My project is a coffee maker, which needs to dispense water into a reservoir in measurable and reproducible amounts. Originally, I was just considering turning a water spout on for a defined amount of time, but this doesn't seem like it will be very accurate. I've read some threads on avrfreaks and elsewhere, and there are lots of interesting ideas, but I wanted to see what people here thought about it. It seems like there are three basic options:

1. Use a pressure sensor under a leg of the reservoir.
2. Use an air pressure sensor with a tube submersed in the reservoir (so as the water rises, the air pressure in the tube will rise)
3. Use some sort of simple conductive thing (wires/resistors) submersed in the reservoir.

Option #2 seems like the most fun (and expensive), and option #3 seems easiest but least advisable (is it unwise to send current through the water?). I don't really need to dispense arbitrary amounts of water -- I'd be happy with specific levels like 8, 12, 16, and 24 ounces. Any ideas? Other options? Has anyone worked with pressure sensors and liquids?

Why wouldn't timing be accurate? If you are always using the same amount of water in the same container it will always weigh the same.

A alternate sensor is optical use a led and a phototransistor projecting a beam across the container when the beam gets blocked turn the flow off. Using optics you have none contact.

Ralph

Timing might not work becUse faucet water pressure can vary by day and location and temperature.

Timing can work if you have a regulator on the water line, set for a certain pressure to give you a known amount over a period of time. We had coffee makers at work that did just that. I think Ice makers in domestic freezers like in your kitchen, also regulate the pressure and just activate the valve for a set time.

Varying pressure was indeed the reason why I thought simple timing might be inaccurate. I'll have to see if a pressure regulator would be small and inexpensive enough for my project. I was also thinking of having a second reservoir that always fills to the top, and a small water pump that pumps from that into the brewing reservoir. The water pump might provide a more constant flow rate.

The optical sensor is a good idea. I'm worried that it might not work if it's trying to "see" though a glass or plastic container, or that water, being relatively transparent, might not block it properly. Any experiences with that?

I had a liquid level that used two clear pieces of tubing with a length of water hose in between.

I think it used a IR Led and Phototransistor. Radio Shack part # 276-0142.

One would have to play to see what worked.

Now if you wanted to get fancy (anal) about your water you would get a Reverse Osmosis water filter, they (a least GE's) come with a 1 gallon storage tank. The water taken from the storage tank would be of a constant temperature and pressure.

I still think timing would work it would not be be more than one or two grams difference from a domestic water supply at best even without using a storage tank.

Again another good reason to do some experimenting (playing around) with electronics.

Ralph

There is always the pre-measured kind, like a toilet tank. Fill the tank with a small line and let the tank do the work.

How bout a float sensor? Say Chip -> turns on water -> Float valve sends an interrupt that shuts the water.

I bought a 1gpm brushless (submersible) CPU cooling pump for around $17. If you used something like that you would get a constant flow and therefore you can use the algorithm time = desired amount / flow rate to find out how long to run your specific pump.

i.e. if you wanted 2 gallons then t = 2/1 or 2 minutes....if you wanted 6 fl oz then first convert the flow to oz (1 gal = 128 fl) so t = 6/128 or .046 minutes