Developing a Thermistor Mod for the Fan Mate 1

[Rory Buszka]

Some time ago, I looked at a Zalman Fan Mate 1 fan controller and thought, "What if I replaced the potentiometer with a thermistor?" If a thermistor can be modded onto a Fan Mate 1, it would offer a very inexpensive, easy-to-use way of adding single-channel thermal control without sending away to Austria for an MCubed FanAmp (and $40 shipping on a $25 item). The Fan Mate 1 is also a linear voltage source, so concerns about PWM noise would be eliminated.

I took my Fan Mate 1 controller and opened it up by getting my fingernail under the outer shell, and then pushing a credit card into the gap formed. This allows the enclosure to be pulled apart fairly easily. Then I traced out the circuit on the PCB, and used my multimeter to measure component values.

There really isn't all that much to a Fan Mate 1. A single power transistor is used to regulate the voltage linearly, which is why the big heatsink is required. The +12V supply enters directly into the emitter of the transistor. The collector of the transistor provides the positive voltage output to the fan, and also supplies voltage for the base, across a 670 ohm resistor. A 1k potentiometer connects from the transistor's base to ground. Maximum speed (4.25V) is accomplished by setting the potentiometer to nearly zero ohms, so all of the current flows across the potentiometer. Maximum speed (10.1V) is accomplished by setting the potentiometer to 1k ohms, energizing the base to 6.7V.

Because of the circuit design employed in the Fan Mate, my original thought of simply putting the NTC thermistor in parallel with the potentiometer won't work, because as the part heats up, the fan would slow down, instead of speeding up. The thermistor needs to control the positive voltage to the transistor's base. I tried putting the 10k thermistor across the 670 ohm resistor, but any change in temperature produced only a tiny change in fan speed -- nowhere near the range of adjustment needed. At 42 deg C, the thermistor's resistance is 7.5k ohms, and at 58 deg C, the thermistor's resistance is 5.5k ohms. A much more satisfactory result was achieved by placing the thermistor directly between the +12V supply and the transistor's base. However, when I placed the thermistor on a heat source, the resistance of the thermistor decreased enough to allow too much current to pass, heating the thermistor even more and causing it to burn up and open.

Though I burned up the thermistor, I still think that pulling up base voltage from the +12V supply is the right track to take. My question, then, is this: How could I decrease the amount of current flowing through the thermistor while still energizing the transistor's base sufficiently? Should the thermistor provide base voltage for a smaller transistor which could handle the additional current involved? Or would it simply be enough to place the thermistor in series or parallel with a resistor?

If it helps, I can supply photos of the Fan Mate 1's innards, including a photo of the PCB traces.

[Rory Buszka]

Today I went to Fry's Electronics and purchased two NTE 123AP transistors and a package of 3.3kohm resistors, as well as a big 10k linear potentiometer. I used one 3.3kohm resistor in series with the potentiometer to mimic the lowest observed resistance of a 10K NTC thermistor when working around 65 degrees C.

First of all, I tried simply connecting up the transistor so that its base voltage was pulled up from the same source as its emitter voltage, and then passed through the potentiometer and resistor, which represented the thermistor. I then placed the transistor across the 670 ohm resistor in the circuit. Miraculously, nothing vaporized instantly. However, the effect of each adjustment was too small. Removing the 670 ohm resistor entirely allowed adjustment of the voltage between 7 and 11 volts, which could be acceptable for most applications.

Another configuration I tried was with the 670 ohm resistor and the 1k pot removed completely, with all base voltage for the power transistor passing through the smaller transistor, and pulled up directly from the 12V supply line. This, however, did not offer a particularly good range of adjustment, only from 9 to 11 volts. I think that this configuration should work, and that I simply need more series resistance before the smaller transistor.

[BillyBuerger]

Wow, I like this project. Please keep us updated on your results. I was always interested in circuits and stuff, but ended up going the software route. Came easier to me.

So is sounds like you're saying a 1K ohm positive thermistor would be ideal. no resistance/low fan voltage when cold, high resistance/high voltage when warm. Drop that in place of the pot and your set. Or am I missing something?

[Rory Buszka]

Um, yeah. That would do the opposite in terms of controlling the fan, and you'd likely end up self-heating the thermistor and burning it out. Thermistors can't handle much current at all. Thermistors have a negative temperature coefficient, which means that their resistance decreases with increasing temperature. As the temperature of the element itself increases due to the current flow, it will only accelerate its own thermal failure.

As I mentioned, current across the thermistor has to be kept very low. That's why I was working with a smaller transistor with a high current gain (around 200) to accomplish what I wanted. This way, the thermistor would only need to pass a small current, and the transistor would pass the higher current. I used the thermistor-controlled transistor in place of the existing 670-ohm resistor, instead of in place of the potentiometer. Now that I know what will work best, however, I need to get another Fan Mate and perform the modification more cleanly so I can take photos. I sort of ripped apart the other one. Perhaps then someone might be able to spot a way to better refine the modification to allow a greater range of adjustment. Right now, the adjustment occurs over a range of 4 volts. For quiet fans, running at 7 volts may be all that's required. I had hoped to be able to get down to 5 volts, though.

I'm no electrical engineer, so I'm sure there is a way to get a broader range of adjustment without using opamps, but I'm not sure what that would be. I wish some of the electrical engineers on this board would take out one of their old Zalman Fan Mates (because everybody's got them) and give me their thoughts.

[cpemma]

There really isn't all that much to a Fan Mate 1. A single power transistor is used to regulate the voltage linearly, which is why the big heatsink is required. The +12V supply enters directly into the emitter of the transistor. The collector of the transistor provides the positive voltage output to the fan, and also supplies voltage for the base, across a 670 ohm resistor. A 1k potentiometer connects from the transistor's base to ground.

I think you'll find the Zalman "transistor" is an LM7805 5V voltage regulator. When the pot resistance is zero, the regulator's 'ground' terminal is grounded and output is the nominal 5V, raising the potential to the 'ground' pin by way of the pot-resistor potential divider raises the output by the same amount up to the 10.5V-ish maximum (on 12V in).

This regulator isn't very amenable to a thermal-control system as you can't easily get near the 5V minimum and still have some voltage increase at higher temperatures with thermistors, plus the regulator has a high quiescent current (4mA). A simple thermal control can be made with the L200 regulator.

Still, best of luck with it.

[Rory Buszka]

I had figured that it would be more similar to the first Emitter-Follower circuit on your (cpemma's) Simply Transistors page. The design of the Fan Mate makes it difficult for me to simply look at the markings on the thing under the heatsink, so I figured it was just a simple power transistor, not a regulator as you've revealed.

This lends some new insight that should help me get something better going. I'm considering simply desoldering all the components on the Fan Mate and breadboarding them -- I don't want to break any more transistor leads. Once I break this next NTE 123AP, I've only got an NTE 11 (which is similar), and then I have to go shopping again. I think I'd use the transistor to apply the potential to the ground pin on the regulator, now that you've explained what's going on here. That way, the transistor carries the current, not the thermistor. The thermistor would then be connected between the +12V line and the base of the transistor.

I don't know how much time I'll have to work on this, now that my classes are starting back up. Not to mention that a ton of what I'm doing is just trial and error. I'm actually surprised I haven't smoked any transistors yet.