PWM Controller



I design cooling systems as part of electronic equipment designs. Although I mostly deal with direct air cooling, once or twice a year I have to deal with water cooled systems.

There are a few automotive cooling myths. Probably the worse and most popular myth is that removing a thermostat will make water flow too fast to cool. In reality, unless flow becomes grossly turbulent  through significant areas of the system, the more flow the better the cooling becomes. I can't imagine an automotive system ever coming close to that point, with the sole exception of excess water pump impeller speed causing pump cavitation and arresting flow.  

To understand cooling systems, we have to understand heat exchange.

Flow Basics

Typical USA water cooled engines flow like this:



Cool water is drawn from the radiator outlet tank (A) into water pump (B).

The pump (B) pushes cool water into and though the block towards the rear (C).

The cylinder head (C) returns water to the thermostat housing (D).

The thermostat (D) blocks flow when cold, opening at a predetermined temperature. This stabilizes the engine temperature around the thermostat setting.

Water from the thermostat flows into the radiator's hot tank (E). This is normally a high spot on the radiator.

Water from the radiator inlet tank (E) is forced through many small hollow tubes. The tubes have fins attached outside to increase air contact area. This cools the water. After reaching the cold tank (A), the water repeats the cycle.

Normally, a limited water volume constantly recirculates from the engine water outlet (D) to the water pump inlet (B).  The recirculation ensures the engine warms evenly, and the thermostat sees a constant water supply at the engine temperature. This circulation from the engine outlet at thermostat is almost always necessary. This small circulation prevents formation of steam pockets or hot spots, or delayed T-stat opening. 

Pressure and Flow

Horsepower are almost meaningless ways to compare or select a radiator, fan, or pump! Pumps, radiators, and fans may be specified that way, but the number does not mean much by itself. Flow without a pressure curve is just as bad. As usually happens, meaningful information is too complex or unfamiliar for most consumers. Since the good information would be confusing, manufacturers move outside the meaningful circle to a fringe specification that the consumer is comfortable with....even if the information is next to useless.

Looking at CFM of a fan or pump, we find graphs like this:


Air flow 50 Hz


The black curves are flow rate or volume, while the red curve is pressure. Notice the maximum pressure (red line) stays about the same over a wide range of flow rates, from zero (dead headed) to around 450 CFM. After 450 CFM, pressure rapidly drops off. This fan flows around 950 CFM of air, but at zero differential pressure. In the middle of an empty room, without any restrictions, the fan would move around 900 cubic feet of air per minute. Most automotive fans and pumps are, unfortunately, advertised by the non-restricted operation. While this makes the numbers look larger, pumps and fans with lower non-restricted flow ratings can move more volume against pressure than higher open flow-rated fans or pumps.

If we put a radiator at the fan outlet, and if that radiator required 1 inch of pressure differential to flow 800 CFM air and properly cool, this fan would only produce about .9 inch static pressure at 650 CFM. The fan would fall short, even though it is a 950 CFM fan and could produce 1.17 inches of static pressure at 400 CFM. The CFM (or GPM for a pump) and pressure at that flow have to match the system, and neither is close to the published ratings.

If we placed a restriction at the fan inlet, making the fan a "puller", the fan would be even more sensitive to restrictions. When restricted, almost every fan or pump works better when pushing into the restriction.

I measured three fans (at exactly 12 volts) pushing and pulling into a stock Ford Mustang GT radiator. The shroud or plenum, as a proper design should, covered the entire radiator area. The shroud was also about 4 inches deep, to even the air pressure and flow across the plenum area. DH is dead head pressure, pressure is inches of water differential. More water differential across a radiator will directly translate to more air volume, while less pressure differential will translate to less air volume (CFM), through the Mustang radiator:


Fan Measurement Comparison

The actual measured fan flow does not agree with manufacturer's rated flow. Because of this we can buy a "bigger" fan but really have a smaller fan.

In measured performance, a low-cost single speed Hayden 3700 flows almost identical air through a radiator restriction as the Derale fan's high speed mode, even though Hayden's advertised CFM is just over half the Derale's advertised CFM! We also see, when restriction is high, all fans work significantly better as "pushers".

The inches water column differential (pressure) was measured this way. Notice it is similar to using a volt meter. I was measuring pressure across a flow resistance to determine flow rate:


Simple electrical equivalent:


For both the radiator and the fan, just as with a water pump and radiator, we must know the back pressure and the flow rate at that back pressure. Flow by itself, just like pressure by itself, is meaningless.











Pushing or Pulling

A fan works by pressure differential. Atmosphere pressure pushes air into the inlet because the blade creates a low pressure area. The blade compresses the air on the outlet side, increasing pressure above atmosphere. This pressure differential causes air to flow.

At very low pressure differentials it doesn't matter much if a fan pushes or pulls.  In other words, if the radiator system does not "block" or choke airflow with backpressure, a pusher or puller will work equally well with a fan blade optimized for the service type. Pushing does require a slightly different blade design than pulling. As a matter of fact, severe restrictions require a centrifugal fan that cups and slings air out into a scroll housing. Think about a supercharger or turbo fan, or your furnace blower. Those are the extremes of working against back pressure, or pushing air through a restriction.

At low restriction, pusher or puller work equally well **if the fan is properly designed for the system**. You may have noticed that earlier, where I made fan measurements.


More Detailed Look

There are some special things going on in the cooling system.

The block has holes directly along the head. These holes, even the small holes, are very important for the following reasons:

1.) The upper areas of the cylinder walls, along with the cylinder head around each combustion chamber and exhaust port, are subjected to very high temperatures. At the least, this results in some areas running hotter than other areas. At the worse, localized heat causes some very hot areas.

2.) Trapped air, steam, or lack of flow can cause localized boiling inside the engine. The localized boiling forms a gas pocket of steam, preventing water from cooling that area. This further overheats the area where steam collects, overheating some small areas even though the overall engine is not overheated.

The solution to this is the addition of small "steam holes" between the block and head. The block typically has dozens of these small steam holes, along with larger water  water passages, through the head gasket into the cylinder head. The gasket holes are often sized to place the largest passages furthest from the water pump, with small steam holes strategically located in places where steam or air might become trapped.  The function of the multiple small holes from block to head is to prevent stagnant areas or steam pockets. The larger passages, mostly concentrated at the rear, ensure adequate water flows to cool the entire block and head.

Almost all systems have a way to bypass a small flow of water around the thermostat. The thermostat bypass ensures a limited water volume circulates even when the thermostat is closed. If the bypass wasn't there, water would sit stagnant. This would cause very uneven water temperatures as the engine warmed up, likely resulting in steam forming long before the thermostat area reached opening temperature.

The opening temperature is roughly 10-20% below the atmospheric pressure boiling point of water. This sets the operating temperature of the engine coolant, which varies roughly around the thermostat temperature with a properly function cooling system. Water leaving the thermostat and flowing into the radiator inlet tank (E), on a fully-warmed engine with properly functioning cooling system, is roughly around the thermostat's temperature.

As water passes through the radiator, heat is transferred from water into the air. This cools the water.

Old data I have from an SAE paper shows a belt-driven water pump has to flow at least 3 gallons/minute for every 10 hp (steady power) to adequately remove heat from an engine.

Contrary to myths and outright falsehoods, lack of a thermostat, or too cold of a thermostat, cannot cause an overheated engine. As water flow increases, temperatures throughout the system become more even. The ambient temperature, radiator design, airflow velocity through the exchanger, and water velocity through the radiator tubes determine the total system heat dissipation. As radiator air or water flow rates increase, more heat is pulled from the water system.   

The graph below shows water flow rate (GPM) vs. heat transfer from water (BTU) for various water flow rates with a small radiator at maximum rated airflow. This is a radiator we use to cool electronics. Electronics, like cars, can be anything from a steady thermal load or heat dissipation rate over a long period of time, to a varying dissipation with very high peaks and long recovery times.

Note the faster the water, the more heat is pulled from the system (until the radiator limits flow). All radiators behave this way:



Pressure is not a factor in cooling, with the exception water's boiling point increases about 3 degrees F for every 1 psi pressure increase. Pressure caps increase the coolant's boiling point. Pressure does not increase cooling rate, but pressure does allow hotter temperatures without boiling.

Cooling Fan

During operating times when external airflow or convection does not allow adequate water-to-air heat exchange in the radiator, the fan's role is to force air through the radiator. Open flow is not important. The important parameter is flow through the radiator's restriction. This is measured as a pressure differential between the radiator inlet side, and the radiator outlet side. Despite pressure differential between inlet and outlet being the determining flow factor, most fans are specified by open air flow!

There is a good electrical analogy for airflow:


My Cooling System


This is my home made fan shroud.

An ideal shroud would cover all of the radiator fin area, forcing air through all of the cooling fins. Additionally, the fan shroud should be as deep as possible. A deep shroud, especially deeper away from the outer edges, makes pressure more uniform across the surface of the radiator. To limit having to weld pieces together, I made my fan shroud as deep as possible just with box and pan brake bending.

My deep shroud, because of limited water pump clearance, required an inset fan. The inset moves the fan away from the water pump. This was an unfortunate compromise over ideal.


Radiator fan and shroud



For physical reasons, I elected to use an electric water pump. I used the Meziere 300 series electric water pumps WP311S.  This pump supposedly flows 55 GPM.

Pump and Fan Controller link


Center brass barb Derale


Teflon Insert adaptor Derale controller probe



Outside shell from barb fittting




Intake manifold adaptor Derale controller



Derale fan controller water pump







End of text for now.