You got it a bit wrong.
This program (a sub-module in Siesta) self-consistently solves the NEGF equations to calculate the system at a finite bias. Upon self-consistency, the Siesta program will write out the Hamiltonian under the effect of the bias as given in the input file.
This effectively means that for each bias-point one needs to self-consistently solve the NEGF equations to obtain the true electronic structure for a given bias point.
Since the electrodes are in equilibrium (even under finite bias) one can use the same electrode Hamiltonian for all bias points, by definition!
This program can calculate elastic transport by inputting the electronic structure of the electrodes (still under equilibrium) and the device region for a given bias point. If the electronic structure for the device region is solved for a given bias $V$, then TBtrans can only correctly calculate the transport for that same bias $V$, if you give TBtrans any $V'\neq V$ your calculation will be wrong.
What TBtrans can provide is a way to interpolate the electronic structure of several bias-points (several TSHS files calculated with TranSiesta for different bias points) and thus give an estimate of the true transport characteristics for a bias point where the electronic structure has not been explicitly calculated. This interpolation should of course be done with care and before-hand knowledge of the electronic structure should give you details about adequate interpolation ranges. Say for one system you can interpolate in a range of 0.5 V, whereas for another system, you can only interpolate in a range of 0.25 V.
TBtrans does not create the
- Calculate the electrode electronic structure
- Calculate the 0 bias case using TranSiesta
- Calculate any non-bias cases using TranSiesta and by using the closests bias point previously calculated using TranSiesta (copy that TSDE file). Be sure to set the flag:
DM.UseSaveDM T in your input.
- Finally, calculate the transmission for the bias points done in steps 2 and 3