Thermocouple - MAX6675 vs AD595?

Thermocouple - MAX6675 vs AD595?

What are the advantages of each of these chips?

From what I've read I'm still not sure what the differences inf unction/performace might be, maybe someone could clue me in.

What I do know is: The MAX6675 uses SPI. I have never used SPI, but it looks like it would be handy if you don't want to use an arduino. I can't see why you would want to use SPI if you have an arduino. Looks like it takes up a few more pins than the AD595. Although, you can buy this handy board from Ryahn Mclaughlin's site and the terminal and everything is put together there.

The AD595 is nice because it's DIP so you can stick it right in the breadboard. But it costs 6 bucks more than MAX6675, and if you buy the thermocouple off Sparkfun you also need to buy a terminal block for your cold junction there. All in all price is probably not a huge factor in the decision here.

Let me know if I'm mistaken in any of these judgements. Peace, Josh

Unless you need the high temperature range that thermocouples can have, there are certainly more accurate, cheaper, and easier temperature sensors one can use. RTDs, thermistors, semiconductor based, etc.


One big difference is that the MAX6675 (and 6674) have built-in 12 bit ADCs. The AD595 is an analog device, so you either use the Arduino-s built-in 10 bit ADC and get 1/4 of the resolution of the Maxim devices, or you have to add your own 12 bit ADC to the AD595 circuit.


I have used both.

Ryan's option is superb and works great with the Arduino. I use it with a 2000 degree F kiln, so type K hi-temp thermocouple is needed. Perfect match.

I much prefer the SPI approach.

Just general differences, thermocouples are more exchangable and consistence. Doped semiconductors (thermistors) are less exchangable, meaning you replace it with another one, their readings may differ by 1DegC or more, even for good ones I buy for physics research. Thermocouples don't. If a German researcher says he got something at 120DegC with a thermocouple, I should get that too at 120DegC.

Being consistent is reporting same value over long period of time without drifting. Semiconductors break down at high temperatures so they don't read the same temperature several years after you first install them.

But in order for a thermocouple to be accurate, you need a reference temperature, we do this by using a fridge of ice+water mixture so the temperature is 0.01DegC. I don't know what reference the chip uses, maybe just another semiconductor.

In fact, each SPI peripheral needs an independent Chip Enable pin, so for each SPI peripheral that you add you must provided another CE pin.