Hot carrier solar cells promise energy efficiencies up to 85%. However, the structure realizing such a high efficiency requires perfect and thermodynamically reversible carrier extraction only at a single energy in each of two bands of the absorber. This approach tends to increase the time required for carrier extraction and results in a strong parasitic carrier cooling domination within the absorber. As a practically feasible alternative, the simple and fast high-pass energy filtering (e. g. across band offsets in a double heterojunction) has been proposed, resulting in some additional entropy generation and in some loss of the limiting efficiency. It was not clear though, whether the remaining efficiency is superior to a conventional cold carrier single bandgap p-n junction with a band gap equal the sum of that in the absorber plus the two band offsets at the interfaces. Further, it was not clear whether the I-V characteristic is expected to be linear like that of a thermoelectric converter, or curved like that of a conventional diode. Finally, it would be interesting to know how much efficiency can be expected in this alternative concept with a realistic carrier cooling time in the absorber and which figures of merit additionally to the cooling time may improve the efficiency. The theory should remain preferably simple, to facilitate engineering calculations and cell design.