Building a Water Flow Switch for a Water-Cooled PC

Detailed How-To – Ron Wlock

This article describes the construction of a water flow switch from simple inexpensive materials for a water-cooled PC. The flow switch can be connected directly to a pump or installed “in-line” in a section of tubing. The flow switch monitors water flow and is connected to an electronic circuit which uses light emitting diodes (LED) to indicate the flow of water.

A large passive radiator is used to water-cool my PC, in order to eliminate fan noise. An article about this project can be found HERE. Shortly after the development of the passive radiator project, the benefits of water-cooling compelled me to water cool the hard drive. An article about the hard drive project can be found HERE.

With dependency upon the movement of water to cool the CPU and the hard drive, it was apparent that some form of protection to prevent overheating would be prudent. Two different options were considered: One option was to monitor the CPU and hard drive temperatures and the other option to monitor water flow. The decision was made to design a system to monitor water flow.

Several water-cooling enthusiasts believe that PC water-cooling systems are reliable and it is unlikely a water pump would stop working. I agree with the low probability of pump failure, but wanted an extra margin of safety.

The original plan was to purchase a reasonably priced water flow switch, with low flow resistance that would connect to 12 mm (½”) inside diameter (ID) vinyl tubing. After researching a few products, this did not seem viable. For example, the Gems Sensor #165841 is reasonably priced at $49.00 CDN, but after experimenting with an actual sample, it was concluded that the flow resistance is high. The Gems Sensor #155484 appeared promising, but is $210.00 CDN. The Crydom #FS15A also seemed promising, but a supplier could not be found in North America.

Feeling frustrated, I decided to design and build a do-it-yourself water flow switch.

The following requirements were set as targets for the flow switch project:

• Simple, reliable and inexpensive to build
• Easy to build from common parts
• Sensitive to a low water flow rate of 3 litres per minute
• Low water flow resistance
• Easy to connect to 12 mm ID vinyl tubing
• Easy to connect to a device that signals that water flow has stopped

After researching water flow switch technology, it was decided to use a magnet and a magnetic reed switch as the principle components in the design.

The plan was to build a switch that operates as follows:

• When water flows through the switch, a paddle is pushed to the “on” position. The paddle contains a magnet and the magnetic field actuates a magnetic reed switch which “closes”
• When water stops flowing, the paddle moves to the “off” position and the magnetic reed switch “opens” as the magnetic field is moved away
• The action of the magnetic reed switch controls an electrical device that signals the flow of water with LEDs

In order to make the switch sensitive to a water flow rate of 3 litres per minute, the plan was to make the paddle partially buoyant using an air space. It was assumed the partial buoyancy would make it easier to move the paddle to the “on” position.

The plan also included the use of PVC material for the switch body, since a magnetic field is unaffected by plastic.
{mospagebreak}

Ron Wlock

1. The magnetic reed switch is a “normally open” model, 6 mm (0.25″) in diameter X 28 mm (1.12″) long. This type of switch is commonly used to monitor doors and windows in residential security systems. Magnetic reed switches “close” or “open” with the presence of a magnetic field, depending if they are “normally open” or “normally closed”. If the switch is a “normally open” model, the contacts remain “open” until they are exposed to a magnetic field, then they “close”. The contacts remain “closed” as long as a magnetic field is near the switch.
2. The reed switch holder is a 20 mm (0.80″) long tube cut from the barrel of a BiC pen.
3. The Tee is a 12 mm schedule 40 Tee with SLIP connections. The SLIP means the Tee does not have threads and is normally connected to pipe using PVC solvent cement. The inlet and outlet connections, as well as the PVC plug, are attached to the Tee using epoxy. The key to joining plastic parts with epoxy is to lightly sand the surfaces, wash with soap and water, and then allow to dry before joining. This provides a secure water tight seal. An added advantage of attaching the plug with epoxy is that the plug can be precisely positioned for proper paddle operation.

   The PVC Tee can be found at any major hardware store for under $1.00 CDN. An alternate switch design uses a Tee with threaded connections and the plug is threaded into position.

4. The outlet connection is an Orbit #37160 plastic hose barb for 12 mm ID tubing, c/w MIPT male adapter.
5. The paddle tube is 28 mm long and cut from the barrel of a BiC pen.
6. The magnet is 6 mm in diameter X 3 mm (0.125″) long and is a rare earth type. A rare earth magnet has a strong magnetic field for its small size. Note: the magnet shown in the picture above is 6 mm long. The magnet can be purchased for $1.00 CDN and suppliers are easily found on the Internet.
7. The wooden plug is 6 mm in diameter X 3 mm long and is cut from a wooden dowel.
   

8. The axle spacers are 2 mm (0.08″) in length and center the paddle on the axle. They are cut from a BiC pen ink cartridge.
9. The axle is 20 mm long and made from 14 gauge solid copper house wiring. It is used to attach the paddle to the PVC plug.
10. The paddle stop is a piece of plastic 5 mm x 15 mm x 2 mm thick (0.20 x 0.60 x 0.08″) that positions the paddle horizontally to ensure the magnetic field is in-line with the magnetic reed switch. A spare plastic drive cover from a PC case is a good source for this piece. The stop is attached with epoxy.
11. The PVC plug is 12 mm with pipe thread
12. The PVC close nipple is 12 mm with pipe thread and measures 35 mm (1.40″) long. It is used to connect the Eheim fitting to the Tee. The Eheim connector has a 21 mm (0.83″) ID and 12 mm pipe has a 22 mm (0.89″) outside diameter (OD). A piece of 12 mm PVC pipe could be sanded down to fit, but the close nipple with epoxy is a simpler solution.

13. The inlet connection is an Eheim 1250 pump adapter & O-ring, # 7438510. For some pumps, such as the Pondmaster model 3 which has a ½” NPT male discharge connection, a Tee with thread connections could be used. The Eheim # 7438510 adapter is usually shipped in the box with the pump.

{mospagebreak}

Ron Wlock

An alternate method of installation is to use 2 hose barbs on the switch and install the switch “in-line” in a section of tubing.

Care must be taken with the “in-line” design to ensure the flow switch is mounted in a vertical position so that gravity pulls the paddle down when the flow of water stops.

If the flow switch is intended for a horizontal operation, the paddle and plug assembly must be installed with the paddle “stop” on the up-stream side. Also with this design, a ‘normally closed’ magnetic reed switch is used because the magnetic reed switch must “close” when water pushes the paddle.

The disadvantage of the in-line installation is the resistance of 2 hose barbs added to the water cooling system. The additional resistance may lower the water flow rate and care should be taken to ensure the water flow is adequate to move the paddle.

An alternate design is a compact version to minimize space requirements.

The overall height of the flow switch is reduced by cutting 6 mm off of the Eheim connector, each leg of the Tee (6 mm x 2) and each leg of the 45 elbow (6 mm x 2). The picture above shows the parts assembled with epoxy, except for the elbow to Tee connection, which is assembled with PVC cement. The hose barb shown above is for 16 mm ( ¾”) ID tubing.{mospagebreak}

Ron Wlock

The plastic tube for the paddle is cut from the barrel of a BiC pen that has an 8 mm OD and a 6 mm ID. The magnet and wooden plug slide perfectly inside the barrel. Both the magnet and wooden plug are secured in place with epoxy. The magnet is located 2 mm from the end of the tube and the wooden plug 6 mm from the other end, creating an air space between.

As stated earlier, a design objective was to make the switch sensitive to a low flow rate of 3 litres per minute. It is assumed the force exerted against the paddle by the flowing water is proportional to the flow rate. For example, a flow rate of 3 litres per minute would exert a smaller force than a rate of 6 litres per minute. To ensure the paddle would be moved by a low force (low flow rate) the paddle is made partially buoyant with the air space between the magnet and the wooden plug.

The law of buoyancy (also known as Archimedes Principle) states:

“Any object, wholly or partly immersed in a fluid, is buoyed up by a force equal to the weight of the fluid displaced by the object.”

Using buoyancy terminology, a negatively buoyant object sinks in water and a positively buoyant object floats.

The correct buoyancy for the paddle is important. If it is very negatively buoyant it would appear as a “heavy” object in the water and be difficult to move with a low water flow rate. If the paddle is very positively buoyant, it would float up to the “on” position and stay there even when water stops flowing. This is not a desirable situation.

The plan was to design a paddle with the right amount of buoyancy to ensure it sinks, yet can be moved with a minimal water force.

After several “trial and error” buoyancy calculations, it was decided to use a paddle 28 mm long. This length of paddle, combined with the weight of the paddle, makes the finished paddle negatively buoyant by 0.183 grams (1.593 – 1.410 = 0.183). See the calculations that follow.

A flow test conducted on the switch indicates the paddle moves to the “on” position at a water flow rate of 2.6 litres per minute.

{mospagebreak}

Ron Wlock

As stated earlier, an objective was to design the switch with a low flow resistance to ensure it did not impede the performance of the PC water-cooling system. All plumbing devices, such as hose barbs and elbows, will create some amount of flow resistance. The goal is to minimize the resistance in order to maximize water flow.

The three components that contribute the majority of the flow resistance in the switch include the pump connector, the hose barb and the paddle. The plan was to use the flow switch to replace an existing pump connector/hose barb (shown below), therefore the only concern was the potential resistance of the switch paddle.

A test was conducted to identify the flow resistance caused by the paddle and consisted of pumping water through three different plumbing configurations. The three configurations are illustrated below:

The test equipment included:

• A Pondmaster model 3 pump with an Orbit #37170 hose barb on the discharge. The pump is rated at 1300 L/hour (350 US gal/hour), max. pump head 3.2 metres (10.5 feet), 45 watts
• 12 mm ID x 16 mm (¾”) OD vinyl tubing by Kuritec
• Orbit #37160 plastic hose barb for 12 mm ID tubing, c/w MIPT male adapter
• 12 mm PVC coupling used to connect the two hose barbs together
• The flow switch

The procedure followed for the test:

• The pump was placed inside an 18.6 litre bucket filled to the brim with water
• The pump was started and when water started to flow from the tubing, the end of the tubing was placed in a 4 liter container
• The time to fill the 4 liter container was recorded
• Seven trials were done for each case. Before each trial, the 18.6 litre bucket was filled to the brim with water

The data collected during the test is shown in table 1:

Comparing Case 1 to Case 2, the restriction caused by the “2 Hose Barbs and Coupling” reduced the water flow by 11%.

Comparing Case 1 to Case 3, the restriction caused by the “2 Hose Barbs and the Flow Switch” reduced the water flow by 13%.

Since the major difference between Case 2 and Case 3 is the paddle obstructing the flow of water, it can be concluded that the paddle only contributes a 2% reduction in water flow.

This low flow resistance of 2% is due to the free area around the paddle that provides a path for water flow.

The PVC Tee has a 21.5 mm ID which equals an area of 363 mm² and the obstruction created by the paddle is 172 mm². Therefore, the free area for the water to flow around the paddle is 191 mm² (363 – 172 = 191 mm²). For a comparison, consider the free area of a typical hose barb for 12 mm ID tubing. The hose barb has an ID of 9 mm which is equal to a free area of 64 mm². The free area in the PVC Tee is three times the hose barb free area.

The round shape of the paddle also helps to minimize flow resistance by creating a laminar flow.{mospagebreak}

Ron Wlock

The magnetic reed switch on the water flow switch is connected to a simple electronic circuit which indicates the flow status. When water is flowing, a green LED is illuminated. Should the water stop flowing, a red LED is illuminated. The circuit is connected to the 12 volt wires of a disk drive plug on the PC power supply.

The interface circuit is assembled on a piece of perforated board and packaged in a Hammond box # 1551HGY measuring 60 x 36 x 28 mm (2.36 x 1.38 x 1.10″).

All of the design objectives for this project were achieved – the flow switch performs well.

The interface circuit and LED’s are a terrific addition to the water-cooling system, since the water pump is located in the garage and I cannot easily confirm that it is running. However, the interface circuit is only a temporary solution because it has the disadvantage of requiring the presence of a human to take remedial action if the water flow stops.

My future plan is to design and build another interface circuit which will automatically shut down the PC (without human intervention) if water flow stops. I am currently at the preliminary stage of designing such a circuit which functions without the use of a software program.

I would like to thank Rich Beketa for his assistance with the design of the interface circuit.

Ron Wlock