Binary phase diagram using maps and phb

Dear Axel,
To plot a binary phase diagram, maps suggested 6 different poscars where the number of atoms was between 2 and 4. Two other suggestions with pure metal contained one atom each.
I used vasp with the tags suggested by the manual to find the energies of these structures.

This is the eci’s in the eci.out file:
0.898037
-0.022933
-0.116050
-0.136303

and this is the fit.ot file:
0.000000 0.000000 0.047861 -0.047861 1.000000 0
1.000000 0.000000 0.001994 -0.001994 1.000000 1
0.500000 0.835250 1.034340 -0.199090 1.000000 2
0.500000 0.823000 0.953329 -0.130329 1.000000 3
0.333333 1.074667 0.833244 0.241423 1.000000 6
0.333333 1.078233 1.041984 0.036249 1.000000 8
0.666667 1.166767 1.026695 0.140072 1.000000 9
0.500000 1.268475 1.306945 -0.038470 1.000000 27

I followed the manual (e.g. pages 27-28).
Running phb:
$PHB/phb -gs1=0 -gs2=1 -dT=100 -dx=5e-3 -er=50 -k=8.617e-5 -ltep=5e-3 -o=ph01.out

where phb1.out is:
0 -0.022933 -1.000000 1.000000 0.024928 0.024928
100 -0.022933 -1 1 0.024928 0.024928
200 -0.022933 -1 1 0.024928 0.024928
300 -0.022933 -1 1 0.024928 0.024928
400 -0.022933 -1 1 0.024928 0.024928

2600 -0.022933 -1 1 0.0249286 0.0249286
2700 -0.022933 -0.999999 0.999999 0.0249291 0.0249291
2800 -0.022933 -0.999999 0.999999 0.0249298 0.0249298
2900 -0.022933 -0.999998 0.999998 0.024931 0.024931
3000 -0.022933 -0.999997 0.999997 0.0249327 0.0249327

16400 -0.0229813 -0.473506 0.469403 0.602691 0.604487
16500 -0.0229959 -0.434665 0.428922 0.621943 0.624468
16600 -0.0230166 -0.406457 0.396007 0.635856 0.639372
16700 -0.0230474 -0.35812 0.353528 0.654747 0.656645
16800 -0.0230581 -0.285269 0.240724 0.676366 0.686213

next
$PHB/phb -T=16800 -mu=-0.0230581 -gs1=0 -gs2=-1 -dT=100 -dx=5e-3 -er=50 -k=8.617e-5 -o=ph0d.out
and ph01.out:
0 -0.022933 -1.000000 1.000000 0.024928 0.024928
100 -0.022933 -1 1 0.024928 0.024928
200 -0.022933 -1 1 0.024928 0.024928
300 -0.022933 -1 1 0.024928 0.024928
400 -0.022933 -1 1 0.024928 0.024928
500 -0.022933 -1 1 0.024928 0.024928

16200 -0.0229772 -0.526674 0.525605 0.571219 0.571553
16300 -0.0229737 -0.504146 0.50246 0.585175 0.586116
16400 -0.0229813 -0.473506 0.469403 0.602691 0.604487
16500 -0.0229959 -0.434665 0.428922 0.621943 0.624468
16600 -0.0230166 -0.406457 0.396007 0.635856 0.639372
16700 -0.0230474 -0.35812 0.353528 0.654747 0.656645
16800 -0.0230581 -0.285269 0.240724 0.676366 0.686213

then
$PHB/phb -T=16800 -mu=-0.0230581 -gs1=-1 -gs2=1 -dT=100 -dx=5e-3 -er=50 -k=8.617e-5 -o=phd1.out
and phd1.out:
16800 -0.0230581 0.256646 0.297374 0.681858 0.673401
16900 -0.0218221 0.274415 0.266868 0.683052 0.684717

In the meantime this was output too:
Phase 1 size: 39 39 39
Phase 2 size: 39 39 39
Phase 1 n_equil= 0 n_avg= 7000
Phase 2 n_equil= 0 n_avg= 2000
16800 -0.0230581 0.256646 0.297374 0.681858 0.673401
Phase 1 n_equil= 1500 n_avg= 4000
Phase 2 n_equil= 2500 n_avg= 1000
16900 -0.0218221 0.274415 0.266868 0.683052 0.684717

I appreciate if you let me know which step I did wrong.
Thanks

The cluster expansion is not very well converged (too few structures and large fitting errors).
But otherwise, this does not look terribly wrong.
What is the problem?

BTW, the clusters.out and gs.out would be useful to check/see.