Problem in processes involved with W boson in MadDipole

Dear MG experts,

I'm trying to use MadDipole (v4.5.1) to compute the subtracted terms for the photon radiation of some pure EW processes. Using the affiliated checking program to validate the generated terms, I have several questions:

1) For the (n+1) processes only involving neutral current, eg. u u~ -> z z a, the soft/collinear limit behaves perfect.
If request one identified photon, eg. u u~ -> z a a, the limits becomes just acceptable, the ratio between subtracted term and (n+1) tree doesn't approach 1 strictly. Is there any way I can approve this?

2) For the (n+1) processes involving W boson, eg. u d~ -> w+ z a, the ratio is far away from 1 in the soft/collinear limit.
If request one identified photon, eg. u d~ -> w+ a a, it even doesn't return a value.
If request multiple W bosons, eg. u u~ -> w+ w- a, it goes into the following error during compilation:

'intdipolesqed.f:370:54:
        CALL EPSFIQED( 0, 1, -1, 1,sikzone,mass_i,zi,esq,e)
                                                                                             1
Error: Symbol ‘zi’ at (1) has no IMPLICIT type'.

If let W boson decay instead, eg. u d~ -> e+ ve z a, the limits behave correctly again, so it seems there exists some problems when the W boson couples to the photon. Is there any way I can correct these? Or do I miss something which should have made it work?

Many thanks,
Huanfeng