I'm a complete newbie at electronics and would appreciate any help with this.. I've have managed to buld these two fan-controller circuits: http://members.aol.com:/ns1496/Fancontrol.bmp http://www.cpemma.co.uk/graphics/393pwm_th_basicsch.gif Both work but unfortunately neither do quite what I want. My aim is to have the fans start at 25.5 (78) ramping up to full at 27.5 (81.5). I'm using them to cool an aquarium hence the narrow range. Both the above circuits seem to rely on a range of at least 10 degrees. Looking at cpemma's spreadsheet it seems that such a narrow range is not possible with thermistor/resistor networks. I therefore used an lm35 and op amp circuit to provide 4-12v over my 2 degree range. This works but has all the disadvantages of linear voltage control. Ideally I'd like to use PWM as in cpemma's circuit. I then looked at the 5v PWM chips by Microchip - the TC64x family - http://www.microchip.com/1010/pline/analog/anicateg/thermal/dcfan/tc6xx/tc648/ These provide PWM fan control and expect 1.25v to 2.65v input for 0-100% duty cycle. The datasheet suggests using a thermistor circuit for this but I thought I could using my op amp/lm35 circuit instead. I now have a circuit with lm35\opamp and TC648. The op amp is tuned to produce 1.25v to 2.65v over my temp range. However, it doesn't seem to be provide enough voltage to turn the fan :-( . The transistor is the suggested SI4410 though I've also tried a BUZ10 which seems similar (?). Is there something really basic I've missed that means I have to use a thermistor circuit with this chip? Is there another way I can use PWM over my 2 degree range? Thanks
It may surprise you to know that a simple 1N4001 diode can be used as a very accurate temperature sensor. I used to have some circuits using a 1N4001 in an application similar to this as I can't stand using thermistors with slow response times. I'll have a look and see if I can dig these circuits up. Meanwhile try a search on google for a temperature sensor using a diode.
Welcome, reefer1. If you can rig your LM35/OpAmp to give a range of around 4v, you can use that for the reference voltage in a PWM circuit. Taking the classic triangle-wave/comparator circuit - You can move the triangle wave up or down to suit your minimum reference voltage by varying R4 & R5 (or replace them with a 100k pot, wiper to U1a pin#2), and also vary the height of the wave if needed by adjusting the ratio of R2:R3. Be worth simulating (CircuitMaker Student is freeware) to get the resistor values, or use a 47k pot and 33k resistor instead of R3. Then just feed the reference voltage from your LM35/opamp into U1d pin #12 or U1c pin#9 and tweak to suit. (May be possible to use U1d as the LM35's op-amp, it's not doing anything important in the above circuit, just a unity gain voltage follower). Any dc voltage should be fine for the reference, whether from a pot, a thermistor pot divider or opamp output. If you've got the TC648 running on 5v there may not be enough output voltage to hard switch some mosfets but I can't see why an NPN switching transistor shouldn't be OK.
Thanks for the replies. For the moment I've abandoned using the tc648 as I couldn't get consistent results with it. Just to explain what my lm35/op amp is doing: op amp 1 compares the difference between the lm35 voltage and a threshold voltage set by a pot; op amp 2 then adds a gain of about 500 to make the output big enough to be useable. The LM35 on its own puts out 1mv per degree celsius. I've connected the circuit kindly provided by cpemma to the output of my lm35/op amp at pin U1C10 - I found if connected to U1C9 fan speed dropped as the sensor output voltage rose. I'd like to add the suggested kick start capacitor - would this go to U1C10 and ground? Also I'm finding the 120mm fans growl a bit - would this be helped by varying the wave?
Yes, voltage across a NTC thermistor drops as temp increases, but voltage from LM35 increases, so a swap-over at the comparator is the fix. The effect of the kick-start capacitor is to make that ic pin lowest for a short time, so in your case I think it should go between the reference voltage pin & ground, so a cold sensor/low LM35 voltage still starts with the fan at high speed. Most PWM-run fans growl at low speed, I think the bearings play a major part in the noise level. You could try increasing the PWM frequency (reduce C1), but YMMV. Another trick is to use a bipolar transistor switch and blunt the PWM pulses with a small capacitor on the base, see Telcom's AN58 for details.
