And here we have the global perspective, on this newest monster in DK's stable of automation tools.
The concept is basically the same as the swimming pool skimmer, with the exception that I have not included a weir (I learned a new word, in this project - there is always something I learn...) in the design.
The chamber has a waterfall edge that is lower
than the tank water level. So of course, the water then falls into the chamber and fills it up, sucking the surface water to do so. The rate at which it sucks surface water is a function of the depth of laminar flow at the waterfall surface - that's the part that is sorta Bernoulli but not exactly. I chose the side surface of the chamber rather than the larger front surface for the waterfall edge because of three things:
- There was already a hole in the side of my chamber, where a bulkhead had gone. (least important reason)
- Using the end, with a smaller waterfall, would make the water flow faster for the same amount of pumping.
- I wanted to utilize a circular current on the tank surface that would feed surface water into my waterfall. By locating the waterfall on the side and having the powerhead shoot out the other side, it causes an overall circular current on the tank surface which continuously feeds new surface water and debris into the waterfall. Pulling the waterfall from the front edge would have pulled surface water from the center of the tank, with no real method to create a "feed" stream.
Because the waterfall edge is lower than the tank water surface, I figured I'd need some sort of weight to keep pulling the chamber down into the water, so I used the weight of the powerhead, which is why I mounted it to the bottom of the chamber. The weight, and pull, of the powerhead stabilizes the chamber's position in the water, if the chamber starts to empty too much.
The powerhead pulls water from the chamber from two places:
- From the general chamber water, fed by the waterfall
- From the auxillary feed port, the second bulkhead in the chamber bottom.
The trick is to encourage MORE pull from the waterfall feed than from the auxillary feed, and balance the flow so the pump keeps pulling what is falling over the edge. I placed a sponge prefilter over the auxillary port to add a scootch of resistance, making less resistance pulling from waterfall water than auxillary port water, at least until the net clogs with debris. It still pulls from both sources, but the ratio is favored toward the waterfall water so long as its resistance is slightly less than the resistance of the auxillary port.
If the pump is too weak, it still pulls, but the waterfall is slow because the chamber water level is nearly the same as the tank water level and doesn't suck surface water fast and therefore doesn't surface vacuum the duckweed very well. If the pump is too strong, it pulls not only the waterfall water but also the auxillary feed water, and if it's strong enough, it will start to pump more water out of the chamber than is being replaced by those two feeds into the chamber. The result of this is that the chamber water level lowers over time, until it reaches a point where the chamber starts to float up and the whole system stops working. So the most important balance issue is to not pull TOO much water from the chamber - that would tip the balance toward emptying the chamber.
As the water falls over the waterfall and into the chamber, it is pumped out the bottom of the chamber by the powerhead, continuously. The system will work continuously unless the flow balance is altered, for example if the catch net clogs with debris such that now most of the water is now pulled from the auxillary feed port.
Points of control can be added into the system to help balance adjustability, for example: a variable plate at the waterfall base to adjust the height of the waterfall edge - the deeper the waterfall layer, the slower the surface vacuum; a valve on the auxillary port to adjust the resistance across that port via the degree to which the valve is opened or partially closed. And of course, if one has a variable output pump, that is also a point of adjustability.
At any rate, here are the global pics, of the set-up.