In over my head, but Im learning! Interfacing a RTD1000 to a uC

[rosscopico0]

Hi all,
I am working on a project to fit a rtd pt1000 into the grouphead of an espresso machine and measure with high accuracy (0.1C-0.5C) the water just before it reaches the coffee.

Im planning to use a 4 wire rtd, to minimize switching noise (is this correct?)
I know I require a constant current source for excitation. From what I have read, I require 1mA or less, to prevent self heating errors, I was thinking around 250uA-500uA.

My range of interest is 85C to 95C which corresponds to 1328Ω to 1366Ω, so with such a small difference in resistance, the accuracy of the current source and signal conditioning is quite important.

Using 5Spice, Ive developed a circuit that I think could work, I have attached it below for critique.

The output of the interface cct will go to an Arduino analog 10 bit input.

Just need to figure out the constant current part.....help!!

 

[rosscopico0]

Confused!

Ok, so I think I may have found a cct that will suit my needs.

I was wanting to stay away from ccts with several resistors due to cost restraints.

Here is one that uses only one external resistor. The cct at the top of the page.

It lists a series of chips, depending on the required current range.

Now my question is, how do I determine which one I need?
Im looking for 250-500uA for a RTD PT1000.

 

[MikeMl]

Here is something to try. You don't need a real current source to drive the bridge because you are interested in such a narrow temperature range.

Set up the A/D's voltage reference to be the same 5V Vdd as is shown driving the bridge. If you are stuck using the A/D's own internal voltage reference, buffer it with a voltage follower opamp and use that to drive the top of the bridge. Either of these connections makes the opamp output ratiometric with respect to the A/D full-scale reading, and that makes the circuit insensitive to small fluctuations in the Vdd voltage.

Note the tricky way of of using a voltage source to vary the simulated RTD resistance over the temperature range from 80degC to 100degC. The 100Ω trim pot is used to zero the system at your lowest temperature of interest. The opamp feedback resistor changes the span (gain). Opamp must be a rail-to-rail out in order to recruit the whole A/D range.

 

[crutschow]

Im planning to use a 4 wire rtd, to minimize switching noise (is this correct?)

A four wire connection is used to cancel the effect of the wire resistance between the sensor and the the rest of the circuit, since you have separate wires to carry the sensor excitation current, and separate wires (which carry very little current) to measure the voltage across the sensor. It has nothing to do with (and will not reduce) any switching noise.

 

[rosscopico0]

Thanks Mike!

Can you pls explain what you mean by "Opamp must be a rail-to-rail out"

I think it means, that the gain of the opamp is to be setup so the high end of the range of interest produces as close to +5V going into the ADC, while the low end of the range of interest produces as close to oV going into the ADC.

Is this correct?

Also, should I be using a 2 wire, a 3 wire or a 4 wire RTD?

 

[MikeMl]

The opamp is powered from 0V and 5V. The A/D input range is 0v to 5V. Only a modern CMOS rail-to-rail opamp will drive its output pin very close to 0V at one end and very close to 5V at the other. Go to the TI website. They have a selection tool where you can search for rail-to-rail opamps which will run on a single 5V supply.

The advantage of using a 4wire RTD over a 2wire one is if the RTD sensor is a long way from the circuit. If you can mount the circuit within a few inches, and your accuracy requirement is ~ 0.2degC, then you can get by with a two wire one...

 

[rosscopico0]

Change of sensor

Ive been talking to someone, apparently I would have faster response times by using a NTC thermistor.
Can I still use it in a bridge? Im assuming I connect it differently since it is a negative temperature coefficient as opposed to the rtd's positive temp coefficient?

 

[Ubergeek63]

Ive been talking to someone, apparently I would have faster response times by using a NTC thermistor.
Can I still use it in a bridge? Im assuming I connect it differently since it is a negative temperature coefficient as opposed to the rtd's positive temp coefficient?

i am afraid your "someone" is talking out his backside.

thermal response time is primarily determined by thermal mass, it has NOTHING to do with the element material or type. platinum does not generally do well in a bridge, but then platinum is generally only used for high temperatures.

in a full sensor, you need to add the mass and thermal resistance of the structure encasing the element.

 

[KeepItSimpleStupid]

RTD's have good accuracy specs. I used them in an envionmental chamber that went from -80C to 200 C. Thermisters typically have lousy accuracy specs. A few tenths of a degree are common. Thermocouples do to. Type T is generally used from cryogenic to room temperatures.

The thermal mass of the sensor and the thermal resistance enters in all of the applications.

 

[Ubergeek63]

Hi all,
I am working on a project to fit a rtd pt1000 into the grouphead of an espresso machine and measure with high accuracy (0.1C-0.5C) the water just before it reaches the coffee.

Im planning to use a 4 wire rtd, to minimize switching noise (is this correct?)
I know I require a constant current source for excitation. From what I have read, I require 1mA or less, to prevent self heating errors, I was thinking around 250uA-500uA.

My range of interest is 85C to 95C which corresponds to 1328Ω to 1366Ω, so with such a small difference in resistance, the accuracy of the current source and signal conditioning is quite important.

Using 5Spice, Ive developed a circuit that I think could work, I have attached it below for critique.

The output of the interface cct will go to an Arduino analog 10 bit input.

Just need to figure out the constant current part.....help!!

wait a minute! 10C and you are worried about accuracy? use a thermistor, calculate the slope and calibrate one point.

 

[aljamri]

my range of interest is 85c to 95c which corresponds to 1328Ω to 1366Ω, so with such a small difference in resistance, the accuracy of the current source and signal conditioning is quite important.

85C= 132.8 Ohm
95C= 136.6 Ohm

 

[rosscopico0]

85C= 132.8 Ohm
95C= 136.6 Ohm

I was planning on using a PT1000, so 85C=1328 Ohm and 95=1366Ohm