Hi! What is the best way to measure the temperature of a liquid metal?
The most important and difficult condition is that the induction furnace is always on. There will be a lot of noise.

 

why not infrared pyrometer?

 

why not infrared pyrometer?

These are the initial conditions. Type B and S thermocouples will be used.

 

so you already have decided on method. yes, those would be suitable thermocouple types. and since you already know that, what are you asking exactly?

 

so you already have decided on method. yes, those would be suitable thermocouple types. and since you already know that, what are you asking exactly?

which circuit design would be the best for noise removal? The inductor will produce a lot of noise.

 

which circuit design would be the best for noise removal? The inductor will produce a lot of noise.

The inductor will be operating at >100kHz, your signal is <0.1Hz. With a twisted pair thermocouple, all the noise will be common mode. Shouldn't be that difficult to filter. Common mode choke on the thermocouple lead and some capacitors to ground?

 

A good commercial thermocouple display unit should filter out that noise. If you're building a circuit yourself, use a proper thermocouple interface IC with internal cold junction compensation; it will save you a lot of hassle. Analog Devices/Maxim's MAX31855R or MAX31855S, for type R and S thermocouples respectively will give ±4°C accuracy between 700 and 1700°C with an I2C interface to an MCU and a DIP or SOIC package. Microchip's MCP9600 series supports S & R types to 1664 °C and also supports type B thermocouples between 500 and 1800°C, also I2C interface but is in a less convenient VQFN leadless package.

Alternatively, you can integrate directly with a MCU - Microchip's AN844 application note shows how.

 

Ian0
Irving
Thank you for your help! I tested the max31856 and it works fine with the inductor turned off.
I got an overvoltage error when the inductor is on. Test circuit:


Should I add LC filter?

Or is it better to choose a EMI Suppression Filters from Murata's catalog?

 

That does surprise me as the datasheet says:

"Because long thermocouple wires may pick up noise
from a variety of sources, including AC power cables, the
amplified signal is lowpass filtered before being applied
to the ADC. The ADC provides further digital lowpass and
notch filtering to attenuate input noise. The notch frequen-
cies are either 50Hz and its harmonics or 60Hz and its
harmonics, selectable using bit 0 of the Configuration 0
register (00h). In addition, bits D6:4 of the Configuration
1 register (01h) enable an averaging mode that provides
additional filtering with an associated increase in conver-
sion time. 2, 4, 8, or 16 samples may be averaged using
this mode."

How long are your thermocouple wires?

Did you try a resistor in the bias connection as per fig 9 in the datasheet?

Can you put a 'scope on both sides of the thermocouple - 2 channels/x10 - at the MAX inputs to see what the common mode/differential voltages look like with the induction running.

 

How long are your thermocouple wires?

10m

Did you try a resistor in the bias connection as per fig 9 in the datasheet?

No. I will try later.

Can you put a 'scope on both sides of the thermocouple - 2 channels/x10 - at the MAX inputs to see what the common mode/differential voltages look like with the induction running.

Also later

 

An isolated thermocouple and an electromagnetic analog meter display will not be affected by the electrical noise from the heating system.