We overclockers often use a DX evaporator on the CPU to chill the chip (example: my own DX evaporator to be placed on the CPU).

Besides getting the evaporator to do it's job, providing good expansion is still a major headache. The reasons why providing good expansion is so difficult:

- Huge load variations. The heat output of a CPU depends greatly on the amount of work it is doing.

- Low output power. I usually calculate with 150W maximum, and this includes heat ingress from the surrounding. Minimum power can be as low as 30W, including heat ingress.

The two most common expansion devices used and their disadvantages for us:

The TEV
These valves have two major disadvantages: their nominal capacity is too large for us, even the capacity of a 1/8 ton valve. The second disadvantage is reaction speed. When the TEV decides that more refrigerant is needed, the evaporator is already starving. When the TEV decides less refrigerant is needed, the suction line is already filled up with liquid.
This is because the heat capacity of most of our evaporators is insufficient to maintain a stable regulation loop. Temp just rises/drops too fast.
Also, TEV superheat setting seem to skyrocket below -40C/-40F.

The capillary tube / fixed orifice / manual expansion valve
These can be adjusted for our loads, and since they do not sense anything, there is no control problem. But fixed restrictions are hard to finetune, and their ability to cope with load variations is mediocre at best.

Now, I see one solution for our problems: the EEV, and then especially the stepper motor operated EEV such as the Sporlan SEI-.5 (1/2 ton EEV stepper type, 1596 nominal steps). This valve has a nominal capacity of 1.7kW / 0.5 tons (R22 @ 8.5 bar/115psi PD, SST = -7C/20F).

We need only about 8% of this capacity. Will such a valve with custom electronics be able to control these light loads? This would leave about 127 useable steps. With R404a even less.