs to get a 20m column of water moving before things start turning normally (I've no idea if they incorporate any kind of soft starter these days).


Because the difference between tripping and holding in might be so slight, simply replacing one RCD with another apparently identical one might make the difference between it tripping or not (or similarly replacing the pump), but if the  RCD has a relatively low current carrying rating (25A, 40A, 63A say) replacing it with one with a higher rating (80A or 100A) is perhaps more likely to help. Similarly if an instantaneous 30mA isn't actually required (if say it's hard wired rather than plugged into a socket and cables aren't concealed in walls etc) than a higher residual rating and/or time delay version would reduce the risk of nuisance tripping - or perhaps no RCD at all if it's a TN installation and Zs is within tolerance.


Or of course it might be a simple N-PE fault anywhere downstream of the RCD (not necessarily on the pump circuit) that's leaking an extra bit of current when the pump startup current increases voltage drop in the supply and N drifts a little further away from Earth than normal.


   - Andy.

There have been reports of RCDs tripping on large startup currents (often largish motors or welders). The problem seems to be due to RCDs only being accurate within a certain range of currents - as I understand it the effect is down to things like manufacturing tolerances so that the L and N coils don't precisely cancel even when exactly the same current is flowing in and out. Say the two current carrying coils on the torroid are only accurate within 0.01% of each other, so with say 63A flowing the residual current might be out by 6.3mA but as that would still be below the RCDs tripping threshold it still wouldn't trip. But if the load current was say four times that, even if only for a short time, the error might climb above 25mA and be enough to trip the RCD.


If would imagine that a borehole pump might indeed  have a difficult time starting, if it needs to get a 20m column of water moving before things start turning normally (I've no idea if they incorporate any kind of soft starter these days).


Because the difference between tripping and holding in might be so slight, simply replacing one RCD with another apparently identical one might make the difference between it tripping or not (or similarly replacing the pump), but if the  RCD has a relatively low current carrying rating (25A, 40A, 63A say) replacing it with one with a higher rating (80A or 100A) is perhaps more likely to help. Similarly if an instantaneous 30mA isn't actually required (if say it's hard wired rather than plugged into a socket and cables aren't concealed in walls etc) than a higher residual rating and/or time delay version would reduce the risk of nuisance tripping - or perhaps no RCD at all if it's a TN installation and Zs is within tolerance.


Or of course it might be a simple N-PE fault anywhere downstream of the RCD (not necessarily on the pump circuit) that's leaking an extra bit of current when the pump startup current increases voltage drop in the supply and N drifts a little further away from Earth than normal.


   - Andy.