Comparing temps

[Jerry In Maine]

I have a room in my basement where the furnace and hot water heater are installed. It gets really warm, especially in the winter when the furnace runs a lot, so I'd like to be able to detect when the inside room temp is higher than the area outside the room where I work. When this occurs I like to run a wall mounted exhaust fan (via a solid state relay) to push the heat out of the room and hopefully on to me. For now I'm just using an LED to indicate if the output is on or off.

I have a 741 op amp with two voltage dividers on the + & - inputs. On the ground side of the dividers I have thermistors. They're the same except that the + side divider has a 330 ohm "offset" resistor to raise the + input up a bit even when temps are equal. This is to ensure that the output of the op amp remains off when temps are close.

To test this I tape both thermistors to a common surface so they're kept at the same temp. Voltages then are as follows

Neg level is 3.2
Positive is 3.0 V
Output is 1.3 V

The - thermistor is warmed by touching it, which causes the - input voltage to go lower than the + level. Output of the op amp goes to 5.2V and LED glows. This is what I want, and I'd expect the LED to go back off when the temps equalize as they were prior.

Trouble is the output never goes back off even when the temps normalize. I'd expect that the - & + levels would go back as they were if the temps of the surface they're attached to hasn't changed.

I can ground the + input so its level falls below that of the -. The LED will then go off and stay off.

I can also "force" the output off by warming the other thermostat and driving it voltage well beyond the opposite input.

What seems to be happening is that when the output goes high it raises the level of the + input above that of what is being represented by the ambient temp. With this "additional" voltage hanging around on the + input the - input level can never catch up, so the circuit is in effect locked. Grounding the + and dropping the output brings all levels back to normal.

Can anyone suggest a fix?

Thanks!

 

[MikeMl]

Too much positive feedback. Increase the value of the feedback resistor (much more than 500K) by putting in a 1meg pot and/or putting a 1-2meg fixed resistor in series with the pot. Post the nominal values of the bridge resistances, the nominal R and delta R of the thermistors at the desired delta T, the supply voltage, the desired amount of hysteresis, and we can calculate the feedback resistor...

 

[Jerry In Maine]

Parts are what I had laying around in the Junque Box;

Resistors on the dividers are 68.8K

Got the thermistors from a surplus outlet. Best I know about them is that they're 48K at ~72*F and drop with increasing temp. The temp they'll sense in use will range from ~60 to ~90*F.

Supply Voltage is a 9V battery.

Hysteresis: Don't have an exact number...long as the output dosen't "chatter" I guess is OK.

 

[MikeMl]

ok, eliminate the 330 Ohm offset resistor. Make the feed back resistor about 2.2meg.

Correction: see the attached simulation in LTSpice. Make the feedback resistor about 4.7meg.

 

[Jerry In Maine]

Thanks. I'm out of the meg range resistors but will pick up some this evening.

If I understand correctly hysteresis will cause the output to go high when the + input is "more than just slightly" higher than the - input. This is why I don't need the offset resistor - correct?

But it'll snap low when the + goes even slightly lower than the -.

Don't want to waste a "learning moment" here

 

[MikeMl]

Thanks. I'm out of the meg range resistors but will pick up some this evening.

If I understand correctly hysteresis will cause the output to go high when the + input is "more than just slightly" higher than the - input. This is why I don't need the offset resistor - correct?

But it'll snap low when the + goes even slightly lower than the -.

Don't want to waste a "learning moment" here

Yes, look at the red trace in the simulation results (voltage at the + input V(p)). Note that it snaps above and below the the green trace (the voltage at the - input V(m)) during the switching points. The magnitude of the step is what creates the hysteresis. The magnitude of the step is controlled by the current that flows through the feedback resistor.

 

[Jerry In Maine]

Installed 4.7m Ω as feedback and the circuit works just as it should...

thanks Mike!

 

[MikeMl]

Isn't LTSpice neat? It will answer a lot of "what if" questions...