Cold-junction point

Question

I'm converting an kitchen top oven into a reflow oven and all seems fine except for the temperature measurements.

Using the same K-type thermocouple, but an external cold junction, the measurements are correct. Both the K-type's that have the 'cold junction' in the ovens control compartment, which heats up by design (for the bi-metalic temperature controller), are off by about 30 or even 40 degrees C.

Now I think that the cold-junction reference of the MAX 6675 and the actual cold junction temperature value don't match, as I think that the chip heats up much slower than the exposed metal of the thermocouple connectors. I've soldered one of the thermocouples directly to the PCB as the screw terminals broke, so heat transfer should be much better, which didn't make any difference unfortunately.

I'm using a predesigned PCB that features the MAX 6675, looks quite similair to this one: https://www.newegg.com/p/295-0026-001J6

I've covered the inside of the oven with heat reflective tape, which seems to work fine, but the control compartment still reaches temperatures of 40C when the temperature inside the oven is around 140C.

So how should I deal with this issue?

• Add heatsinks on top of the MAX6675's to allow for better heat transfer into the chip?
• Place one or multiple fans inside the control compartment to pull cold air through it?
• Use soft fluffy isolation material that they use inside house roofs?
• Preheat the control compartment to a fixed temperature of 40/50/60C?
• Using thermally conductive epoxy? (I think it's better to attach the metals directly to the PCB instead of running them through a connector first)

EDIT: I'm testing the oven with the MAX6675 chips out side of the hot compartment and the temperature is still way off. The only difference is that the probe connected to my DMM, has about 8 or 9cm of wire inside the actual oven and the probes that I installed inside the oven, only have the hot junction inside the oven. Could this be the explanation of the big temperature difference?

EDIT 2: I've wrapped the thermocouples wiring with heat reflective tape and sticked them right above the surface of a metal bin. The temperature readings make much more sense now, I'll try a PCB later on to see if it still comes out black.

Answer 1

Measuring TC properly is not trivial, that is why RTC is usually preferred when the temperature range allows it.

You have thermocouple effects everywhere, not only on your TC probe but on the TC wire to the solder, to the solder to the pad, to the pad to solder and pin of your chip, and also inside the gold wire bonding of the chip to the die. Basically everywhere 2 different metal touches.

By extent, you cannot extend the wires of a thermocouple, you need the thermocouple with the wire length you need, if you extend it, especially in the region of heat, you introduce additional TC effects that will offset the measurement.

You need to keep your measurement chip and board at ambient temperature to reduce the effects of those. The MAX6675 itself is somewhat compensated in temperature, but not your PCB connections. Put it far from the oven, not in the oven nor the top of the oven.

As the datasheet states:

Optimal performance from the MAX6675 is achieved when the thermocouple cold junction and the MAX6675 are at the same temperature. Avoid placing heat-generating devices or components near the MAX6675 because this may produce cold-junction-related errors

The negative side of the TC probe needs to be grounded.

Also, make sure the TC probe itself is not in electrical contact with anything and not picking up too much EMI.

Reducing Effects of Pick-Up Noise The input amplifier (A1) is a low-noise amplifier designed to enable high-precision input sensing. Keep the thermocouple and connecting wires away from electrical noise sources.

One may guess an electric oven generate quite a bit of noise at 50/60Hz, you can try to shield the cables.

Others:

Thermal Considerations Self-heating degrades the temperature measurement accuracy of the MAX6675 in some applications. The magnitude of the temperature errors depends on the thermal conductivity of the MAX6675 package, the mounting technique, and the effects of airflow. Use a large ground plane to improve the temperature measurement accuracy of the MAX6675

the module you linked has no ground plane.

When using long thermocouple wires, use a twistedpair extension wire.

Use extension wire only at low temperatures and only in regions of small gradients.

Answer 2

The MAX6675 is a bit half-a**ed for any kind of precision work but it's okay enough for reflow solder.

You should put the MAX6675 somewhere that does not change temperature quickly or by much, and run proper thermocouple wire or thermocouple extension leadwire to that area. The goal is to have the 2 junctions between copper and thermocouple material close in temperature to the chip (as isothermal as possible).

Since you're using a base-metal thermocouple the extension leadwire is typically just thermocouple wire with a bit relaxed specification at worst.

Make sure you pay attention to the polarity of the wire or the error will be doubled. That error will be approximately double the temperature difference between the junction to the leadwire and the cold junction sensor on the chip.

Note that color codes vary, for example for type K (Chromel-Alumel) China uses JIS (Japan Industry Standard) color codes so + is red and - is white. The USA uses ISA (now part of an ANSI standard) color codes, so + is yellow and - is red. So if you have wire color coded red & white and a sensor coded yellow and red (or vice versa) and connect red to red you'll have a bad day, and it may not be immediately obvious why because the temperature reading will still increase as it heats, just not as much as expected.