measure 1300F to 1700F with 10F accuracy

There are off the shelf controllers which will do that. Eurotherm 91e Should do it. IIR goes to 1000C

What does it generate, mV? First thing is to scale it to the uC operating range (usually, 5V) with an op amp.

You know they make heat controllers and monitors only just for that ready built and accurate with your specs? but then again I guess you maybe trying to experiment with this your self?

btw, if you want simple go here

You can eliminate CJC, but there will be approximately a 1 degree error for every degree of change in instrument temperature (including self-heating), which kind of eats into your 10°F error budget pretty quickly, especially since 2°-4°F is used up by your sensor.

Best regards, Spehro Pefhany

check out the MAX6675

thermocouple to digital output

At first glance, it looks really mediocre. Are they serious about the equation at the bottom of sheet 4?

John

maybe they are assuming the device is being used in a very small temperature range where a type k would be very linear?

Yes, it is a little simplified, but if you plot in degrees C with a fat Sharpie it's hardly noticable...

Best regards, Spehro Pefhany

According to the data sheet, the temperature error in the 700-1000 deg C range is +/- 17 LSBs, or +/- 7.7 deg F, which is already more than required.

In addition there's a +/- 3 deg C error in the cold junction temperature measurement.

Also, the cold junction measurement is done on the die, not at the actual junction, so there may be an error there too.

Those are just the measurement errors introduced by this chip which come on top of the inaccuracy already present in the thermocouple itself.

Nominally, Maxim only *advertises* neat stuff, they never have it available unless you are willing to buy a few $100K and wait for them to dig a new sand mine...