IR camera for PCB debugging

I’m considering springing for an IR camera to scan boards for hotspots. Anyone care to recommend an entry-level device for hobbyist use? Can I get away with paying $250-$300 for a serviceable unit?

I am curious if anyone has had good results using one of the cheaper options like a FLIR One Pro. I have only used the more expensive FLIR cameras and they work well for this kind of thing.

I’ve made a couple of gizmos to see IR/full spectrum… Have done it with $10 USB webcam’s… Google it (many posts on How To Do It), many on youtube.

It starts by removing the IR filter that covers the lens…

The one below is still on my desk - Camera and Gizmo that receives the light and spectrum-izes it, so to speak… Homemade and works well… taken apart for the screenshot…

Alternatively, you can use a $12 Temp reading gizmo - you know the one’s I mean…

Not suggesting that you don’t buy one, perhaps a fun toy. But, perhaps making one is as much fun… Here’s one link…

I have found assembly faults on several boards with very basic FLIR cameras. You do not need very high resolution to be useful to see the hot spots. Much easier than the old four wire voltage sensing days

The original post (and my response) inspired me to dig out some earlier IR and Imaging gizmos I made… one thing leading to another ended up by watching this YouTube video. This video is what you want to start with for a more current approach that’s Low Cost, easy and inspirational…

…such that I wouldn’t be surprised to find myself making another gizmo today :nerd_face:

[EDIT] Adding info…
I mentioned using a Webcam and the many posts on doing it…
While drinking coffee (after ordering an AMG8833), I dug up a useful YouTube video showing use of Webcam with the IR filter removed… Here’s the Video

I’ll report back after I receive the sensor and build the Gizmo with it - thus far mine use PIR’s and Webcams so, I’m looking forward to another quick project…

CMOS sensors as in webcams are capable of seeing near infrared. The FLIR cameras use more exotic sensors like lead telluride, that can detect much longer wavelengths ie cooler sources

Well that explains. I had heard of the webcam trick and never tried it despite having something of a history of dismembering webcams. I tried it on one of the dozens of Logitech B500s I have lying about and it revealed nothing about the heatsink’ed linear regulators I had on my bench, which were just hot enough to be uncomfortable to touch.

This device looks about right: https://www.flir.com/products/flir-one-gen-3/. I’ll follow up here next month with the outcome.

I’ve been following this project for a while now. The project creator is making good progress on an open source instrument that uses automotive imaging sensors (same ones used in FLIR instruments). Looks like you will be able to order some PCB’s from him in September.

https://www.eevblog.com/forum/thermal-imaging/openirv-isc0901b0-(autoliv-nv3-flir-e4568)-based-opensource-thermal-camera/

If you want to just buy something off the shelf, the thermal imaging forum has a lot of good info. Best solution is to by a FLIR E4 and then upgrade the firmware to increase the resolution. You 3D print some adapters to add an IR lens.

Finally got around to (almost) finishing up my effort on making one. Only remaining work is to make the PCB and Design and Package it into 3D printed parts.

I made several of them - trying out various schemes and got the Webcam working with an IR color scale (typically used for Spectrograph IR projects). But, I decided to try an AMG8833 sensor. Dialing-in the code and using a TFT worked the Best.

[Edit - corrected the Temp spec/range. Vid below does not reflect corrected code limits] The AMG8833 Temp range is 0ºC to 80ºC.
All is working well. I used different parts (Nano, Pi-Pico, ESP32 and Teensy4.0).

I’ve fallen in-love with the Teensy4.0 so, will use it (though may upgrade to 4.1).

The Teensy has selectable Speeds and over-clocking but, dialed the speed down to what seems the best (for my preference, only).

Running at 150Mhz and using the ILI9341_t3 (the verrrry Fast library)

Parts:
• Teensy4.0 ($25 with soldered Pins, PJRC, about same price at Amazon)
• AMG8833 ($45 from Adafdruit but, half that if buying on eBay)
• TFT 320x240 ($15 Amazon)
• 3D Printed Parts (call it, $2, using my printers)
• PCB (I’ll make my own $1 for single-sided board)
• Misc… $peanuts

Results - Video shows Temperature on SN754410 Stepper Motor Driver Chip
(Excuse the mess on my desk…)

Thermal_Cam_Motor.mp4

2 Likes

I use the FLIR ONE to get an idea of what’s going on. Th only issue is as you get close-up, the thermal image gets offset from the outline of the actual part. So you start 4’ away, move in, and you can figure out what thermal signature matches what part.
Then you take a thermocouple bead thermometer (Fluke in my case) and see what you really have. Since the the sensor (bolometer) and software change the color to match the thermal range, sometime a “hot” part is not really that hot.
Enclosed is a picture of a small Atom Flex PCB. You can see the hot M2 SSD in the lower left, and in the upper part you can see the 2nd DRAM stick. Here the thermal image is shifted up from where the part is. Due to the fan positioning in the chassis, you can see the the 2nd stick runs hotter than the 1st one. I added a 30mm fan to the chassis to move the air around, things got better.
The 2nd image is my dual 50W Boost-Buck supply where I got close in. The thermal image is shifted down from the parts. Looking at the left side, you can see where the 4 screw terminals are on the PCB (J3), and a “cool spot” below in the thermal image. That lets me “calibrate” my eyeballs so I can see what is what. And sure enough, the hot spot is the small switching chip itself (U60). But it was only about 6 degrees F hotter than the inductor (L60, the big square above the terminal strips). 3rd image is the actual PCB. I went with a much bigger inductor than the Texas Instruments eval board since those ran hot, and the thermal image verified that choice. So you can both find problems and verify decisions.
For me, having the thermal imager is a “How did I live without it” moment. I don’t miss the burned finger when I found a really hot part the hard way. I have a drone with thermal imager. Becuase the subject matter is so far away, it can you a temperature reading by taping on the controller video screen. Same with the hand-held imagers in the fire department. But for the low cost of a FLIR ONE for your phone you have to think and work a bit. Still it’s a win.

thermalView

SEEK also has relatively inexpensive thermal imaging dongles which use smartphones as processors / displays.

Most general purpose thermal imagers don’t focus closely enough for PCB inspection, although of course even a fuzzy component will be evident when it’s hot enough. One fix for this is to add a ZnSe auxiliary close-up lens. They’re used with CO2 IR lasers and can be had cheaply from eBay etc. Add a 3D printed mount and you’re in business. Using one will affect the measured temperature, of course, but will still clearly indicate if a component is significantly hotter than its neighbors.

Reporting back: Finished printing the housing (a quick, rough prototype) - it contains batteries, power-dongle and USB port access…

screenshot shows it measuring blackcoffee (top cover removed)

FYI - I used Kris Kasprzak’s code as baseline - (it has many features but, I blew them away and customized it for my needs. I Added code to grab the center pixel’s temperature reading and display&hold it with button press. Other major tweaks, too.
The interpolated pixel reading can ‘interpolate’ well beyond un-interpolated sensor reading - I measured 460˚F oven (baking bread).

1 Like

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.