MAX6675 temperature instability

sky999

MAX6675 temperature instability [Copy link]

 

When MAX6675 is connected to a K-type thermocouple probe, the temperature is fairly accurate, about 32 degrees, with a positive or negative error of 2 to 3 degrees. However, as soon as your hand or skin touches the K-type thermocouple wire or probe, the temperature immediately jumps to 10 or 20 degrees. When you take your hand away, it returns to normal.

maychang

The electromotive force of a thermocouple is only in the mV range, and the voltage of the human body is also several mV. If your hand touches the thermocouple, it will certainly affect the measurement results.

As for the voltage on the human body, you can see it by looking at the electrocardiogram. Each grid on the vertical axis of the electrocardiogram represents 1mV.

sky999

maychang posted on 2019-8-17 16:49 The electromotive force of a thermocouple is only in the mV order of magnitude, and the voltage of the human body is also several mV. If your hand touches the thermocouple, it will certainly affect the measurement results. As for the human body...

The K-type thermocouple probe has two wires, one positive and one negative. Touching both of them with my hand will not affect the result. However, if I just touch the iron sheet outside the wire, it will jump around. Why?

maychang

sky999 posted on 2019-8-17 16:53 The K-type thermocouple probe has two wires, one positive and one negative. I touch both of them with my hand without affecting the result. Just touching the iron sheet outside the wire will...

I guess the iron sheet outside your thermocouple probe is not well grounded.

sky999

maychang posted on 2019-8-17 17:06 I guess the iron sheet outside your thermocouple probe is not well grounded.

Does this iron sheet need to be connected to the ground? I have never seen anyone do this.

maychang

sky999 posted on 2019-8-17 17:23 Does this iron sheet need to be connected to the ground? I have never seen anyone do this

This question is rather complicated.

It's hard to say without seeing your thermo-electromotive force amplifier circuit (MAX6675, including AD conversion, microcontroller, display and power supply) or the actual object.

I guess your amplifier circuit (including power supply) is not well grounded, so the amplifier circuit, power supply and AC mains will be coupled through the distributed capacitance between the primary and secondary of the transformer (whether it is powered by a power frequency transformer or a switching power supply), and the thermocouple skin is not well grounded. There is a large distributed capacitance between the human body and the AC mains, and the AC mains will be applied to the amplifier circuit through the distributed capacitance between the human body and the AC mains wire and the distributed capacitance between the thermocouple skin and the thermocouple, causing interference.

It is best to connect the amplifier circuit to the ground and the outer skin of the thermocouple to the ground. Or connect the outer skin of the thermocouple to the "ground" of the amplifier circuit. You can try it and see the result.

sky999

maychang posted on 2019-8-17 18:14 This question is quite complicated. I haven't seen your thermo-electromotive force amplifier circuit (MAX6675, including AD conversion, microcontroller, display and power supply), ...

I didn't connect the amplifier circuit, the probe is directly connected to the T+T-

maychang

sky999 posted on 2019-8-17 20:53 I didn’t connect the amplifier circuit, the probe is directly connected to the T+T- of MAX6675

I didn't connect the amplifier circuit, the probe is directly connected to the T+T-

A1 and A2 inside MAX6675 are both amplifiers, and there are analog switches at the input and output ends of the amplifiers.

sky999

maychang posted on 2019-8-18 16:09 I didn't connect the amplifier circuit, but the probe was directly connected to the T+T- of MAX6675. A1 and A2 inside MAX6675 are both amplifiers, and the amplifier input terminal and...

No, the chip must be connected to GND. Now it can display normal temperature when connected to a 1-meter probe, but it is abnormal when touched by hand. When connected to a 5-meter K-type probe, it is always 0 and no temperature can be collected.

sky999

maychang posted on 2019-8-18 16:09 I didn't connect the amplifier circuit, but the probe was directly connected to the T+T- of MAX6675. A1 and A2 inside MAX6675 are both amplifiers, and the amplifier input terminal and...

By the way, are the two switches automatically turned on when CS is pulled low?

sky999

maychang posted on 2019-8-18 16:09 I didn't connect the amplifier circuit, but the probe was directly connected to the T+T- of MAX6675. A1 and A2 inside MAX6675 are both amplifiers, and the amplifier input terminal and...

Should the negative terminal T- of the thermocouple be connected to the external ground or the circuit board to GND?

maychang

sky999 posted on 2019-8-19 14:22 By the way, are those two switches automatically turned on when CS is pulled low?

"Are those two switches automatically turned on when CS is pulled low?"

You need to read the datasheet carefully and analyze its working process to know it.

maychang

sky999 posted on 2019-8-19 14:36 Should the negative terminal T- of the thermocouple be connected to the external ground or the circuit board to GND?

"Should the negative terminal T- of the thermocouple be connected to the external ground or the GND on the circuit board?"

The thermocouple should of course be connected to the chip pins and then to the power supply that powers the chip.

sky999

maychang posted on 2019-8-19 16:03 "Should the negative pole T- of the thermocouple be connected to the external ground or the circuit board to GND?" The thermocouple should of course be connected to the chip...

Then I am very curious, should the T+T- trace be thicker or thinner? Also, what is the distance from the K-type thermocouple to the circuit board? If copper wire is connected, will it cause errors?

maychang

sky999 posted on 2019-8-22 09:28 Then I am very curious, should the T+T- trace be thicker or thinner? Also, what is the distance from the K-type thermocouple to the circuit board? If copper wire is connected, will it...

If the distance between the thermocouple and the circuit board is far, a compensation wire should be used in the middle instead of ordinary copper wire. Different types of thermocouples must be equipped with corresponding compensation wires and cannot be mixed.

maychang

sky999 posted on 2019-8-22 09:28 Then I am very curious, should the T+T- trace be thicker or thinner? Also, what is the distance from the K-type thermocouple to the circuit board? If copper wire is connected, will it...

“Should the T+T- wiring be thicker or thinner?”

Thickness doesn't matter much, since amplifier input resistance is usually quite high and the current is small.

sky999

maychang posted on 2019-8-22 09:48 If the distance between the thermocouple and the circuit board is far, a compensation wire should be used in the middle instead of ordinary copper wire. Different types of thermocouples must be used...

OK, then I feel that the traces in the PCB cannot be too long. After all, they are all copper film lines. Is it necessary for T+T- to be of equal length?

maychang

sky999 posted on 2019-8-22 11:03 OK, then it seems that the traces in the circuit board cannot be too long, after all, they are all copper film lines, so is it necessary for T+T- to be of equal length?

"Does T+T- have to wait that long?"

Thermocouple signals have a very low frequency and it is not necessary for the wires on the circuit board to be of equal length.

maychang

sky999 posted on 2019-8-22 11:03 OK, then it seems that the traces in the circuit board cannot be too long, after all, they are all copper film lines, so is it necessary for T+T- to be of equal length?

"Then I think the traces in the circuit board can't be too long, after all, they are all copper foil lines."

You don't understand how thermocouples work. Thermocouples measure the temperature difference, which is the hot end temperature minus the cold end temperature. The hot end is of course the point where the temperature is to be measured, and the cold end is your circuit board. If the temperature point to be measured is far away from your circuit board, the thermocouple must be made very long. However, the two metals of the thermocouple are expensive (especially precious metal thermocouples, such as platinum and rhodium), so the thermocouple cannot be made too long. If the thermocouple is connected to two copper wires, the connection with the copper wire becomes the cold end, and the temperature here is not zero degrees Celsius. Extending from the point where the temperature has dropped, if two metal wires (compensation wires) with the same temperature characteristics as the thermocouple (but cannot withstand such high temperatures) are used to connect the thermocouple, then the cold end is at the other end of the compensation wire, that is, your circuit board. Therefore, the compensation wire cannot be used with the wrong type, and when the correct type of compensation wire is used, the two compensation wires cannot be swapped (the two compensation wires are different alloys).

maychang

maychang posted on 2019-8-22 11:53 "Then I feel that the traces in the circuit board cannot be too long, after all, they are all copper film wires" You don't understand the working principle of thermocouples. Thermoelectric...

"Then I think the traces in the circuit board can't be too long, after all, they are all copper foil lines."

The length of the copper wire in the circuit board has nothing to do with the measurement results, provided that the temperature at the connection between the copper wire and the compensation wire is the same as the temperature at the connection between the copper wire and the chip pin.