We propose an extension to other high-ir, compounds of a model introdu
ced earlier for YBa2Cu3O7. In the ''self-doped'' compounds we assume t
hat the doping part [namely the BiO, HgO, and TlO planes in Bi2Sr2CaCu
2O8+delta (BSCCO), HgBa2Can-1CunO2n+2 (HBCCO), and Tl2Ba2Can-1CunO4+2n
(TBCCO), respectively] is metallic, which leads to a multiband model.
This assumption is supported by band structure calculations. Taking a
repulsive pairing interaction between these doping bands and the CuO2
bands leads to opposite signs for the order parameter on these bands
and to nodes whenever the Fermi surfaces of these bands cross. We show
that in BSCCO the low-temperature dependence of the penetration depth
is reasonably ac counted for. In this case the nodes are not located
near the 45 degrees direction, which makes the experimental determinat
ion of the node locations an important test for our model. The situati
on in HBCCO and TBCCO is rather analogous to BSCCO. We consider the in
dications given by NMR and Rnd that they rather favor a metallic chara
cter for the doping bands. Finally we discuss the cases of Nd2-xCexCuO
4 and La2-xSrxCuO4 which are not ''self-doped'' and where our model do
es not give nodes.