electronics experts: what component am I looking for?

I’m a mechanical engineer by trade, but I’ve fiddled with electronics just a little bit in the past. I need some advice for an electronics project I’m working on.

I have a 24 Vdc power supply for a project, and I need to power a set of small fans with it (total draw 0.1 amps).

From a pot, I have a control voltage that I can vary from 0-10 Vdc.

When I select a pot output voltage below 1.5 volts, I want the fans to be off; for 1.5 volts and higher, I want the fans to be on.

My 10-volt source for the control pot can’t provide much current, a few mA at best.

Question: what component do I need for closing/opening the connection between the fans and the power supply?

Sounds like a job for a relay, but any relay I’ve found so far (solid state or otherwise) that is closed when there’s 1.5 volts on the trigger will be damaged (or will draw too much current from my pot supply) when the full 10 volts gets applied. I’ve looked at op-amps and comparators, but it seems like the ones I’ve found (at least while searching on Digikey) all can’t pass enough current to power the fans.

So what do I need to make this work?

Is the power to the fans independent of the pot voltage?

You do this with a comparator, voltage divider (or voltage reference) and transistor (to drive the relay, or control the fans directly).

Look for a simple circuit using a CMOS op-amp wired as a comparator. One input of the amp is used to set the comparison voltage, the other is used to monitor the voltage you want to control fan operation, and the output (though a transistor or very light-duty relay) is used to switch the fans on and off. You can use almost any op-amp from the venerable 741 up, but a CMOS variant will draw very little operational current.

The circuit will require a stable supply voltage, so don’t plan on running it from the varying fan-power voltage. There are ways to make self-stabilizing comparators that run from the same battery, though. Your main problem will be turning the very light-duty output of the op-amp to a switching signal sturdy enough to handle the fan power. (A beefy transistor for up an amp or so DC; a light transistor to a suitable relay for almost anything else.)

PM me if you want a quick circuit diagram - include specifics about all the voltages, loads, etc.

A comparator may not be able to drive a fan by itself, but you can use a transistor on the output; for example a LM393 (pot input to inverting input, 1.5 volts to non-inverting input) driving a 2N2907 transistor (if the positive line is switched) with a 10k resistor to limit base current (2.4 mA is enough for a 100 mA load); switching the negative line needs a 2N2222 or equivalent, with the resistor connected between base and 24 volts (no resistor needed to the comparator since it is open-collector; you also have to reverse the inputs so the output is low when it is “off”).

Here is a schematic illustrating how to connect the parts, with both variations shown (the LM393 is a dual comparator, you only need one section):

Note: If your 24 volt supply is regulated, you can just use a couple resistors for the 1.5 volt reference, such as 100 k to 24 volts and 6.8 k to ground, with the comparator input to the junction. The PNP transistor is also shown with an additional resistor from base to emitter to ensure it is completely cut off when the comparator’s output is high (floating).

You don’t really want to use an op-amp as a comparator because they can misbehave, up to output inversion (the output switches unexpectedly to the opposite rail) and increased power consumption* and comparators offer better performance (since that is what they were designed for).

For similar reasons, I wouldn’t recommend a relay for the OP’s needs unless he is switching AC mains voltages; even a small low-power general purpose transistor (2n2222/2907) is sufficient (at heavier loads, I’d use a MOSFET, which can easily be found, for 24 volts, with on resistances low enough to switch 10 amps or more with no heatsinking).

*PDF from Texas Instruments discussing these issues

For simple DC sensing where speed is irrelevant and simplicity is called for, I can’t imagine what else you’d use. There are many ways to make an op-amp comparator circuit stable and reliable in the ways the OP is seeking to use it. I’ve used generic 741-class op amps for this kind of duty for more than 30 years.

What exactly are you proposing as an alternative?

A comparator?

Also, using the correct part makes it easier to go back and figure out what the circuit was for; op-amps for amplifiers and comparators, for well, comparators. Not that I haven’t done the same thing, but only because I was already using an op-amp and didn’t want to add another chip (even then, if I needed two op-amps and a comparator, I’d use a dual op-amp and a comparator in separate packages instead of a quad op-amp unless I didn’t have any on hand).

Ah, the transistor seems to be the piece I was missing; thanks.

For the reference voltage, I was planning to use a multi-turn pot so I’ll be able to adjust it to the exact value I need.

For anyone who’s curious, my project is as follows:

I have a torchiere lamp with a 300-watt halogen bulb in it. A fair mount of light at a color-temp of 3000K, and a lot of heat. It’s in a big room with a high vaulted ceiling; I want MORE light, less heat, less power consumption, and better color temperature. Solution: I’m converting to LED’s. 2.2 time the lumens, 40% of the power consumption, and a color temp of 4000K.

The challenge is that the LED’s need to be kept cool. Each of the six 20-watt LED’s gets its own small heat sink and fan. Power to each LED is managed by a FlexBlock, which incorporates a dimming feature. All six will be dimmed by a single pot on the lamp’s main post that works with ten volts (supplied by the FlexBlocks). The LED’s dim in a continuous manner down to a dimming pin voltage of 2 volts. When that dimming pot dials the dimming pin on the FlexBlocks down to 2 volts, the LED’s turn off; when the pot is dialed down to 1.5 volts, I want the heat sink fans to shut down, and that’s where I was running into trouble. The FlexBlocks can’t provide much current (they’re expecting a multi-Kohm dimming pot), and so I didn’t have much to work with for controlling the connection between the fans and the power supply.

Why not just put a temperature sensor on one of the heatsinks?

If that particular LED goes open circuit for some reason, the sensor won’t get warm, and the fans won’t turn on - and pretty soon the remaining five LED’s will pop.

A temp sensor (RTD probably) will still require an op amp + MOSFET transistor to trigger the fans.

Your objection is reasonable, but I can’t help but think this lamp is going to be annoyingly loud. A variable-speed temperature-controlled fan might help with that.

12 dB each, six fans should add up to about 20 dB. Ought to be barely audible hidden up above the lamp’s reflector/dish.

What is the θja of the heatsinks you are using with .11 m^3/min of airflow?

1.9 degrees C per watt for the sink itself. the LED’s dissipate 10-12 watts each, and have an internal resistance of about 0.8 degrees per watt (IIRC). So 2.7 degrees per watt, times 12 watts, plus maybe 30 degrees ambient, gives an operating temp of about 62 degrees C. The LED’s are characterized at 85C, and it looks like they’ll tolerate quite a bit more than that, so I think they’ll be OK.

Plus the resistance of your thermal attachment.

But, it still looks OK. I’d measure it with an IR thermometer after it’s been running for a while, and see how close it measures to calculated values.

I’ve done several HB led lamp replacement projects (using the Seoul Semi ACRiche LED). I’ve always been able to do them with passive cooling, but I’ve had plenty of space.

It should be noted that that value depends on the amount of airflow, which in this case is 200 linear feet/minute, and to make things complicated, the fans in the datasheet you give are rated in cubic meters per minute (if I am correct, you can convert that to cubic feet/inches and then to a column of air with a cross-sectional size equal to the heatsink to get LFPM (I get 225 LFPM; 0.11 m^3 = 3.88 ft^3 = 6712.6 in^3 / 2.48 in^2 (area of heatsink) / 12 = 225); although airflow also depends on how much pressure drop the fan has to work against, the rated CFM being for free air, and again, the units used for pressure drop are inconsistent).

Also, the datasheet you give for the fans only lists 5 and 12 volts, so are you really running them off of 12 volts? I am guessing you are using the F410T-12LC, 12 volts at 40 mA and intend to put them in three strings of two fans each, for 24 volts and 120 mA. Don’t try to connect two fans in series off of 24 volts because they don’t draw continuous current (being the same, it would be safe to put a capacitor, around 100 uF, across each set of 3 parallel fans to smooth the voltage out).

Although I think it would be better to use one big heatsink and perhaps one fan (if you don’t want or can’t use too big of a heatsink), rated at 24 volts and physically much larger (bigger fans can push more air at lower RPM, leading to less noise, especially at higher frequencies).

I love the electronics discussions here. Problems are posed that any hobbyist could solve in about three minutes with junk from his or her parts bin - solve well, easily, permanently and with no unpleasant side effects.

By the third response here, someone is suggesting an elite bit of hardware only available to corporate engineers working under NDAs.

By the tenth post, two or more posters are furiously squabbling over esoteric specs that don’t actually apply to the OP’s situation.

By the fifteenth post, the physics geeks have gotten into it, greek letters, equations, physical constants and all.

Not every problem needs maximum engineering and a solution that didn’t exist two years ago. Just sayin’.

Actually, planning on using the 5-volt model, running them off of a 5-volt DC-DC converter powered by the 24V source. Between 12 and 5 volts it would have been a coin toss (either one requires a converter to step down from 24V), but the 5V setup is just a smidge more efficient.

I plan to angle the six LED’s about 15 degrees off of plumb so they aren’t all blasting their output straight up at a single spot on the ceiling; this precludes laying all of them on a single flat heat sink. If six small fans proves to be silly-noisy, I might consider pairing the LED’s onto three larger heat sinks, which would still let me angle the three clusters independently. Will see what happens when I start testing.

We hobbyists wouldn’t have much to do if we always settled for off-the-shelf solutions with inherent compromises. :smiley:

Well, in a way, that is what you are doing; when I need a fan, I grab one from my collection of fans removed from discarded electronics (maybe 50 fans, also have a huge box filled with heatsinks), same for many other parts, except for those that I use a lot of or “special” parts (e.g. high power MOSFETs, SiC diodes). Similarly, I’d build my own DC/DC converter, which can be built using as few as two transistors and a handful of other parts (this took about 10 minutes to design; 12 volts is better for efficiency since the diode will drop about half a volt and also conducts less at a higher duty cycle/output voltage; you can also easily add a shutdown disable without having to switch the power) if you don’t mind not having current limiting, and run at a very decent efficiency for such a simple circuit.

Of course, I know you mean just going out and buying some LED bulbs and a fixture.