Chip Temperature Measurement 1

[mikelum]

Hi,

We have been measuring the surface temperature of a problem chip using a thermocouple and an IR gun but are getting temperatures that are approximately 30 deg C different, the IR gun providing the higher reading. Other companies have been measuring this chip with a thermocouple and the results always conflict with the IR gun.

Anyone got any ideas why?

Thanks,
Mike

 

[Skogsgurra]

The "Stem Effect". The thermocouple wires lead heat away from the thermocouple. Always use non-contacting (i.e. IR, mostly) measurement for this. Or forward diode voltage drop internally in the chip - it is often a very reliable way of doing it. Provided there is a free diode available.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[MetalworkerMike]

IR guns can be fooled due to 'reflected heat' and give an inaccurate result. That's what the guy from Flir told me, anyway. I tend to trust thermocouples more than I trust IR guns.

Mike

 

[Skogsgurra]

I do not think that you are measuring a little chip, weighing around 1 gram with a thermocouple. Do you really? Also, the reflected temperature is mostly a problem on shiny metal surfaces. No chip I know about has that property.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[mikelum]

Thanks guys for your comments.

I tend to agree that we don't have a reflection problem and there is no heat source to provide a problematic reflection. There is conformal coating on the chip however which is more reflective than the chip on it's own, I think I'll scrape some off as a test to see if that makes any difference.

Gunnur, we are measuring a fairly small chip, the chip casing is approx. 10mm x 6mm. Is this is size of chip not a suitable for using a thermocouple on?


Mike

 

[VEBill]

If you need a third vote to break the deadlock, there are stick-on decals that will record the peak temperature to within whatever step size you select. Of course, the decal blocks some heat flow and also affects the reading slightly - that darn Heisenberg is everywhere.

In case this applies in this case, generally when people start to argue about exactly how hot the chip is, the real answer is that it's simply too hot. That's a common organizational behaviour.

 

[melone]

IR guns can be yield different results based on many factors: package size, angle, material / color of the surface being measured (black mold cap will yield a different result than shiny metal), smooth vs rough, etc., can contribute to conflicting data.

IR camera's are great for monitoring an entire system, and will result with relative temperature differences between different parts of the system.

Thermocouples, TC's, are great for monitoring a few pin point locations in a system. TC's often result in a more accurate measurement. However, the obvious drawback with a TC's are that you need a TC for every location you want to monitor. Also, it is often very difficult to deduce the temperature in other portions of the system based on the data from a few discrete locations.

 

[IRstuff]

The measurements can be misleading, depending on the conditions. You state the usage of conformal coat, but not whether the thermocouple was on top of the conformal coat. This would then be measuring a conglomerate of the air temperature and the surface temperature of the coating, neither of which are the temperature of the package.

The conformal coat might be sufficiently transparent in the IR to get you a reading that's closer to the desired measurement, but it might not.

At the very minimum, you definitely need to remove the conformal coat and do both measurements again.

TTFN

FAQ731-376

 

[MetalworkerMike]

I have little faith in IR guns as anything more than a ballpark reading. Take a piece of pipe with paint on part of the surface, and you will get different temperatures on the threads, the plain surface, and the painted surface. By different I mean that I routinely see differences of 10 degrees or more. They're as handy as all get-out, but they can stab you in the back if you're not careful. Oh, and before the question is asked, my IR gun is a RayTek, which I consider to be one of the most trusted brands of IR guns.

Mike

 

[itsmoked]

I have very, very, good results with IR guns if, IF(!), I respect the fact that they are looking at light and reflections will affect them. Shiny metal is going to give you offset readings. (always) The higher end units like the one on my desk will allow you to change the emissivity setting to get a correct reading off of any particular metal.

So, if you have stainless steel pipes with molten chocolate running thru them all the time and you need to keep track of them, you can measure the temperature with contact then adjust the gun so that you can now walk thru the plant and correctly read your chocolate pipes.

If you have oxidized Al or anodized AL you should have no problem. Same with metal that has a dull finish.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[MetalworkerMike]

Itsmoked: probably my biggest problems with these guns has to do with condensation on the pipes. That sends the reading way off the fairway and into the long grass. It's a pain to keep wiping pipes to get a reading from them (and defeats the purpose of a non-contact reading). Still, handy devices particularly if you're using them on the same material all the time.

Mike

 

[itsmoked]

Interesting observation Metal. I will have to check that out. Must be the reflection ability of the droplets directing other temperature surfaces towards the receiver.

Sounds like you could use a thermal imagining camera. They work differently than the less expensive IR guns. I went to a cardboard factory with a salesman and when we were done the salesman whipped out a small handheld imager and showed it to the cardboard guy. He stuck his hand on the whiteboard for about 2 seconds then pointed the imager at the whiteboard. You could clearly see the hand print on the whiteboard in bright purple. You could then compare the color to a temperature color bar down the side, or you could move a cursor over the print and get a precise temperature value. He could also print out a color picture of any scene. It was pretty neat!

It was a Raytek of some sort.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[VEBill]

An IR Gun could probably take a reading via a mirror. On the other hand, it would have major problems measuring the temperature of a mirror (at least the front surface).

 

[IRstuff]

Water is very absorptive in the IR. That makes the IR reading of the water temperature itself, not the stuff underneath.

Most of the imaging IR sensors would have the same problem with water, since that's a property of the water itself.

There are some minor differences depending on the exact wavelength of the sensor.

Note also, that a measurement that's non-perpendicular to the surface will aggravate any anomalies with emissivity as well as reflections, particularly if the IR imager uses 3-5 um wavelengths. The Brewster angle of water in the 3-5 is sufficient high enough that serious glints from off-axis sources can show up.

TTFN

FAQ731-376

 

[analogkid2digitalman]

Does the IR gun have a means to adjust the calibration based on the emmissivity of the object you are measuring?
Here are some values for different materials:

http://www.monarchserver.com/TableofEmissivity.pdf

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"It's the questions that drive us"
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

 

[itsmoked]

Nice find!

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com