Temperature Emissivity Experts?

[TimWilborne 108]

One of my customers has some 4-20mA UV temperature sensors that they are having some trouble with. Because of the varying amounts of UV reflected from the materials, they can't accurately read the temperature. I'm looking at a UV temperature sensor that has a selectable emissivity and I'll probably go with one of these. But it got me thinking. What does setting the emissivity do to the 4-20mA coming from the temperature sensor. It has to adjust some curve based off of the materials UV reflectivity. Does anyone know how or if this could be done in the PLC? Thanks TW

[JeffKiper 9]

TW No help with the UV from me but I worked at a shop that used IRCON non-contact temp cameras. We used them to measure 2300 degree forgings. To the best of my memory they made a low temp unit as well. These units were not effected by reflectivity / emissivity of the material or inductive fields.

[TimWilborne 108]

Just glancing over their site, looks like they would have the same problem with varying materials. Do you remember which model?

[JeffKiper 9]

TW I will but in a call and hopefully there will be someone on site sat. If not I will have an model number for you monday. You maybe able to contact BobLfoot he may remember which model we used. Can you talk about what you are doing or what you are going to do this on?

[paulengr 44]

It's IR, not UV. Emissivity is the ratio between a perfect blackbody radiator and a white body radiator...it is a measure of the efficiency of radiation. Unfortunately, it is also nonlinear. Depending on your temperature range, you can also run into a nasty little problem that things like CO2 and O2 and such have selective optical transparency which is wavelength dependent. I have made corrections at the PLC in the past. A pyrometer absorbs radiation, call it j, and converts it to temperature via the following formula: j=e*B*T^4, where B is the Boltzmann constant, e is the emissivity, and T is the temperature in Kelvins. Taking logs, we get: log(j) = log(e*B) + 4*log(T) So if you plug T into the above formula, you can use a log, add/subtract a different "fudge factor", and then exponentiate it back to get your corrected temperature. It's not pretty but it works. Check out this web site for a better introduction to not only pyrometers but a small idea of some of the problems I alluded to. http://www.omega.com/literature/transactio...rmometers1.html Ircon's claim to fame is ratiometric pyrometers. Unfortunately they are way behind the times and they haven't made any improvement to their stuff in years. Land and Williamson are both the current leaders in the field.

[TimWilborne 108]

Paul I've been pulled in several different directions since last week. I'll get back to this as soon as I get caught up Thanks TW

[DanW 41]

>"j=e*B*T^4, where B is the Boltzmann constant" I had thought that was accounted for in the electronics of the modern non-contact IR sensor. I thought most commercial IR sensors, the Ircon, Mikron, Land types did that correction internally, using an 'external; linear adjustment for emissivity. Paul, did you make these math corrections to a 4-20mA conditioned output, or to a signal from a raw thermopile? But to get back to the original issue, non-contact IR sensing provides an average of the temperature in the target area. So if the emissivity varies dramatically within the target area, the readings sill, too. Or is this a situation where poduct A has one emissivity, but when switching to product B, the readings shift because product B has a different emissivity? Dan

[paulengr 44]

The case where I was manipulating the math is when I had to "calibrate" a perfectly good IR sensor to match another measurement. In this case, I was aiming for a known measurement, so I could first take my measured value and calculate j. Then by dividing by B*T^4 where T was given to me, I could calculate a corrected e. After a few measurements like this, I could target a specific e which would give the expected result. Emissivity is essentially an attempt to treat all objects as grey body emitters. They deviate from black body emitters by a constant (0-1), called emissivity. That sounds nice but in reality as you get closer to the temperatures where self-luminescence kicks in, never mind melting temperatures, emissivity starts to do all kinds of strange things and the simplified world of the grey body emitter becomes utterly useless. At best you can sometimes do piecewise linear adjustments to fix the problem. The issue is not only that emissivity is a complex number in reality, but that it is not even a linear relationship. But as long as you stay away from the liquid metals world, "simple" emissivity measurements will suffice.

[TimWilborne 108]

Sorry for the delay. I'm finally getting back around to this one. This is a situation where Part A is has one Emissivity and Part B has another. Part A has a reflectence of almost 100 and an Emissivity of almost 0. I'm using .1 which seems to work ok. Part B has a reflectence of 54 and an Emissivity of .46. I'm still confused on Paul's equations and what would I set the baseline emissivity on the sensor at? Thanks TW

[paulengr 44]

Reflectance is only useful if you are bouncing a signal off of the object. In your particular case, since you have mutliple emissivities, a single setting will never work correctly. You can't get an accurate reading in this case. It is also extremely difficult to get readings off objects with an emissivity of almost zero. Part A sounds exactly like fresh aluminum by the way which is notoriously difficult to deal with in any event. In this particular case, you need to use an IR sensor which reads radiation at multiple wavelengths. Since emissivity is also wavelength-dependent (here is where I said that some materials deviate from the theoretical grey body emitter), a multiple-wavelength detector can accurately deal with this situation. There are a bunch of modes to set up to make this work. In that particular case, I'd strongly recommend you talk to LAND (www.landinst.com) or Williamson (www.williamsonir.com). Both sell multiple wavelength pyrometers specifically for these kinds of problems. I recommend you stay away from Ircon, even though they are the best known name. This is for two reasons. First, the selectivity on Ircon's sensors (the bandwidth of the sensor) is very broad. This makes it very difficult to deal with nonlinear emitters. It is so bad that with iron & steel, it is strongly affected by the amount of CO2 & CO in the air (they work very poorly in furnace situations). It will drive you nuts trying to make them give you consistent results. I've tried to do this when guys are armed with Land handhelds and we're trying to make an Ircon give similar results.

[TimWilborne 108]

Great info! I'm going to go with as sensor that can do multiple wavelengths. Part A is a galvinized material

[paulengr 44]

In that case, the plant manager where I work at worked around a galvanealing line for a few years. He swore by Williamson's for their strip sensors (line scanners) for furnace control.