looking for thermocouple amplifiers

ryuji

2[H]4U
Joined
Jun 11, 2004
Messages
3,115
preferably one that has internal linearization and cold junction compensation

working on designing/building a fan controller, and been looking at DS18B20's for commodity sensors to place wherever, but i would like to measure water fluid temperature too, and thermocouples would be the most precise way to do it, as i dont need to worry about sealing/isolating as much as sticking a DS18B20 in there

reason why i need water temps is my controller is going to control pump speed too ;)

only ones i've found so far that look nice are from analog devices and only handles up to 50C(might want to watch cpu temp, etc too)
 
Last edited:
Thermocouples can be a pain to work with, especially if you're directly measuring water temperature (think galvanic corrosion for non-insulated ones). About the only advantage would be the very small size. I've used small thermisistors in the past - all you have to do is drill a small (about 3/32" depending on your thermisistor size) hole in the metal tubing/fitting and JB weld it in. They can be calibrated using ice water and boiling water, but generally I've found them to already be very consistent. I'm assuming you're using a microcontroller due to looking at the DS18B20s, adding an individual correction for each thermisistor would be trivial. If you're adamant about thermocouples, I'd go with a bridge and a low noise op-amp to give you better resolution and feed an ADC to your uC.
 
i was somewhat planning on using a stainless steel tube to insert the thermocouple into and then use arctic silver epoxy to seal it all in, but sounds like the gain in accuracy is too minimal to be even worth assembing the opamp circuit or paying for a IC solution. so your thinking using ds18b20's for all my temperature sensing, which i guess works for me
 
ds18b20 =/= thermocouple =/= thermisistor

They are separate technologies. Silicon based temperature sensors work on the basis of differing current flow through a P-N junction based on temperature. Thermocouples use the voltage generated by a dissimilar metal junction that varies based on temperature. Thermisistors are resistors which vary in value based on temperature. Each has advantages and disadvantages in size, cost, accuracy, consistency, linearity, output format (digital vs analog), temperature range, etc.

How are you planning on reading and using the temperature? Are you going to use a microcontroller? Use USB or serial ports and have the computer do the control? Or is this to be stand-alone with a display and independent analog controller? The best type of sensor will be based on this.

IC based sensors are very accurate and give either digital or analog outputs, but are more expensive and physically bigger.
Here are some IC based temperature sensors:
National Semiconductor page on IC temperature sensors:
http://www.national.com/en/tempsensors/index.html
TI page:
http://focus.ti.com/paramsearch/doc...log&familyId=611&uiTemplateId=NODE_STRY_PGE_T
Maxim IC:
http://www.maxim-ic.com/appnotes10.cfm/ac_pk/24
Analog Devices:
http://www.analog.com/en/mems-sensors/analog-temperature-sensors/products/index.html

Thermocouples can be built with a simple pair of dissimilar metal wires, there are many types and they can be made for cheap. I used them alot in wind tunnel experiments years ago, but they were a bit of a pain to get the automated data collection to work right with late 80's PCs.

Thermisistors are cheap and relatively accurate (do you really need 0.065 degree resolution?) and interface easily with analog circuitry.
 
Last edited:
I'm using a microcontroller, so a ds18b20 is fine to interface with, I was considering thermocouples as they can be very flat and small
 
Back
Top