I'm trying to build a reliable circuit that is based on analog components to detect the temperature of a device and shut it down if the temperature gets too high potentially damaging the device.

So far, I have brainstormed and have this layout in my head:

1. Thermistor is affixed to device to be monitored, and is part of a voltage divider.

2. A Shmiddt trigger op-amp monitors voltage from voltage divider (1), if voltage corresponds to a temperature that is below the maximum, op-amp output is low and relay providing power to device is de-energized

3. If voltage corresponds to a temperature that is higher than maximum, op-amp output is high and the relay providing power to the device is energized and cuts the power circuit powering the device, while the cooling system on the device remains on

4. Once device temperature returns to normal, op-amp output goes low, and power to the device is restored

Is this reasonable?

 

I'd use the base-emitter junction of a transistor as the temperature sensor.

 

Instead of an op-amp, I'd use a comparator. The differences are slight but a comparator is purpose-built for this chore. Just like an op-amp, you can build in hysteresis so the relay doesn't chatter.

An op-amp or comparator is not likely able to drive a relay directly. You'll want a MOSFET as a switch to drive the relay. Depending on what the load is (if it's DC powered), you could eliminate the relay and just use the MOSFET. But if the power is AC, a relay is a good choice for isolation from your DC circuit.

Personally, I'd use a LM35 thermometer IC instead of a thermistor. It doesn't require any calibration for most applications and you can validate its output signal by simply measuring the output voltage with a multimeter.

Since you emphasized reliability, one option is a commercially available thermostat instead of any DIY solution. Thermostats are available at every price and quality level. You can even get data logging and remote access (for instance over the internet) for not much money.

 

Sounds ok to me, op amp with hysteresis and a Lm35 as a sensor.

 

Is this reasonable?

It is certainly reasonable.
But I think one of these would be easier, cheaper, and more reliable: http://uk.farnell.com/webapp/wcs/st...tegoryId=700000005944&langId=44&storeId=10151

 

Thermistors have a non-linear T/V curve and is best used with a uP and a look-up table, I'd go with the LM35.

Some time back I needed something similar for the heatsink of a power amp. I used the LM35 to drive a LM3514. With some parts you can scale the LM to the temperatures required. E

 

I'm trying to build a reliable circuit that is based on analog components to detect the temperature of a device and shut it down if the temperature gets too high potentially damaging the device.

So far, I have brainstormed and have this layout in my head:

1. Thermistor is affixed to device to be monitored, and is part of a voltage divider.

2. A Shmiddt trigger op-amp monitors voltage from voltage divider (1), if voltage corresponds to a temperature that is below the maximum, op-amp output is low and relay providing power to device is de-energized

3. If voltage corresponds to a temperature that is higher than maximum, op-amp output is high and the relay providing power to the device is energized and cuts the power circuit powering the device, while the cooling system on the device remains on

4. Once device temperature returns to normal, op-amp output goes low, and power to the device is restored

Is this reasonable?

What I did... It works!

 

Titles like this always make me wonder: Where does one study to make un-reliable circuits? Sales and marketing?
Anyway, this is a set point device. I did it with a thermistor because I had some laying around. This one has been in service for nearly 40 years.

 

Where does one study to make un-reliable circuits?

The internet?

 

Titles like this always make me wonder: Where does one study to make un-reliable circuits? Sales and marketing?
Anyway, this is a set point device. I did it with a thermistor because I had some laying around. This one has been in service for nearly 40 years.

A 723 as a power output comparator - interesting.

ak

 

Thermistors have a non-linear T/V curve

Don't care; this is a single trip-point controller, not a thermometer. Yes, thermistors ae non-linear. They also have a much greater volts-per-degree change than a diode or transistor. This reduces the effect of the comparator input offset voltage as an error term.

My go-to part is a Thermometrics DC95F. Its resistance at any temperature characterized to six decimal places. In a comparator circuit with 1% resistors, it is interchangeable without any calibration.

If the comparator circuit drives a transistor to drive the relay, better to go with an opamp; it can source current into the base, rather than have a pull up resistor with the comparator pulling the base close enough to GND to turn it off.

ak

 

A 723 as a power output comparator - interesting.

ak

Many years ago, I was the designer at a power supply company, so I had some LM723's laying around, too.
They are just a voltage reference and an op-amp with a collector and emitter accessible npn output and a defeating transistor in the package. Similar to an LM10 IIRC.
You don't need the defeating transistor for a thermostat, but the output driver is rather convenient. Bit of an op-amp with balls.

 

I'm trying to build a reliable circuit that is based on analog components to detect the temperature of a device and shut it down if the temperature gets too high potentially damaging the device.

So far, I have brainstormed and have this layout in my head:

1. Thermistor is affixed to device to be monitored, and is part of a voltage divider.

2. A Shmiddt trigger op-amp monitors voltage from voltage divider (1), if voltage corresponds to a temperature that is below the maximum, op-amp output is low and relay providing power to device is de-energized

3. If voltage corresponds to a temperature that is higher than maximum, op-amp output is high and the relay providing power to the device is energized and cuts the power circuit powering the device, while the cooling system on the device remains on

4. Once device temperature returns to normal, op-amp output goes low, and power to the device is restored

Is this reasonable?

What temperature are you talking about?
Space available? Cost?

 

Many years ago, I was the designer at a power supply company, so I had some LM723's laying around,
You don't need the defeating transistor for a thermostat, but the output driver is rather convenient. Bit of an op-amp with balls.

I've used the 723, but I've never been a big "fan". I think the part was a bit ahead of its time in terms of process; the amplifiers are ok but not great in terms of errors, noise, and GBW; the reference is ok but not great, too high for many of today's applications, ; the output darlington is ok but it sets up a pretty large headroom requirement (probably because IC process PNP transistors sucked back then), especially if you are adding an external power transistor or power darlington. It is a great part for an average power supply. Reading this, I sound like a bit of a snob. That's fair - I wouldn't use it for a lab supply.

ak

 

Where does one study to make un-reliable circuits?

Thanks for all the replies. I'll digest the suggestions and do some research on them on my end.

What I meant by reliable is I'd like it to be a simple analog solution vs a micro-controller, which are cheaper than even just a few analog components nowadays.

 

Here's a simple 18 to 30C sensor....
Decrease R2 for Higher temperature.

 

What I meant by reliable is I'd like it to be a simple analog solution vs a micro-controller,

Good point. Today I am working on taking out the 6th malfunctioning microcontroller in my 2005 Ford.
It sure would be nice to have a dome light when I open the door, but there is a microcontroller between the door switch and the dome light.

 

the amplifiers are ok but not great in terms of errors, noise, and GBW;

So who needs those for an adjustable set point thermostat? Meh. Junk in a drawer becomes a thermostat that has worked beautifully for 37 years. Some jobs are so crude that crude components work just fine.

 

alberthall said: It is certainly reasonable.
But I think one of these would be easier, cheaper, and more reliable: http://uk.farnell.com/webapp/wcs/st...tegoryId=700000005944&langId=44&storeId=10151

Thank goodness for some common sense! I know the OP said he wanted an analogue solution, but I suspect that this is just that he is less familiar with digital circuits than with analogue ones.

Why does everybody have to complicate things? The device above will do everything that the OP requires at lower cost and a fraction of the effort than most of the other solutions put forward.

Remember - KISS!

 

I'm trying to build a reliable circuit that is based on analog components to detect the temperature of a device and shut it down if the temperature gets too high potentially damaging the device.

So far, I have brainstormed and have this layout in my head:

1. Thermistor is affixed to device to be monitored, and is part of a voltage divider.

2. A Shmiddt trigger op-amp monitors voltage from voltage divider (1), if voltage corresponds to a temperature that is below the maximum, op-amp output is low and relay providing power to device is de-energized

3. If voltage corresponds to a temperature that is higher than maximum, op-amp output is high and the relay providing power to the device is energized and cuts the power circuit powering the device, while the cooling system on the device remains on

4. Once device temperature returns to normal, op-amp output goes low, and power to the device is restored

Is this reasonable?

If you don't need tight accuracy - you might not need any circuitry.

There are various sealed/enclosed bi-metal thermal circuit protectors.

Possibly the best known package style is found on the side of the magnetron in a microwave oven, they've been around a lot longer in washing machines and tumble dryers.

There are much more compact package styles embedded in many rechargeable battery packs - and probably in temperature ranges better suited to your application.