View previous topic :: View next topic |
Author |
Message |
jetskier
Joined: 09 Dec 2003 Posts: 287 Location: Nevada |
Posted: Thu Jun 26, 2008 1:40 pm Post subject: Hysteresis in a simple thermostat fan circuit |
|
|
Well after having a closet full of equipment, I finally added a vented to the outside to help keep the closet a little cooler. A few years back I built this little circuit for my entertainment center cabinet. I used an old 12V AC-DC transformer (300mA) and a 3" PC fan in conjunction with the simple circuit. Since moving into this house and using my whole-house system, all the equipment moved out of the cabinet to the closet. I upgraded to a larger 4" fan to match the new exhaust (dryer vent) and a larger ac-DC transformer (700mA) to handle the bigger fan.
In both cases, the circuit is too sensitive to minute temperature fluctuations near the off threshold. Can any experts see a simple way to add a little hysteresis into the circuit? After searching someone suggested adding a 500k or 1M ohm resistor (or 1M trim pot) between pins 3 and 6 of the 741 IC to adjust the on-off time (aka hysteresis). My goal is to get the fan to cycle down about 2 degrees below the "on set point" before turning off. Right now is shuts off for about a second or two and then goes to full on. The fan doesn't even get a chance to spin down. |
|
Back to top |
|
|
Tommy Tyler Expert
Joined: 21 Sep 2003 Posts: 412 Location: Denver mountains |
Posted: Fri Jun 27, 2008 9:02 am Post subject: |
|
|
This is a strange little circuit. Brush noise filters for DC motors are usually much smaller than C1, and some of the higher RF frequency noise won't be eliminated by an electrolytic capacitor. The author says C1 also "provides a capacitive start for the motor" (whatever that means) and should not be omitted. This is a DC motor that needs no capacitor to start, as some AC motors do. The DC motor does have a starting current surge, but rather than helping provide that, C1 actually diverts some of the otherwise available starting current to charge itself up. This circuit shows poor design practice because every time the fan starts, the completely discharged C1 presents a momentary dead short to transistor Q1. In fact, I'm a little surprised it's still alive after "a few years". Maybe R5 limits Q1's collector current to an acceptable value. Well, whatever works beats theory any day, I always say.
To predict hysteresis from a 1M feedback resistor Rf from IC1-6 to IC1-3, assume Vcc is 12V and the op amp output switches "rail to rail" as they say, or from 0V to 12V. We know that the circuit switches when the resistance of R1 drops to slightly below that of R4, or 8.2K at 88 degrees F. We also know that R1 is 10K at 70 degrees F. That's a change of 1.8K over 18 degrees, or 100 ohms per degree. (We can assume linearity over such a small temperature range.)
Before fan turn-on, Rf is in parallel with R2, effectively reducing its value to (10K x 1M)/(10K + 1M) = 9.9K. The threshold reference voltage at IC1-2 is (8.2K/18.2K) x 12V = 5.41V. At fan turn-on, (R1/R1 + 9.9K) x 12V = 5.41V. Solving for R1 gives 8.13K as the turn-on value. If you draw a straight line curve representing thermistor temperature versus resistance between 70 and 88 degrees, you'll see that the turn-on temperature is actually 88.8 degrees instead of 88 degrees.
After fan turn-on, Rf is switched in parallel with R1, effectively reducing its value to (8.13K x 1M)/(8.13K + 1M) = 8.06K. Now R1 has to cool down to where the parallel combination of R1 and Rf looks like 8.2K for the fan to turn off. That means R1 must increase in value to 8.27K, which in parallel with 1M gives 8.2K. Back to our temperature versus resistance curve, that's a temperature of 87.3 degrees. So the circuit SHOULD have a hysteresis of (88.8 - 87.3) = 1.5 degrees F when using a 1M hysteresis feedback resistor.
Sorry for all the boring math. I can't explain why the circuit doesn't work for you after adding the hysteresis resistor. I just wanted to offer some encouragement and let you know the suggestion you were given seems like a good one to me. Note that connecting Rf from the collector of Q1 to the inverting input (pin 2) of IC1 should provide about the same amount of hysteresis. You could try that to see if it works any better.
I notice the author used a 2N2907 and said the fan should be 200mA or less. You may need a transistor with more beans for your bigger fan.
Tommy |
|
Back to top |
|
|
jetskier
Joined: 09 Dec 2003 Posts: 287 Location: Nevada |
Posted: Fri Jun 27, 2008 9:31 am Post subject: |
|
|
Tommy, I was hoping you would reply!
The 2907 I have is rated at Ic=600mA.
I think I understand the math. I used a 10k trim pot for adjustability of the set temp in place of the 8.2k for R4.
I appreciate your input. I'll let you know if I can achieve any hysteresis. 1.5° would be great. |
|
Back to top |
|
|
jetskier
Joined: 09 Dec 2003 Posts: 287 Location: Nevada |
Posted: Tue Jul 29, 2008 3:44 pm Post subject: |
|
|
Update:
So far it seems to be cycling at about 1-1.5 degrees. I need to adjust it a little, but it is keeping everything a little cooler. |
|
Back to top |
|
|
fuji0010
Joined: 29 Apr 2009 Posts: 1
|
Posted: Wed Apr 29, 2009 5:19 am Post subject: |
|
|
A programmable timer circuit that can be connected directly to the fan switch of any thermostat with a forced air heating and/or cooling
_________________
programmable Thermostats |
|
Back to top |
|
|
|