IR Temperature probe with same spot size regardless of distance

Den mandag den 3. marts 2014 13.57.47 UTC+1 skrev Klaus Kragelund:

how about doing an x-y-z pattern, either by probing the pcb with a proximity sensor or by inputing coordinates?

-Lasse

Yes, good idea :-)

I thought about adding a lense to reduce the angle of the beam, but that would need to be a special lense, since glass will just reflect most of the radiation

Cheers

Klaus

a water bag?

don't know these people still exist, but experts when I used them

Optical Coating Lab, Inc.

2789 Northpoint Parkway Santa Rosa, CA 95407-7350 707 545 6440

From memory one time they made a 'lens' that was opaque to visible, but transparent to infrared, and focused it. May have been a 'plastic' lens, can't remember.

ZnSe lenses, which are used for CO2 laser applications, are relatively inexpensive and do pass IR wavelengths. The ones commonly available are for cutting/engraving applications and so have fairly short focal lengths but that's probably okay (or even a good thing) for IR thermography.

You can use a very thin HDPE Fresnel lens, or else a fancier one made from zinc selenide, silicon, or germanium. Oxides are pretty well completely opaque in the thermal IR.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

Nice. I didn't know the name. Thanks for the hint, a quick search brought up literally hundreds of different types :-)

Thanks

Klaus

You'll need to recalibrate, of course, because the HDPE isn't really all that transparent in the thermal IR. Your sensor will see some combination of the lens temperature and the object temperature.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

You'd need some pretty expensive optics to make that happen. I'd try to find a lens, and focus on the board. You shouldn't need a big one.

Be sure to figure out what wavelength your temperature probe is operating at, so you can select the right lens material. I'm pretty sure that bolometers all operate at 8-12 micron, but when I worked in thermal imaging I was always off controlling platforms or designing circuitry, I only hung out with the optical guys in meetings and around the coffee machine.

--
Tim Wescott 
Control system and signal processing consulting 
www.wescottdesign.com

The problem with thermal imaging is that everything glows at thermal wavelengths.

If you use a small lens at any distance from the sensor, the sensor sees a small area of lens and a big area of other stuff, namely the interior of its own case, all glowing. That seriously dilutes the IR from the object with IR from everything else. Some of that can be mathed out, if you know the ambient temp and calibrate often.

Our FLIR imager has tiny pixels and a huge (about 1" diameter, kilobucks) germanium lens. A pixel, looking up at the world, sees mostly lens. The lens can focus practically touching a part. Cheap IR thermometers probably have one big sensor, so it will be hard to focus a small object area onto one big pixel.

Incidentally, some surfaces, like brass and copper, are practically perfect mirrors at thermal wavelengths. Something might be possible with reflective optics.

Visible, hand and copperclad FR4, with a bit of kapton tape

Thermal IR

The copper is mostly reflecting the thermal image of my hand and the ceiling, which was cool that day. The kapton, like most plastics, is nearly black at thermal wavelengths, so appears as the temperature of the FR4.

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation

When I'm looking for hot spots, I use a liquid crystal sheet. They're not accurate, clumsy if you have tall parts, slow, but cheap and convenient.

If your IR thermometer responds in about 500 msec, and you step on a

0.1" grid (100 points per square inch), you'll be running at 50 seconds per square inch. So, a common smartphone size PCB (2.5 x 5") of 12.5 square inches will take 625 seconds or 10.4 minutes. Not exactly blindingly fast, but probably tolerable. The trick will be to get the spot size down to a 0.1" dia or smaller if you want to use a smaller step size.

Germanium or polyethylene IR lenses. You'll want something that will

Also check eBay as they occasionally have cheap germanium lenses.

The problem with the small diameter lenses is that you'll still get a rather wide viewing angle. As I vaguely recall, a 1" lens gave me about 10 degrees or: 1 / tan(10) = 5.7:1 At 30mm distance, and a spot size of 2.5mm, you'll need: 30 / 2.5 = 12:1 or arctan(1/12) = 4.8 degrees That's quite doable as there are IR thermometers on the market with distance to spot ratios of 12:1 and higher such as: I don't know anything about the optical arrangement inside a 50:1 (1.15 degree) device.

However, if you decide to simply buy a 50:1 IR thermometer, and attach it to your CNC machine, the scan time will be greatly increased. At

30mm distance, a 50:1 optical system will produce a 0.6mm (0.024") diameter spot. That will take 4 times as long to scan as the 0.1" spot size version.

Good luck.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

That is, indeed, a drawback to radiometric thermal measurements. The thermometer he has should be doing that already.

Decent thermal thermometers and imagers have what's called a "warm stop", and the optics are designed so that what they see is the target and the warm stop, and as close to nothing else as the optical guy can make happen. Then the temperature of the warm stop is either measured or controlled, and its effects are either actively subtracted out (if its temperature is just measured) or implicitly subtracted out at calibration (if its temperature is controlled).

GOOD thermal imagers hold the sensor at cryogenic temperatures (77K is common, because it makes testing easy if you have some LN2 handy). The sensor is mounted on a pedestal inside of an evacuated flask, and is surrounded by a blackened optical "cold stop". Again, the optics are arranged so that the sensor "sees" only the cold stop and the thing to be imaged.

If the lenses are decent, they are transparent to IR and they have anti- reflective coatings on them. Consequently, they are invisible to the sensor. The effect is never perfect, but with appropriate effort you can minimize it.

It is imaging in 8-12 micron, then. Silicon is much cheaper, and works nicely at 3-5, but not 8-12.

If the lens occludes the entire viewing area of the sensor, then the sensor will "see" only its own warm stop, however much of the lens isn't rendered invisible by transparency and AR coatings, and the scene. So using a lens to do the job is a quite valid method.

You probably do want the lens's blur spot to cover the whole "pixel", though.

That's probably a wacky notion, but maybe not. To really work well you'd need a parabolic section that's not cylindrically symmetrical, which would be a bitch to machine accurately -- a $1000 lens would probably be cheaper in the end. But since he just wants to make the focused spot smaller, he may be able to use a spherical mirror, bounce off of that, and accept the resulting comatic abberation.

--
Tim Wescott 
Control system and signal processing consulting 
www.wescottdesign.com

d to a DMM

Great info.

Regarding the scan time, I would let it run a course scan first, and then f ocus on the hotspots. So in principle add a dT/dx function, which should re duce the scantime perhaps 10 fold

Cheers

Klaus

Thanks. I'm not an expert in optics. Just a dabbler.

Good idea. Image enhancement tricks are also useful, but the I suggest you make improvements to the system before you attempt those.

I predict you'll see a giant blurrrred image instead of anything resembling the PCB. You may find that the slower you scan, the better the image. The problem is that the sensor is getting warmed by the target and needs some time to recover. There are also a variety of IR sources contributing to the IR "noise" level, such as reflections, lens heating, air currents, room lights, your body, the CNC machine motors getting warm, etc. The general effect is to reduce the sharpness and contrast of the image. Cooling the PIR sensor with a Peltier junction device will help, if you can get rid of the water condensation problem. I tried it with dry ice (frozen CO2) which worked amazingly well.

Incidentally, I posted to sci.electronics.design a previous rant on my assorted failures doing something similar with a supermarket bar code reader modified for IR. Also, my partial successes with a pin hole lens and liquid crystal sheet in making an IR camera obscura.

Argh... I'm late... gone.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

You need to use a FLIR camera for that however, I do have an IR laser pointer probe I got somewhere, which works very well for doing component ball park temperature monitoring..

With the IR pointer, surface types make true temperature vary, the FLIR seems to work much better if you're looking for comparison data.

What works the best, I have found, is a K probe directly on the component.

Jamie

A FLIR has the same emissivity issues as an IR thermometer, namely a partially reflective (low e) part will look partly like the ambient (walls, ceiling, whatever) temperature. I image a board and wave my hand around as if the board were a mirror. If part temps change, their emissivity is low. Kapton or electrical tape, or black whiteboard marker, will improve emissivity.

You can set the emissivity on a FLIR menu, and it will try to compute it out, but that's a mediocre approximation.

--

John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Time for a nice custom IR telecentric lense. Do these even exist?

Klaus, You can try this: cut a small hole in a piece of aluminum foil and put it over the opening of the probe. See if the signal strength is OK, and you may have to recalibrate.

I have a Ryobi which can "image" small things a good distance away. Looking at the 80t, I'm surprised it goes 1:4 aspect, as the snout looks long and narrow. Are you 100% sure it expands like that? Did they use a metal lining on the snout? Something doesn't add up! jb

The sensor will see a tiny glimmer of the object in the middle of a huge reflection of itself in the foil. The temperature that you measure will be mostly the reflection.

Focussing or masking thermal IR is different from doing that with visible light. Everything is glowing in the thermal IR.

--

John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

t over

have

Yes, everything glowing but the aluminum foil, which has a low e. Now, take a look at the 80t. It has a very long snout in relation to its di a., AFAICS. The first question Ihave is what is this snout made of, low or high e? The second question I have is, if this snout defines the geometrica l optics here, why is Klaus seeing so much beam spread? My solution is to neck down that snout a bit, pretty far from the sensor, w hich is already confined in a long tube. So I feel that an annular ring of foil would not change the optics as much as the snout already does. And that brings up a question here, "How does Klaus know he has that much b eam spread?" And what is that snout made of? Pertinent questions to ask before saddling up...:) jb