To detect speakers are playable

No. The price of a microcontroller running the .NET framework will always be greater than the price for a microcontroller with equivalent performance (or equivalent to the price of a micro with greater performance) which is programmed in assembly, C, or even using the Arduino tools.

When buying a microcontroller for the .NET framework, you pay for small transistors, you pay for more transistors, you pay for complexity, you probably pay for licensing a core developed somewhere else, and, since you're paying for all that, most manufacturers imagine that you'll want to pay for higher pin-count packages since your die budget supports them. If you did this in assembly or C, you could get equivalent performance and pay a lot less.

See also the following questions:

• Arduino / Processing versus .NET Micro Framework? Best language? Best hardware?
• Microcontroller with Java API
• What are the .NET Micro Framework ready systems available?
• Is C++ suitable for embedded systems?

The first question in the list is very applicable to this question, it just discusses the suitability of the .NET micro framework as opposed to the pricing and programming which we're discussing here.

If you still want to use .NET, you think you can get acceptable performance out and you're going to be buying a significant number of parts (at least 1k, probably easier at 10k), you can do the same thing that the vendors of the FEZ parts do: Have it loaded at the factory. It costs somewhere in the region of a quarter apiece to program large files like the one you'll be working with.

To make the fill control for the water tank you should be able to use the same microcontroller as the one shown in the example. In that case an Atmel AVR was used.

The circuit would require the addition of an output from the AVR that would go to a buffer circuit that would in turn be used to turn an electronically controlled water valve on and off.

Then the software if the AVR would have to be modified to control the fill valve ON/OFF state based upon the level sensor states. The simplest scheme would simply turn the valve ON when the level 9 was not active and then when the water filled enough so that once the level 9 changed state the valve would be turned to the OFF state. An alternative would be to select some other lower level to use for detecting of the valve ON state and then turn OFF when level 9 showed the tank was once again full.

The fill valve usage assumes that there is a pressurized source of water from which the water is supplied to the tank. On the other hand if a pump was required to move water from say a pond into the tank then the pump ON/OFF could be controlled in place of the fill valve.