NTC

Sorta looks okay, although you probably want to add a bit of roll-off so it doesnt oscillate around the setpoint.

If im not mistaken this will provide a constant current to the NTC wich will mean the temperature will self stablise but will vary depending on the thermal cooling unless the NTC has a particularly steep slope.

Also the fet will waste power unless this also goes into what you are trying to heat of course, an alternative is to use it in switch mode (just turning it into a relaxation oscillator might be enough)

If you want to make it closer constant temperature you could add positve feedback,... so as the temperature rises the resistance falls and the voltage accros the ntc falls too if you make the current equal to a proportion of the voltage accross the ntc it will give full voltage to the heater untill the temp rises and the resistance drops to some value wich matches the proportion. just adding a resistor from drain to -input will do it.

This will also likly cuase the circuit to oscillate if you have to much feedback, but this wil mean the FET will be on or off and so disipate less power anyway.

Colin =^.^=

Before I put too much energy into this circuit, I have a question for you. How do you know the heating element has a negative temperature coefficient? Have you checked this at anything near the normal temperature the heater was intended to operate at?

I ask this, because many heating elements that are intended to be self regulating positive temperature coefficient devices (and have a very high positive temperature coefficient at some transition temperature), have a much lower negative temperature coefficient at low temperatures.

Use a resistor to heat, not the NTC. So have the NTC sensor at R1, and leave the heater above the FET.

Rene

Huh? Most all are PTC.

"Homer J Simpson"

All Google Gropers know that.

....... Phil

I measured resistance at 32 F, room temp, and 212 F. R at 32 F: about 2.5 ohm room temp (74 F) about 1.85 ohm 212 .96 ohm

I using the Rankine scale I fitted the room temp and 212 deg data points into the Arrhenius equation of the form R = a exp (b/T) and got R = .0762 exp (1700/T) used this equation to predict R at freezing and got a result within a tenth of an ohm of the measured 2.5 ohm. This equation is a good model for the behavior of this heating element between freezing and boiling, and probably quite a bit beyond, but the heating element is meant only to heat vegetable oil to about 170 degrees F, very loosely. Ten or twenty degrees higher than that wouldn't do any harm. Anyway, it certainly has a negative tempco, hard as it may be to believe that they actually manufacture and sell such a thing. I knew something was wrong as soon as I took the thing out and put it on a car battery to measure current draw. The current start out low and very gradually crept up. So now I have a good mathematical model for the R-T characteristic of the material. I also pencil-traced this stuff on graph paper. By the way, after I posted the circuit I saw a mistake. The positive feedback makes the mosfet turns fully off, which makes a muck of current sensing. That circuit would work without the positive feedback, but unfortunately I can't have this thing run linear, there's way too much power going through it. By the way, it operates on 12 - 14 volts (automotive application).

That's GARGLE GROUPER like when a fish makes that funny pucker face

Snip

Close. Not actually constant current. It balances a bridge with the NTC and Rs on one side, and the voltage divider on the other. So it has the effect of maintaining the NTC heater at a constant resistance. Sounds funny, I know, but look at it and you'll see this is actually way better than a constant current circuit. It avoids the thermal vagaries you mentioned that would be concomitant with feeding the heating element a constant current.

Snip

I did put a feedback resistor to the op-amp positive input. It's just that the other end of the resistor is in the wrong place (see below).

That's actually the right place for the feedback resistor, not where I put it. Snip

Well, that's the idea, isn't it?

Nope. PTC = resistance increases with temperature.

NTC = resistance decreases with temperature. Most unsatisfactory for a heating element - good as a surge suppressor.

"Homer J Simpson"

....... Phil

Made no sense so it's you who is the idiot.

"Homer J Simpson"

Go look up the word " irony " sometime - f*****ad.

........ Phil

Okay, NTC heater control won't work with the switching element inside the sensing loop. Needs to look something like this, in essence:

V in | | ,--------------+----------------------, | | | | / / \\ \\NTC /R2 /heater \\ \\ / / | | | | | ,----------/\\/\\/\\/---, | | | | | | | |\\ | | +-----+------|+\\ | | | | >----------+ | | ,-------|-/ | | | | |/ | | | | | | / | | | \\ | | | /R1 | | | \\ | | | / +-----------/\\/\\/\\/---|----------+ | | | | | | | / | | | \\ | | | /Rs | '---||----+ | \\ | | | / | | | | '--------------+-----------|----------' | | | | '--| | ||-------' ,--| | ground

Then the question becomes what to do with the power pins on the op-amp.

You (and the OP) should take a look at the data sheet of a PTC heating element. They are typically _negative_ resistance to just under the transition temperature, then suddenly PTC to limit current and temperature.

The OP doesn't need any of that circuitry. He just needs to hook his heating element up to an appropriate voltage. See, e. g.,

John Perry

I don't think you can turn the power completely off and still have the control loop function. But you could switch it between a high and low poser condition by putting a resistor in parallel with the fet switch. That way, the bridge that compares the heater resistance divider to the fixed resistance divider would always have a bit of excitation. With an opamp that had an input common mode range that includes zero, like the LM358, but not the positive supply rail, this would work a lot better if the switch were on the high side. Then I think you could eliminate the capacitor in the feedback network and DC couple it.

This is a very interesting concept, and the only problem I see with it is how much power you have to waste with the low power feed resistor (nice if it could be part of the heater) to keep the bridge comparison working.

It appears that the OP wants to control the temperature to something well below the transition temperature (if any) of his heating element. He has tested its resistance to well above the intended operating temperature and found nothing but negative temperature coefficient. So it is the wrong heater for his application, but it is what he has.

I understand irony. Do you understand dumbfuck?

...

...

...

ah yes in fact I missed it had any positive feedback at all, but yes it needs to sense the ntc voltage not the gate voltage to keep the NTC at a constant resistance.

Well yes although it depends, if the switching would be too noisy and/or the mosfet is atatched to the thing you are trying to heat up anyway such as an oven for an OCXCO.

But as someone else mentioned are you sure its NTC? it seems rather strange to have this as a heater.

Colin =^.^=