Practical accuracy of IR thermometers on PCBs

[mich41]

I'm thinking about getting an IR thermometer to measure electronics, mainly chips and PCBs.

I suppose plastic chips shouldn't be an issue with emissivity close to 1, but how much error should I expect on those shiny soldermasks?

[wraper]

On the soldermask it will be more or less fine. However on metal parts, especially non anodized heat-sinks, ir will be completely useless. Also IR thermometer is not very useful tool compared to the thermal imager. The biggest issue is that you never know if it shows you some bullshit, like reflected IR and similar. As laser aim usually is not precise at all, you won't be completely sure what part you actually are measuring.

[ez24]

Since the OP does not indicate their country - I will say that I got one from Harbor Freight in the US.  I have found it useful for about $30, but I am sure it is not as useful as an thermal imager for $300 or more.  You can move the spot around to find what you think is correct.  Also you can get some flat black paint or dark nail polish and put on the object to reduce reflections.

Try the corners of the components, but anyway you can learn how to move the IR spot around.  Well worth the money.

If I were a millionaire, I would have a nice thermal imager, but for now my cheap IR will have to do. 

[Gyro]

Don't rule out getting a cheap ebay thermocouple thermometer, and a roll of Kapton tape. They tend to be useful in situations that IR thermometers aren't, eg. shiny heatsinks. They can also be used to give you an indication of how far off the IR thermometer is on a particular surface finish.

[mich41]

Thanks for the responses. Thermocouple is exactly what I've used so far, but it's slow, conductive and a bit of PITA to use at times.

I know that IR doesn't (easily) work on metals, but if it can quickly measure a random spot on PCB without requiring some insulation or risking shorting stuff and without >10°C errors, I'll be all happy

As for thermal imagers, nice toy, but I don't need it yet.

[ez24]

Don't rule out getting a cheap ebay thermocouple thermometer, and a roll of Kapton tape. They tend to be useful in situations that IR thermometers aren't, eg. shiny heatsinks. They can also be used to give you an indication of how far off the IR thermometer is on a particular surface finish.

thanks I just added a roll to my cart.  I am always in the market for new tapes and adhesives.

http://www.amazon.com/Mil-Kapton-Tape-Polyimide-Yds/dp/B006ZFQNT6

But how can this work for a IR meter since it looks like it has shinny surface?

[uncle_bob]

Hi

Solder mask may be "thick" at visible wavelengths. That may not be true at IR. Solder mask over copper can read quite a bit differently than mask over other things. When we do IR profiling of PC Boards (with a camera), we spray paint them black first. Without that step, the data does is not very useful.

Bob

[mich41]

Solder mask over copper can read quite a bit differently than mask over other things. When we do IR profiling of PC Boards (with a camera), we spray paint them black first. Without that step, the data does is not very useful.

And how bad is it? What I'm looking for is some way to check if ICs, FETs etc. stay within their limits. Something like +/-10°C accuracy within 20-120°C range would be totally fine. I rarely deal with heatsinks (and then thermocuples are OK for those), but PCB exists everywhere and I wouldn't want to get bogus results each time the spot covers some PCB area around the component, especially when it's small.

[uncle_bob]

Solder mask over copper can read quite a bit differently than mask over other things. When we do IR profiling of PC Boards (with a camera), we spray paint them black first. Without that step, the data does is not very useful.

And how bad is it? What I'm looking for is some way to check if ICs, FETs etc. stay within their limits. Something like +/-10°C accuracy within 20-120°C range would be totally fine. I rarely deal with heatsinks (and then thermocuples are OK for those), but PCB exists everywhere and I wouldn't want to get bogus results each time the spot covers some PCB area around the component, especially when it's small.

Hi

Once we figured out what was going on and played a bit, we could easily introduce 30C errors on a pretty typical board.

Bob

[wraper]

Solder mask over copper can read quite a bit differently than mask over other things. When we do IR profiling of PC Boards (with a camera), we spray paint them black first. Without that step, the data does is not very useful.

And how bad is it? What I'm looking for is some way to check if ICs, FETs etc. stay within their limits. Something like +/-10°C accuracy within 20-120°C range would be totally fine. I rarely deal with heatsinks (and then thermocuples are OK for those), but PCB exists everywhere and I wouldn't want to get bogus results each time the spot covers some PCB area around the component, especially when it's small.

With a thermal camera I barely see any difference between the black plastic ICs  and the PCB material itself (maybe 0.1-0.2^oC difference at room temperature). No tracks are visible through the solder mask (likely depends on the solder mask material).

[uncle_bob]

Solder mask over copper can read quite a bit differently than mask over other things. When we do IR profiling of PC Boards (with a camera), we spray paint them black first. Without that step, the data does is not very useful.

And how bad is it? What I'm looking for is some way to check if ICs, FETs etc. stay within their limits. Something like +/-10°C accuracy within 20-120°C range would be totally fine. I rarely deal with heatsinks (and then thermocuples are OK for those), but PCB exists everywhere and I wouldn't want to get bogus results each time the spot covers some PCB area around the component, especially when it's small.

With a thermal camera I barely see any difference between the black plastic ICs  and the PCB material itself (maybe 0.1-0.2^oC difference at room temperature). No tracks are visible through the solder mask.

Hi

... and that is what we got after we sprayed our "clear at IR" masked boards with the black paint.

Bob

[wraper]

Hi

... and that is what we got after we sprayed our "clear at IR" masked boards with the black paint.

Bob

That is what I get without any painting with the boards I have on the hand right now. On same of them, tracks a visible a bit, but just barely. Likely depends on the solder mask used.

[uncle_bob]

Hi

... and that is what we got after we sprayed our "clear at IR" masked boards with the black paint.

Bob

That is what I get without any painting with the boards I have on the hand right now. On same of them, tracks a visible a bit, but just barely. Likely depends on the solder mask used.

Hi

Needless to say, we were a bit surprised when we first figured out what was going on. Any shiny metal is going to be a problem, it's not just a mask issue. We figured on it for the exposed stuff, we didn't anticipate it on the mask. Even with metal, it depends on what kind of metal and it's surface finish. Plated copper is a different beast than milled aluminum.

Bob

[xygor]

I stick a piece of Scotch tape (dull not shiny) to shiny objects to measure them using IR.

[xygor]

... When we do IR profiling of PC Boards (with a camera), we spray paint them black first. Without that step, the data does is not very useful.
...

Black what? Why black?  Gray primer should do pretty well.  Applying it to the whole board may affect the result.  If it is too thick, it insulates.  If it is too thin and has good emmissivity, it can cool the board.

[uncle_bob]

... When we do IR profiling of PC Boards (with a camera), we spray paint them black first. Without that step, the data does is not very useful.
...

Black what? Why black?  Gray primer should do pretty well.  Applying it to the whole board may affect the result.  If it is too thick, it insulates.  If it is too thin and has good emmissivity, it can cool the board.

Hi

Black because we got a couple of cans of this and that. The black stuff was best when put on as lightly a we could spray it.

Bob