Second part of the question. I noticed in table 3.7.3 of the on site guide that maximum cable length for spd connections is only defined for a type 1 SPD. Other parts of the osg and BS7671 extrapolate this to type 2 spd's as well. The cable length is to reduce ringing generated by large spikes. I was wondering if this limit was only intended for type 1 spd's dealing with very large spikes and has been incorrectly extended to any spd installation unnecessarily.

As I read BS 7671, 534.4.8 applies to all SPDs - not just type 1s. BS 7671 is definitive of course.

It's not so much "ringing" effects, but simple voltage drops along the conductors when the SPD 'shorts out' the conductors to suppress the surge.  Say for the sake of argument you had 10kV coming into the installation, and a perfect SPD that maintained 0V between its terminals when it triggered - you'd end up with 5kV each along the supply wires (so the SPD terminals would sit 5kV above true earth and 5kV below the peak of the surge). Say (again for the sake of arguments and simple maths) say the supply conductors were 10m long - 10m L and 10m PEN - you'd have half a kV per m along those cables. If the load you're protecting is 'tapped off' say 1m 'upstream' of the SPD on both conductors then it'll see an extra 0.5kV difference on each conductor - so be subject to 1kV p.d. even though the SPD was limiting the difference to zero. Longer connections to an SPD from a bus-bar or terminal block, is equivalent to making the connections upstream of the SPD. So as short as possible every time, total shouldn't really exceed 0.5m and in no case more than 1m. Straight rather than looped (certainly not coiled) helps with the inductive effects (such sharp pulses act very much like very high frequency a.c.).

    - Andy.

Second part of the question. I noticed in table 3.7.3 of the on site guide that maximum cable length for spd connections is only defined for a type 1 SPD. Other parts of the osg and BS7671 extrapolate this to type 2 spd's as well. The cable length is to reduce ringing generated by large spikes. I was wondering if this limit was only intended for type 1 spd's dealing with very large spikes and has been incorrectly extended to any spd installation unnecessarily.

As I read BS 7671, 534.4.8 applies to all SPDs - not just type 1s. BS 7671 is definitive of course.

It's not so much "ringing" effects, but simple voltage drops along the conductors when the SPD 'shorts out' the conductors to suppress the surge.  Say for the sake of argument you had 10kV coming into the installation, and a perfect SPD that maintained 0V between its terminals when it triggered - you'd end up with 5kV each along the supply wires (so the SPD terminals would sit 5kV above true earth and 5kV below the peak of the surge). Say (again for the sake of arguments and simple maths) say the supply conductors were 10m long - 10m L and 10m PEN - you'd have half a kV per m along those cables. If the load you're protecting is 'tapped off' say 1m 'upstream' of the SPD on both conductors then it'll see an extra 0.5kV difference on each conductor - so be subject to 1kV p.d. even though the SPD was limiting the difference to zero. Longer connections to an SPD from a bus-bar or terminal block, is equivalent to making the connections upstream of the SPD. So as short as possible every time, total shouldn't really exceed 0.5m and in no case more than 1m. Straight rather than looped (certainly not coiled) helps with the inductive effects (such sharp pulses act very much like very high frequency a.c.).

    - Andy.

Inductive effects help you when they are between the victim and trhe source, and hinder you when they aer between the SPD and the victim assume about 1microhenry per metre length for meter tail, less for cable constructions that mean youhave loops where flow and return currents are in close proximity - if the magnetic fields that each wire would produce in isolation overlap, they cancel, so there is less round-loop inductance.
So twisted pair is very low impedance and widely separated singles are the highest. Normal mains cables are in between.
Mike

On thing I have wondered is how effective a spd is in a consumer unit is at attenuating spikes generated in an installation when the CU is equipped with RCBO's. Presumably the RCBO coil has some inductance. I know a type 3 SPD is intended to deal with this, but how many are required and how necessary are they.

Would be great if the spd industry actually provided public documents showing actual spikes in a domestic environment and spd performance.