Electronic – Detecting dry-running of water pump

Well, assuming you are operating here on earth (as opposed to in orbit), you can use a float sensor that you measure using the micro. The question is: Do you need to know the exact level, or do you just need some sort of trip when it gets to high/low?

On the 'trip at specific level' theme, you can get switches like this - generally the float contains some kind of attitude sensing switch. When the water level is lower, the float ends up on it's side, and the switch is 'off', while raising the level forces the float upright, and therefore 'on'. These are often used for sump pumps. They are pretty reliable and very easy to deal with, though you have to be careful to anchor them correctly, and you have to be sure that nothing gets in the way of the float.

If you need to know the actual level, then you've got lots of choices, but it's less simple. You can go with something like an ultrasonic sensor (ultrasound beam firing down from the top of the tank to reflect off the water level).

These guys seem to make level indicators for BIG tanks which consist of a float and an external weight cabled together. As the float rides up and down on the fluid, the external indicator moves appropriately. You could build something similar, with nothing more complicated than photo-cells to read the level (if it's a large tank, or you want lots of granularity, you'd need a lot of photo-cells).

Alternately, taking the cabled float idea, you could put an idler wheel that moves whenever the cable moves, and then attach an encoder to that. This would allow you to track the level with pretty high accuracy.

AC motors are not as easy to speed control as DC motors; their torque is related to both the applied voltage and the applied frequency. This is known as the Volts:Hz ratio. In order to maintain torque at a speed other than its rated speed, you need to adjust the frequency in proportion to the voltage.

If you want to do it right, you rectify the AC into a DC bus and then use IGBTs to chop up the bus (using PWM) to generate an AC waveform at the frequency and average voltage you need. This is not quite as simple as it sounds, and single-phase AC motors have capacitors which are the sworn enemies of active devices such as IGBTs.

All is not lost, though: If you can live with the torque drop-off you can simply phase back the voltage applied to the motor using a triac. Use an optocoupler to detect when the AC wave crosses zero and then use that as a reference point to drive a timer which controls where in the AC wave that you turn on the triac. This is classic phase angle firing or phase angle control and works for general dimmer applications for lighting and heating.

You have something in your favour: most water pumps are a cube-log load; the torque requirements drop off drastically as the speed of the pump drops. If you have a positive-displacement type of water pump (piston pumps) you will not be so lucky, but centrifugal and impeller type pumps are fairly easy to slow down. You should be able to get close to a 2:1 speed reduction and still move water, but no more. Also remember that you may not be able to meet the static head requirements of your system if you slow down too much; the pump simply will not be able to make the head pressure required. You won't break anything, but it may not work as well as you want.

All of this should be VERY doable with an arduino or other small microcontroller. I used to design three phase motor controllers using the PIC microcontrollers available in the late 90s. Just keep in mind that you will be playing with line voltage; use optical isolation both for detecting zero crossings and for driving the thyristor, and work slowly and deliberately when you're testing.

Note that if you just want a knob to turn down the pump you can do it without any microcontroller at all using a diac and a triac. Do some web searches for resistive heater controllers or soldering iron temperature control, or "phase angle" triac controllers. Tons of examples online.