Sciencemadness Discussion Board - Measuring very high temperature

Measuring very high temperature

yobbo II - 7-11-2016 at 16:11

Measuring very high temperature

The device MLX90614 shown at link

https://www.sparkfun.com/products/9570

can measure temperature up to 350 or so. Is there a way this could be extended by putting black glass in from on the sensor or some such.

y

BromicAcid - 8-11-2016 at 04:32

Interesting question, it makes sense that if you block a consistent amount of IR reaching the sensor that you will extend the range, but it would still fall to you to recalibrate the sensor. Your glass would also have to be transpartent to IR and I would worry about whatever does get reflected by the glass messing with the reading. Hold on.... a quick trial with a glass in my kitchen and my finger located behind it didn't find my finger, just read the temperature of the glass. Anyway, that would be my concern. But I don't have any experience in the area, just trying to look at it logically.

However, 350C wouldn't normally be considered a very high temperature, you wouldn't have an issue finding a thermoprobe that would go that high. What temperature do you hope to be able to read by blocking a portion of the infrared?

Sulaiman - 8-11-2016 at 07:18

the temperature reading is based on IR power received,
power is proportional to T⁴
and also angle of acceptance cone.
So a small hole in Al foil at fixed distance should work as an achromatic attenuator,
how to callibrate ?

EDIT: AREA of acceptance cone, not angle.

no, I think that (solid angle)² is more like it
who cares ?,

just try a small hole at a fixed distance.

[Edited on 8-11-2016 by Sulaiman]

[Edited on 8-11-2016 by Sulaiman]

yobbo II - 8-11-2016 at 16:45

Thanks for reply. I was thinking of hacking together a high temperature measuring pyrometer but it's probably just wishful thinking.

How do these things work.
http://www.ebay.com/itm/Pyro-X100-Degrees-C-Micro-Optical-Py...

Is there a bit missing (the optical 'sense head')

There is a good one here

http://www.ebay.co.uk/itm/High-Temperature-Infrared-Laser-Th...

Goes all the way up to 2200 C ! (That's my boy)

There is a high/low alarm on it which may make it possible to hack as a contoller of sorts. Its a pity they do not have an electrical output for temperature measured.

Thanks for your time.

Sulaiman - 8-11-2016 at 17:04

infering the temperature of objects based on i.r. power is much easier for higher temperatures than those near ambient.
https://en.wikipedia.org/wiki/Stefan%E2%80%93Boltzmann_law

yobbo II - 9-11-2016 at 11:05

Attached is an old method (1921) for measuring temperature in a furnace. Water is passed through a pipe at a fixed known rate (fast enough so that it did not boil ) and the increase in temperature of the water indicated the temperature. A bit crude I suppose but would do the job in the outback.

Attachment: designtestofelec00naim.pdf (2.5MB)
This file has been downloaded 614 times

aga - 9-11-2016 at 11:31

A simple method is to place lumps of known metals at known points inside the furnace.

As they each melt, you know you reached that particular temperature.

Not much use in a small furnace, or to see what C is in the pot i guess.

[Edited on 9-11-2016 by aga]

Sulaiman - 9-11-2016 at 21:53

in the early '70's we used optical pyrometers,
you look at the heat source with a tunsten filament lamp in the field of view,
turning a dial both indicated and changed the filament temperature
when the filament was at the same temperature as the viewed heat source, it became invisible, indicating temperature.
Callibration required

nux vomica - 10-11-2016 at 02:46

Here is a link for a homemade pyrometer. Cheers Nux.

http://www.metallab.net/pyrometer.php

Sulaiman - 10-11-2016 at 04:36

I only mentioned the optical pyrometer out of interest,
I would not recommend the construction of one other than as a hobby project,
thermocouples are cheap and accurate enough for most purposes,
i.r. thermometers ... I guess that you get what you pay for.

if you do use an i.r. power-sensing thermometer,
choose the spot/distance ratio wisely.
I believe the indication is a summing (averaging) over the area,
so wider cones need relatively closer proximity.

wg48 - 10-11-2016 at 04:58

Calibration using the melting point of relatively inexpensive metals (not Pt) is complicated by the surface oxide that forms at high temperatures. Aluminium, magnesium nickel and chromium for example. Pure iron may be suitable because its oxides melt at a lower temperature than the metal. T shaped test pieces may reduce the oxide effect. Perhaps compounds could be used or special temperature cones with sharp and time independent melting points are available.

A diy hot wire pyrometer is probably the optimum on a cost and calibration basis and very usfull for measuring your element temperature or you will have to be very conservative in its loading. Accurate calculation of the element loading is difficult. Apparently the change in resistance of the pyromters filament with temperature can be used for its calibration.

But don't forget a Pt or TC sensor (1800C) can be bought for a round £20 and used with a controller. Though its difficult to use them for accurate element temperature.

[Edited on 10-11-2016 by wg48]

yobbo II - 10-11-2016 at 11:00

Where can you obtain Pt/Rh thermocouples to go to 1800 C for around 20 pounds. There is a link to one here. It goes to 1600 C (according to the seller).
http://www.ebay.com/itm/WRP-100-225mm-probe-S-type-platinum-...

How long will it last at 1600 ???

'Going to' is not so simple. The ageing of the TC comes into it and theres the length of time it will last (expense). The TC lasts longer if the wire is thick. The ones on ebay are very very thin wire. The TC wire (Pt and Rh/Pt) sold on ebay is also very very thin and costs a fortune PER INCH

.

Reading from here

http://www.technology.matthey.com/article/41/2/81-85/

"The six new thermocouples, Number 3 having a diameter of 0.5 mm, and the others having a diameter of 0.38 mm, were used for 45 days (1080 hours). One of them developed a dark area about 300 mm from the cold end due to a crack on the protecting sleeve. The other five thermocouples continued to work well and retained a clean surface.Compared with the conventional, 0.5 mm diameter, Pt+Pt\Rh thermocouple, the service life of the new one is 1.5 to 2 times longer and can withstand contamination."

There are talking about thermocouples being used in the steel industry in protective tubes. The wires being used are in the region of 0.38 to 0.5mm (I think it is the wires they are talking about, or is it) . If you were to purchase thermocouple wire that thick (assuming you can get it) it will cost you a fortune for 6 (x2) inches or so.
Also the temperature that the steel mill they are talking about are not too high at around 1350 C. At 16 or 18 hundred C the thermocouples will be lasting much less time.
When you read of temperature limits of certain thermcouples there is an assumption that they are in an inert atmosphere.

The pyrometers can be had on ebay like this one.

http://www.ebay.co.uk/itm/Tung-Sol-Ircon-Model-710C-Vintage-...

and

http://www.ebay.co.uk/itm/VINTAGE-INDUSTRIAL-FOSTER-OPTICAL-...

Unfortunately you cannot get them to control temperature, just observe/measure.

Anyone any suggestions?

If you were to have two optical paths and some sort of cheap 'colour sensor'. One path looks at the furnace and one path looks at the filiment in the bulb (which is set to a certain temperature that you want achieve.). Both paths are sampled at say 100Hz (rotating mirror or somesuch), the ac signal obtained would be an indication of temperature difference. No signal = same temperature. Phase would (I suppose) have to be taken into consideration as when the furnace was above the filiment temperature (as opposed to below) the phase of the signal would be inverted.

Another project, another pipe dream........

[Edited on 10-11-2016 by yobbo II]

wg48 - 10-11-2016 at 16:55

yobbo II:

You can get the S type here

www.omega.co.uk/S_Type_Thermocouples‎
0800 488488
Huge range of type S Thermocouples. From £20. UK Manufactured.

Though the "from" part may means twice that price LOL. I was assuming S type go to 1800C but it depends on the housing.

If the manufacture says 1600C as you say that usually means for a very limited time and under the best conditions which may be difficult to obtain from an ebay seller. Hopefully Omega has a real specs for their thermocouples.

yobbo II - 10-11-2016 at 17:35

Thanks for reply.

The link does not work for me.

yobbo II - 27-11-2016 at 16:51

Can anyone shed any light on this item.

http://www.ebay.co.uk/itm/NEW-IRCON-DNS30-20C-DNS30-20C-HO-B...

There is some more info on this item here

http://www.prc68.com/I/EOG.shtml#Ircon

What is in the device from an electrical point of view. It it a diode (+ some filters and windown and some perhaps prisims)
How would I wire up the device seeing that I am not going to purchase the controller. I would hope to just connect up an 'controller' of my own. Thats just something to measrue the diode voltage (if it is just a diode).
Thanks,

There is more info attached.

Attachment: 9260115_ENG_F_W.PDF (7.8MB)
This file has been downloaded 2143 times

wg48 - 28-11-2016 at 05:17

Yobbo: The information you provided states the detector is a silicon diode, It also says the assemble has a 0.7 to 1.0 um spectral sensitivity which strongly suggests it has a filter to reduce the response to visible light.

A typical configuration for using a silicon detector diode is to use it in current mode ie measure the current it produces at close to zero bias voltage, That would make the transfer function (signal out versus photons in) more linear than voltage mode and reduce offset caused by dark current.

I am guessing from my experience I would expect the signal from the emitted visible light or near infrared red (0.7-1.0um) of an object above 1000C to be easily measured from less than a meter away with say a f8 lens.

The assemble probably uses near the infrared band to reduce interference from ambient visible light . That is probably not essential if you looking into a furnace and the optics and detector are shaded from ambient light. That has the advantage that it would produce a larger signal. In any case the signal will need to be amplified for connection to a controller. The amplification could be achieved with a single opamp.

The major problems will be the very non linear signal wrt temperature and calibration. Perhaps if accuretly knowing the temperature is less important than maintaning the temperature at some value it may an ideal method.

careysub - 28-11-2016 at 08:23

In general infrared temperature sensors are expected to perform over a wide temperature range, and to get accurate measurements on a wide range of materials compare two different wavelengths to estimate the black body curve. This is fairly fancy.

If you just want to measure the very high temperature range above what a thermocouple can go (1260 C for the standard Type K, if suitably packaged; or 1600-1800 C for Pt/Rh thermocouples) a simpler sensor system might work just for that higher-than-thermocouple range (you would need to remove your thermocouple before entering that regime).

Temperature color charts describe 1500 C (and above) as "dazzling white", so a simple optical sensor that measures the intensity at this high temperature range, using straightforward black body computations (plug the measurement into a calculator of some sort), calibrated by your high temperature thermocouple at the upper end of its range, should be pretty good. It only needs to be sensitive enough, and proportional enough in the range where you can't use a thermocouple. A band filter would probably help with the accuracy, but something that measures total brightness might work as well.

At 1800 C the peak emission wavelength is still in the infrared, but so much light is being emitted at these temperatures (100 watts per square centimeter) that any wavelength band you like could be used.

Depending on the set-up you may not need any optics at all beyond the sensor and filter, but a simple lens might be useful - you do not need to form an image to measure the band brightness.

[Edited on 28-11-2016 by careysub]

wg48 - 1-12-2016 at 16:37

careysub:

Yes a lens is not essential but it can reduce the required size of the hole in the furnace.

Those 100w/cm of radiance will be mWs or less when you have a 1mm square detector with a limited spectral response one meter away and assuming your not opening the door to view the furnace LOL.

I had a quick look at the numbers for a £2 detector, its low sensitivity was a problem. The price jumped to the £40s for significantly better detector. I may return to it some time but as that cost is comparable to a high temperature thermocouple I lost interest.

careysub - 2-12-2016 at 08:43

One meter seems excessive. If you mount the sensor on the furnace enclosure why would it be greater than 30 cm or so, where the intensity would be ten times greater?

Seems to me that between using an inexpensive quartz window ($10.50 with shipping on eBay right now), and air cooling, and a lens to focus the light on the sensor getting the sensor close enough for even a cheap one to readout should not be a problem.

I was actually proposing this as a method for measuring temperatures which would extend beyond the range of any thermocouple - above 1800 C (very high temperatures).

[Edited on 2-12-2016 by careysub]

yobbo II - 2-12-2016 at 16:04

http://www.ebay.com/itm/Tung-Sol-Ircon-Model-710C-Vintage-Py...

Above is a vintage non contact pyrometer from 1971 no less.

wg48 - 3-12-2016 at 04:08

Quote: Originally posted by careysub

One meter seems excessive. If you mount the sensor on the furnace enclosure why would it be greater than 30 cm or so, where the intensity would be ten times greater?

Seems to me that between using an inexpensive quartz window ($10.50 with shipping on eBay right now), and air cooling, and a lens to focus the light on the sensor getting the sensor close enough for even a cheap one to readout should not be a problem.

I was actually proposing this as a method for measuring temperatures which would extend beyond the range of any thermocouple - above 1800 C (very high temperatures).

[Edited on 2-12-2016 by careysub]

It was just a starting point with round numbers. I imagined a detector viewing a 10mm x 10mm area of the inside of the furnace thru a spy hole of similar dimensions with detector mounted in 0.5m tube with an aperture at the end such that it could only see the 10mm x10mm area of the furnace 0.5m from the end of the tube. So the solid angle would be 1msr.

I assumed your 100W/cm was a radiance into a solid angle of 1sr not 2 x pi sr and that 10% of the 100W was detectable. I assumed 1mA/mW/cm for the detector but I now think I may have used that incorrectly.

3DTOPO - 3-12-2016 at 20:16

Quote: Originally posted by yobbo II

Measuring very high temperature

Is there a way this could be extended by putting black glass in from on the sensor or some such.

y

I have this non-contact IR thermometer (up to 1050C supported):

http://amzn.to/2gTQE9u

I found that putting various welding glass windows in front of the sensor aperture lowers the readings accordingly. Trick is to calibrate the altered readings with a known point (like melting temperature of copper or something) - but I haven't done that part yet.

careysub - 4-12-2016 at 10:47

Quote: Originally posted by wg48

Quote: Originally posted by careysub

Lets see if I am doing this right.
According to this calculator:
https://www.sensiac.org/external/resources/calculators/infra...
The radiance from a 2073 K (1800 C) blackbody is 33.3 W/Sr-cm^2.
A 1 cm^2 circular window at 30 cm subtends 0.01 Sr.
That would mean the irradiance is 0.33 W/cm^2 (I think), or 330 mW/cm^2 to use more typical photosensor units.
Comparing it to the characteristics of this cheap photodiode:
http://www.mouser.com/ds/2/427/bpw34-240374.pdf
Indicates that it is not only easily detectable, it is more than an order of magnitude too bright, saturating the detector.

A common sense way of estimating whether this should work is this: a 100 W light bulb radiates light in all directions. Just about any sensor placed 30 cm (or 100 cm) away from the light bulb can detect that light trivially. The furnace window puts out that same amount of optical energy, put in a unidirectional - far more focused - manner.

The spectral characteristics of the light bulb and the 1800 C furnace aren't far different, a tungsten filament's max temperature is 2550 C, and can be closer to 2000 C depending on set-up (use of a dimmer, etc.).

I imagine the set-up like so:

You have an insulating block in the furnace side with 1 cm^2 hole bored through it that can be raised and lowered to block off the hole when measurements are not being made (reducing heat loss, and unwanted heating of the sensor).

The sensor is mounted on the outside of the furnace, with quartz window, looking into the hole.

A small fan blows air over the sensor to keep it cool while taking data.

An IR cut filter at about 1 micron might be used to block the IR energy that the photodiode cannot detect so reduce the heat load, but based on my estimate above a filter arrangement would be necessary anyway to reduce the intensity to something manageable. Of course a stand-off tube such as you proposed would be a good solution to reduce intensity and heating problems.

[Edited on 4-12-2016 by careysub]

wg48 - 4-12-2016 at 14:20

careysub:

Looking at your calculation and mine I noticed some errors, My detector subtends an angle 0.1msr not 1.0. While yours being approximately 3 times closer (3^2) so about 1msr not 0.01sr

You assumed all wavelength would be detected. I estimated 2 w with a flat sensitivity from 0.7um to 1um and zero at other wavelengths.

So I think your about 2mw?cm^2. from the spec of diode that’s about 100uA versus a worst case dark current of 20nA (@25C). Lets say thats 0-200nA due to temperature. that’s an error of 0.2%. So yes your correct it doable from the view point.

Your radiance calculator is useful It also gives the change in radiance with temperature. At 220K its 0.0660W/K that’s 3% significantly larger than the 0-200nA varition of the dark current. Sensitivity of the detector also changes with temperature which also should be considered.

Note also that the radiation from the spy hole is not focused. If the furnace was at the same temperature as the filament of a light bulb and both have the same emissivity they will have same irradiance.

I can imagine building a small hermetic +2000C furnace with a hot volume of say 200ml using carbon and or MgO with carbon or tungsten (welding rods) heating elements. Perhaps heated with microwaves or induction, it would be fun and it would need a means of measuring the temperature. However its not on my must do list.

yobbo II - 4-12-2016 at 16:40

If one were to set up a (cheap) mass flow controller

http://www.ebay.co.uk/itm/MKS-1179A-hi-vacuum-gas-mass-flow-...

to control a given quantity of gas (air) through a short pipe exposed to the furnace interior and measure temp. in and temp. out you might get a usable system. The air rate would have to be large enough to stop the pipe carrying the gas from melting. A thin fuzed quartz pipe bent into a U shape and put sticking into the oven. It's the same idea as the attached document (from 1921) above only he used water and his furnace was a lot bigger.
The thing would need calibration. It would measure very high temperatures I suppose.

How much of a temperature increase would you get from a gas (air) travelling in a 2 inch long pipe (say 2 inches of U shape exposed to inside of furnace), a few mm in dia. and having air flow at 5 sln (centi litres per minute).
Where would you even start.......
Yob

careysub - 4-12-2016 at 22:39

Quote: Originally posted by yobbo II

If one were to set up a (cheap) mass flow controller

http://www.ebay.co.uk/itm/MKS-1179A-hi-vacuum-gas-mass-flow-...

to control a given quantity of gas (air) through a short pipe exposed to the furnace interior and measure temp. in and temp. out you might get a usable system. The air rate would have to be large enough to stop the pipe carrying the gas from melting.

If you are using an inert gas atmosphere (which I consider a near-necessity for a general purpose ultra-hot oven) then you could use molybdenum tubing which is readily available and not too expensive. Mo melts at 2623 C.