# Thermocouple comparator choose

I measured the parameters of a thermocouple (R1). R1in=5 Ohm, V=1..10 mV (measured by multimetr).

Ipositive = 5v/20000 = 0,25mА, and Upositive max = 0,25 * 0,001 * 100 = 25mV

I have lm339, lm324. Can this comparator catch 1-2 mv? Is exist any comparator catch 1-2 mv? How do I choose the right comparator? What parameter should you pay attention to?

UPD: Later... I found a сircuit that suits me more or less, thanks to the answers I understand how it works.

I have lm339, lm324. Can this comparator catch 1-2 mv?

Nope, because the offset error is ±1mV.

You should use an instrumentation amplifier to amplify the thermocouple's output voltage, and then feed it to a comparator. That's the safer way.

simulate this circuit – Schematic created using CircuitLab

• The LM324 is also amplifier, can I use one element of LM324 as amplifier? One element as amplifier, and two as comparator? Sep 14, 2020 at 10:55
• @nick_n_a That's why I said "instrumentation amplifier". LM324's max offset error is 1mV. If your sensor generates 1mV and the opamp's error is +1mV then the output voltage will be twice the expected. It gets worse if the opamp's error is -1mV. So, you should either a) use an amplifier dedicated to using thermocouples (preferred), or b) pick an opamp having as low offset error as possible, or c) calibrate your system if you are to use a standard opamp/amplifier. Sep 14, 2020 at 11:01
• I expected to see a circuit diagram with elements and values in the answer, but I think today is best answer:) Sep 15, 2020 at 9:45
• For thermocouples OP177GPZ is good and avalible preamp. Sep 15, 2020 at 9:55

Given you have a potentiometer, you can adjust out the offset error.

What will then vary is the change_in_offset_voltage with temperature.

Some comparator datasheets provide that specification. Some do not.

You also need lots of gain, to amplify 0.1 millivolt of thermocouple change (a few degrees C) into 5 volt output changes ----- without oscillation, all inside ONE MECHANICAL PACKAGE with ONE BYPASS CAPACITOR. That means gain of about 100,000 X, or 100 dB gain, very broadband and very likely to oscillate.

And you want a little bit of hysteresis, perhaps 1 degree C? which is about 40 microVolts, for some thermocouples. Whereas the Comparator will have finite Power Supply Rejection Ratio at DC, and essentially ZERO PSRR at 1MHz and higher, where VDD sagging will occur with internal charge demands as the comparator logic output changes. All this suggests AN OSCILLATOR.

Better to use the Preamplifier, the INA.

Yes, you can use a cheap op-amp, however the Vos may prevent you from setting it as low as 1mV. If you bias the T/C positive by greater than Vos can be you can overcome that. The chances are 50/50 it will be in the wrong direction. Worst-case Vos is +/-9mV but most will be within a few mV.

It would be better to use a zero-drift op-amp. You do not need or want an instrumentation amplifier since you have a single ended signal, Vos and gain are the main concerns.

Since you have no cold-junction compensation whatsoever, the relationship between the LED state and temperature will be heavily dependent on the temperature at the points where the T/C is connected to copper. Also the cheap op-amps have relatively large TCVos, unspecified, but typically 5-10uV/°C. Depending on the type of thermocouple and your actual target temperature, both those effects may be large. The cold junction effect is ~1:1, meaning that if the PCB temperature changes by 20°C there will be around 20°C error in the switching point (for most thermocouple types). TCVos can add or subtract from that, but a type K thermocouple has about 40uV/°C output near room temperature so a 20°C change might change the switching point by a few °C due to TCVos.

• I plan to measure the temperature in the region of 200-400 Celsium degrees, turn on/off the heater (soldering iron). Delta - I think 20 degrees will be good. Sep 14, 2020 at 15:03