Quickly Metering Air Temperature

[Dacr0n]

I tried the filament of a lamp and the change is miniscule.

Okay, but at what temperatures did you try it?

Isn't the resistance curve non-linear?

If you did it at around room temp, varied the temp +/- 10 Degrees, and didn't get much change, could it be possible that the ratio of resistance change could be higher at higher temps ?

 

[colin55]

Don't forget the filament goes to 3,000 degrees to produce light and the resistance changes from 60 ohms to 240 ohms.

For 10 degrees, the change is not noticeable.

 

[crutschow]

So if I am going more for accuracy, would going with the tungsten be a better choice since it probably has less mass than a 28AWG wire and respond quicker?

What is the % change per K of tungsten?

I meant that you use one single strand of a 28AWG stranded wire. That's a very fine wire since there are many strands in a stranded wire.

I also stated in my previous post that tungsten has a change of 0.08% per degree C. Copper changes 0.22% per degree C.

Why do you need such fast response? As noted by kchriste the heater likely has a significant thermal lag which would probably exceed the lag in your temperature sensor.

 

[ronp849]

Hi - I have used a very small resistance sensor for controlling a forced airflow to within 0.5C. It was smaller than a pinhead with a nominal resistance of 50k ohms. With a suitable series resistor, the voltage generated by the current through the sensor was fed directly into an A/D input of an Atmel processor. (a Mega16 in my case).

From the A/D value the sensor resistance can be calculated and from the resistance the temperature can be calculated. (floating point math required).

Where are you based - I could drop a sensor in the post to you.

Ron

 

[KMoffett]

Did I miss something along the way? Didn't the OP say:

quickly detect air temperature changes in air flowing...

All this discussion about "wind chill" and heated sensors would relate to measuring air "flow", not air "temperature". Wind chill has to do with the "rate" of cooling, not absolute temp. It can only cool the sensor below ambient if the sensor is wet and evaporative cooling takes place. A bare, fine wire thermocouple will probably have the lowest thermal mass and respond most rapidly to temperature changes. And, they are very accurate.

Ken

 

[Dacr0n]

A bare, fine wire thermocouple will probably have the lowest thermal mass and respond most rapidly to temperature changes. And, they are very accurate.

Ken

I have a bare TC and it is accurate with out a doubt, but its not fast enough...

Also, I am not worried about "flow rate", my example of the airflow meter may be what made you think that, but I am just trying to find the temperature of an airflow, as close to instantly as possible

crutschow:


I suppose I am going to give the 28AWG wire strand a try, but before I do, is it a difficult feat to make a circuit sensitive enough to read the 0.08% change in resistance per degree C of tungsten and use it in part of the control loop?

Or is it simply as colin55 said:

the filament goes to 3,000 degrees, the resistance changes from 60 ohms to 240 ohms.

For 10 degrees, the change is not noticeable.

 

[KMoffett]

I have a bare TC and it is accurate with out a doubt, but its not fast enough...

What kind if response time do you need? I don't know what size of thermocouple you're using, but I don't think you'll get another sensor with much faster response time than a 0.001" diameter, bare wire thermocouple.

**broken link removed**

Ken

 

[Dacr0n]

I'd like a true reading of the air temp in about 1.5 seconds.

Okay... so how would I use that TC wire in the link you posted? I need two different kinds right? Then just make a coil out of it with the two types meeting in the middle or something?

 

[KMoffett]

Even their small surface mount ones with 30Ga (0.010" dia.) show a response of better than 0.3 seconds.

Fast Response Surface Thermocouples with Self-Adhesive Backing

Ken

 

[Dacr0n]

Are those 30Ga (0.010" dia.) (.3 Second response time) TC assemblies much different than this?

Thats the smallest TC I have ever had and its not fast enough. I just think I need something more like a ultra thin coiled wire or something.

With those "Surface Thermocouples", can I remove all that paper or tape or whatever it is from those stick on sensors so that it can just sit in the center of the air flow without restricting it. They're surface sensors originally... which makes me think that they will have to be modded to better meter air temp.

 

[KMoffett]

That's actually pretty big. The smaller the wire, the faster they respond. Though the sense point is at the bead, the thermal mass of the leads add to the response time. Stripping the patch off of the surface mount type should actually improve the response time. If you want professional input, free, Omega has engineers that can help you figure out exactly what you need. And I'm sure there are other companies, but I use their online references a lot.

Ken

 

[ronp849]

One problem with very small sensors that can be overlooked is the thermal effect of the connecting wires - heat can be conducted away from the sensor through these wires giving innaccurate readings. Make sure a good part of the connecting wires are in the airstream as well as the sensor.

Re the sensor I mentioned before - (see pic) - the sensor is shown next to length of standard 0.7mm cored solder. (The blob on the right hand side is the sensor).

 

[crutschow]

crutschow:

I suppose I am going to give the 28AWG wire strand a try, but before I do, is it a difficult feat to make a circuit sensitive enough to read the 0.08% change in resistance per degree C of tungsten and use it in part of the control loop?

If you have a tungsten filament with a room temperature resistance of 60 ohms and you ran 17mA of current through it, the voltage would be about 1V. 0.08% of this is 0.8mV, which is a very small signal. If you amplified that by a gain of 10 with a low-offset op amp, you would get 10V with an 8mV change per degree C. This is still a small signal but possible to work with.

To get a larger signal you could add a (negative) offset to the amp to give near 0V output at room temperature. Then you could perhaps amplify the signal by 50 to get 50mV output per degree.

But in any case you have to be careful about shielding, layout and grounding of the circuit to avoid noise and power line pickup. Such pickup will likely be the biggest problem in working with such a low signal. You will likely need twisted-shielded-pair for the signal and the circuit will need to be in a grounded metal box.

I did a quick calculation with copper and, if you maintain the power dissipation at a level comparable to the tungsten filament (to minimize self-heating), then the signal voltage is comparable. So copper has no advantage as a sensor on that basis.

 

[colin55]

It seems the output for a thermocouple is about 1mV for 10 degrees rise for the temperature you are measuring.

See if you can get a more-sensitive thermocouple.

That's why they use PTC or NTC resistors. They have a sensitivity that can measure 0.1 degree and cost less than 30 cents.

I dont think a termocouple is the way to go.

 

[KMoffett]

It seems the output for a thermocouple is about 1mV for 10 degrees rise for the temperature you are measuring.

colin,

K-type thermocouples @ 30C change 41mV/Δ0.1C

https://www.omega.com/temperature/Z/pdf/z204-206.pdf

You will never find an RTD/thermistor with the response time of very small thermocouple wire. Its just a matter of thermal mass. Accurate measurements with thermocouples take more care in the circuitry than RTDs, but the T/C interface ICs available make that simple.

Ken

 

[Dacr0n]

Its just a matter of thermal mass.

Thats exactly what the main hurdle is here... I need one with the least possible mass that can still give me a fast and accurate T/C reading.


I'm going to have a chat with omega and see if they can help me with what I need.


crutchow:


About that interference.

If I have my 2 TC sensor wires running next to wires that power the heating element VIA the PWM signal.. Is that going to cause problems?

Is there a way to make the 4 wire run a 3 wire run by having some kind of common?

 

[kchriste]

Is there a way to make the 4 wire run a 3 wire run by having some kind of common?

Don't even think about it!
This is especially true with a thermocouple.

 

[KMoffett]

A little more info.

Thermocouple time constant in water:

https://www.omega.com/temperature/Z/pdf/z052.pdf

Thermocouple time constant in air:

**broken link removed**

Ken

 

[ccurtis]

It seems to me, all you need to do is reduce the gauge of the thermcouple wire.

 

[KMoffett]

Was I being too subtle?

Ken