> Briefly, take a long pipe and block it at about the midpoint with a wall.
> Make a small central hole in the wall. Introduce air tangentially into the
> pipe next to the mid point wall. Throttle the air flow out of the side that
> the air is introduced into so that SOME of the air will tend to flow through
> the hole and out the other open pipe end. (Actually slightly more complex
> than this in practice but not very much so).
>
> The air coming out the throttled end is noticeably hotter (up to around 100
> kelvins hotter). The air coming out the end which is fed via the central
> hole is noticeably cooler - up to about 50 kelvins typically depending on
> hot/cold split. Several somewhat incompatible theories explain this device
> rather well.
>
> Unlike Graham's device above, this one does need a net energy input. And,
> no, it cant be used regeneratively to give successively lower temperatures
> (unfortunately).
>
> RM

There are comercial & industrial products available based on that
system. I use to receive literature from them. A small cilinder (5
inches) and a compressed air connector, can refrigerate machines,
enclosures, cabinets, and so on.

Wagner