A Quiet Fan Controller Speed varies with temperature
Many modern radios and other electronic devices rely on muffin fans for cooling.These can be loud and annoying. Some run continuously, while others cycle on and off, either when needed or just on transmit. In some radios, the fan cycling results in a small frequency shift as the oscillator is heated and cooled.
Wouldn’t it be preferable to have a fan with a variable speed, responding to cooling needs? I’ve thought so for a long time, but never got around to doing something about it. Recently, I decided it was time. I figured this was an obviously useful thing, so there would be lots of circuits available on the web. NOPE! The only things I could find were microprocessor circuits, many of them relying on fancy fans with internal tachometers – none of those in my junkbox. Also, the microprocessor controls the speed by turning the fan on and off rapidly; some of the notes suggested that the results are audible.
Most muffin fans use DC brushless motors, so the speed is easily controlled by varying the motor voltage. 12-volt fans are convenient and readily available. Also, there are a number of inexpensive temperature-sensing ICs available. What we need is a simple circuit to vary the fan voltage in proportion to temperature – basically, an amplifier.A couple of opamps should do the job.
I sketched out some circuits and simulated them with the free SwitcherCAD III software from www.linear.com. None of them worked satisfactorily, so I called the opamp guru, Byron, N1EKV. He agreed that it sounded simple and would look into it. He soon called back to say it wasn’t as simple as it sounded, because of some choices I had made: to keep one end of the temperature sensor and one end of the fan grounded, and to drive it with a power FET for minimum voltage drop at full speed. The result is that the sensor is referenced to ground, but the FET is referenced to the positive voltage. The final complication is that there is a huge gain in the circuit due to the transconductance of the FET, about 2.5 Siemens (in tube terms, this is 2,500,000 μmhos – a typical tube is 5000) or more. To make things worse, the FET is operating in a non-linear region, and having non-linear elements inside a feedback loop is never a good idea.
I went back to engineering basics: find a circuit to steal. One of the microprocessor fan controls used an interesting circuit to drive a FET and shift the reference from ground to high side. The circuit, in the area of Q1 and Q2 in the schematic, looks like the Widlar current mirror used in many integrated circuits. I added this circuit plus the PNP emitter follower, Q3, and fiddled with the resistances to get it going. Then I consulted Byron again and added capacitors C4 and C5 to stabilize things.
Download Fan speed controller PCB, Schematic, part list..