PDA

View Full Version : Back on the temperature measuring camera subject...



J Tiers
05-10-2013, 12:55 AM
The original question was interesting to me, and I was surely ticked when I found it closed.... so let's continue the useful part, if possible......

I have the same issue, wanting to find hot parts on circuit boards quickly. Things can go bad in a hurry with power circuits, especially with power levels of 10 to 200 kW.

For various reasons, it is impractical to coat the parts with anything, and in some cases they are already coated by a conformal coating anyway.

measurement with any sort of contact device is impractical, for several reasons.

1) as the parts are, in many cases, smaller than the "bead" at the end of even a very small thermocouple. Any practical sensor would act as a heatsink, changing the temperature more than the emissivity differences are likely to.

2) there are a jillion of them on the typical board, and trying to plod through the lot would take far longer than is available, the damage would be done long before a fraction were tested

3) There is no practical way to attach a probe to most of the parts, anyway.

4) voltages can be 750V or higher

As a result, imaging would be best, and the more accurate the temperature measurement is from the image, the better. Probably nothing more than a 30C error is remotely tolerable, when operating in the range up to perhaps 150C. Around 10C would probably be sufficient error band.

nothing above 150C is of interest, since by that temp on the case, the device is already damaged. Better is from 60 or 70C up.

I have not seen anything capable of doing the job for less than several thousand dollars, and at that price the performance is marginal. From the prior discussion, it seems unlikely that anything less expensive will work, as the good sensors are expensive.

But in case anyone has a clever idea, I thought I would ask if there were any somewhat cheaper ways to get there, even if the performance is a bit limited.

As it has been said, when you want to see where you are going, bright light is nice, but compared to the dark, a candle is an improvement.

The Artful Bodger
05-10-2013, 01:08 AM
See if you can find what happened to all the SARS cameras they used to have at airports.

Optics Curmudgeon
05-10-2013, 01:46 AM
Some years back Cadillac offered a thermal imaging camera as an option. Maybe a junkyard will have one?

Evan
05-10-2013, 10:44 AM
I am going to do some more experimenting with luminescent paint. I'm going into town today and will be looking for some that sprays on. It's a water based paint containing zinc sulphide so I plan to first spray a board with waterproof acrylic spray and then apply some luminescent paint. I read that it won't do for Jerry but it may well be useful for others including myself.

lwalker
05-10-2013, 01:23 PM
How large are the boards and how rapidly do you need to image each one? I'm thinking of a raster scanner (maybe get a scanner out of an old barcode reader) using one of the Melexis IR chips (http://www.melexis.com/Infrared-Thermometer-Sensors/Infrared-Thermometer-Sensors/MLX90620-776.aspx).

applescotty
05-10-2013, 01:40 PM
Maybe this was mentioned in the old thread (I don't know which one it was?), but there's this:
http://www.rhworkshop.com/

I can't tell if it meets all your requirements.

Scott

Evan
05-10-2013, 01:46 PM
I will have to also try my IR Camera and see how sensitive it is. It easily detects stars in IR.

dp
05-10-2013, 02:27 PM
I will have to also try my IR Camera and see how sensitive it is. It easily detects stars in IR.

I'm interested in seeing how extended or multiple exposures work on fixed targets like a circuit board. It isn't practical where succh exposures are impractical, but for the limited needs of seeing thermal variations on circuit boards it may work. Sensor noise being the big problem especially at high ISO values. This is where stacking several images taken at lower ISO and with an intervalometer may work. Treat them as you would star trail images to hold the noise down.

J Tiers
05-10-2013, 09:58 PM
How large are the boards and how rapidly do you need to image each one? I'm thinking of a raster scanner (maybe get a scanner out of an old barcode reader) using one of the

The ones of interest at the moment are about 100mm square, and 60 x 200mm, but we have had boards up to 300 x 400mm or larger, and "area monitoring" is another use, so I guess the answer to that is "various".


Refresh time is not necessarily very fast in an absolute sense, since we are talking about thermal responses, which are typically on the order of several seconds at least. I would suppose that a refresh time of a second per frame would work very nicely, and so would 2 or 3 seconds, most likely. If it got a lot longer than that, I think it would be too slow. Problems often start small, and after a matter of "X" number of seconds start getting bigger quickly. You'd want to see a "bright spot" pretty early, well before it starts emitting visible light!

There are really two uses for it.

1) circuit boards, especially first prototypes, where one *thinks* everything has been dealt with, but may be wrong. Here fast is good, as the thermal capacity of parts is typically small.

2) larger tests, where a camera and screen might be set up to monitor an "area" to see if any places are getting hotter than expected. When one has boost transformers, rectifiers, high power inverters, and so forth operating in an area of several metres square, at 480VAC, you don't necessarily want to be bending over the bits and pieces sticking thermocouples in to check, nor do you really want an extra bundle of thermocouple wire coming out from parts at various voltages, even if it were possible to monitor everything.
An overall picture could show the distribution of heat very nicely, possibly from a position where it is inconvenient or dangerous to stand.

Fasttrack
05-10-2013, 11:37 PM
First, thank you JT for opening up another thread on the subject. I am very interested to see what progress we are able to make in this area. The board I'm interested in checking is the small SMPS I've discussed with you in the past. It works very well in my basement but I want to be sure it handles high ambient temperatures, too.


I am going to do some more experimenting with luminescent paint. I'm going into town today and will be looking for some that sprays on. It's a water based paint containing zinc sulphide so I plan to first spray a board with waterproof acrylic spray and then apply some luminescent paint. I read that it won't do for Jerry but it may well be useful for others including myself.

Did you find any? I went out two days ago with the same intention. I was going to spray some clear acrylic and then top coat everything with Rustoleum's "Glow in the Dark" paint. Both of the local hardware stores were sold out!

Alternatively, I may eventually try to come up with something using the rastering technique. Right now, I have too many projects on my plate to consider another one.

J Tiers
05-11-2013, 12:45 AM
[QUOTE=Fasttrack;850348 Right now, I have too many projects on my plate to consider another one.[/QUOTE]

I know about THAT problem.....

As for the temperature issue, if you don't need real-time data, and can use just the information that a part, or location got "hotter than "X" degrees", McMaster-Carr sells temperature "crayons". You make a mark on the part with the "crayon", and if the material gets hotter than its melting point, it changes appearance, so you can go and look later to see if some particular temperature has ever been exceeded.

Accuracy is stated to be 1%, and there are a rather wide range of temperatures available. For after-the-fact diagnosis, that can be quite helpful. Back in the music company days, I would mark likely parts on prototypes with several different temperature materials (we used the Omega "lacquer", not the crayons). Later, after the protos were returned, or after various tests and abuses in the lab, we'd check the marks and see what the maximum temp had gotten to.

There is another type, not as good. Omega sells the not-as-good ones, where you have to try the "stick" on the hot material, if it makes a streak then it is hotter than the melting temp of the stick. Obviously much less useful.

Neither gives real time warning of a problem, and for my purposes are not really suitable. In this case, I want advance warning, not confirmation after the fact.

I'll look into the ideas given above. Thanks

Evan
05-11-2013, 03:44 AM
I'm interested in seeing how extended or multiple exposures work on fixed targets like a circuit board.

I have a "real" IR Camera with photomultipler, generation 2 I believe. I am not sure what the IR sensitivity wavelength range covers. It is Russian surplus but it has an enormous objective lens compared to any of the others from that period. The lens is around 60mm aperture with extremely short focal length.

Evan
05-11-2013, 03:52 AM
Did you find any?
No, but I do have some for brush painting. I want spray paint to obtain an even coating. The brush paint is thick and I'm not sure how well it will spray from a touchup gun. Also a lot more trouble with a lot more waste.

ikdor
05-11-2013, 04:24 AM
I think your best bet will be to get a second hand thermal imager from eBay. Just saw a Flir unit for $825. There's no way you can homebrew something for that kind of money.
Igor

Evan
05-11-2013, 09:11 PM
You might be able to rent a FLIR from a local Private Investigator. They like those sort of toys. I had a chance to play with a very good one from a PI that was a customer of mine. The price varies dramatically according to resolution.