EEVblog Electronics Community Forum
Products => Thermal Imaging => Topic started by: TomS_ on June 12, 2020, 10:40:39 am
-
Hi all.
I am buying a board from an ebay seller. Its a backplane, so largely busses between sockets.
The seller got back to me before shipping saying they detected a short between two power rails, and would therefore send me the board for free if I wanted to have a go at repairing it, and indeed I would be interested in trying that. Its a 4 layer board, so if the short is internal then Im stuffed, but I think its worth a shot.
I am thinking quick and dirty: apply some voltage and limited current and simply using a thermal camera to help locate it.
But I dont have a thermal camera at present, and as much as a Flir would be nice, they are quite pricey.
So what can people suggest as a reasonable "my first thermal camera" ?
Thanks!
-
First think through the basic choices:
Do you want a 'stand-alone' either video out or with a display, or do you want something attached to a smartphone ?
Just to see hot things either will do, the phone dongle type is more use for taking photos and later analysis.
A lot offer visual overlay, but for close work this is worse than useless.
Even basic resolution (160x120) will do fine for hot spotting, aided by a cold screwdriver.
You will need close focus. There are secondary lens methods described well in the forum, and there are cameras with main lenses you can focus.
Not that many 'phone dongles' are good for that.
Are you up for a bit of building / modification, or want an off the shelf ?
How much is 'too pricey' - although that may vary with what get for it.
Bill
-
I am thinking quick and dirty: apply some voltage and limited current and simply using a thermal camera to help locate it.
That will work, in fact a few seconds of power is likely enough rather than leaving it on. You will only need to raise by <1°C to see it easily, and thermal conduction will tend to smear out the hotspot.
Bill
-
Hi Bill,
After reading your post I can see that my initial message is a bit vague. Let me see if I can expand some points:
- Ive always been wary of dongles that attach to phones. Lots of technologies surrounding phones change rapidly, and it seems some companies have a pretty poor track record of keeping phone apps up to date as OSes progress, so I would worry too much about it becoming unusable in the future. If I could get loads of short-medium term use out of it, maybe I would consider it, but I would probably be likely to use it lightly over many many years.
- So long story short, in that respect, I think I would prefer a self contained unit with its own built in display. Image capture would be ideal, but I dont think I need to go as far as video capture.
- Focus is not something I had thought of. I dont need "macro ability", I suppose if you were holding it 30-odd cm away from the object you were observing and could make out individual hot spots that would be ideal.
- Pre-built, I dont fancy mucking around trying to build something myself in this case (I do enough building and just want something that works).
- I suppose Im not really sure what I should be expecting for something reasonable. But since Im going to have to choose a price range, I suppose £500 give or take would be ideal. I will have uses for the camera going forward, so I dont want a cheap throw away (plus Im tired of buying cheap tools), but I dont think my requirements justify spending 1000's
Thanks ^-^
-
How about renting a thermal camera if it is just for fix one board.
I have done some thermal imaging in the past and found that if you are going to get down on component level you defenly want a zoom or macro.
By renting the camera you get one that is within calibration and can get the additional zoom/macro.
-
Take a brush, and with it flood the board with IPA (alcohol). When powered up, the short circuited component will heat first, thus evaporating the alcohol in the area where the shorted component is.
-
Hi Tom,
Overall budget should be fine. Those who have used a wider selection of cameras should be able to make useful suggestions. I think Fraser had a FLIR PCB inspection camera ?
The focus / lens you will get might be surprising. Lenses are expensive, especially long focus ones, and because thermal cameras are generally wide aperture you tend to get wide field of view.
For PCB work that means needing to set very short focus (as you prefer not to add extra lenses).
This is basically keep undoing the lens and stop just before it falls out. :phew:
My cameras are very wide angle (50° field) I end up focusing at more like 50-75 mm.
Attached an image from a working 70mm sized PCB and soldering iron, there's a couple more images / video down the bottom of my page here:
http://www.fire-tics.co.uk/project1/index.htm (http://www.fire-tics.co.uk/project1/index.htm)
The core there is near to what you'd like, but need boxing up.
regards
Bill
-
Take a brush, and with it flood the board with IPA (alcohol). When powered up, the short circuited component will heat first, thus evaporating the alcohol in the area where the shorted component is.
Interesting technique...!
-
How about renting a thermal camera if it is just for fix one board.
I have done some thermal imaging in the past and found that if you are going to get down on component level you defenly want a zoom or macro.
By renting the camera you get one that is within calibration and can get the additional zoom/macro.
Im not organised enough to know in advance when I want or am going to do something (this case would be an exception as I havent received the faulty board yet). :-DD
I tend to just need to grab something and go, so renting might be an option if I really needed that particular functionality at a particular time, but otherwise I would probably prefer to have it at hand.
-
Lenses are expensive, especially long focus ones
mmm yes, as someone who owns some Canon L series lenses, I know the pain all too well. ^-^
I noticed Flir has a model, the TG267, which would seem to "fit" within my budget (or really, consume most of it). Ive seen good things about Flir so they are kind of top of my list in terms of potential purcahses.
-
You could use a sheet of LCD material from LCR Hallcrest, which is a UK company, try different temperature ranges depending on your ambient temperatures. Its cheap, non-conductive (but doesnt like extreme heat) and it works great. As you said, make sure to use a current limited (best) or simply a variable supply to find the short so you dont damage it.
An LCD is probably as good or better than a cheap thermal camera and will give high resolution as to heat spots if you can put it horizontally above a PCB thats lying flat. The further away it is the less resolution. You could cut smaller pieces to get into tighter areas.
Hi all.
I am buying a board from an ebay seller. Its a backplane, so largely busses between sockets.
The seller got back to me before shipping saying they detected a short between two power rails, and would therefore send me the board for free if I wanted to have a go at repairing it, and indeed I would be interested in trying that. Its a 4 layer board, so if the short is internal then Im stuffed, but I think its worth a shot.
I am thinking quick and dirty: apply some voltage and limited current and simply using a thermal camera to help locate it.
But I dont have a thermal camera at present, and as much as a Flir would be nice, they are quite pricey.
So what can people suggest as a reasonable "my first thermal camera" ?
Thanks!
-
Consider alternative approaches as well. Send an AC signal at audio frequency between the two shorted rails, and trace it using a small pickup coil connected to an audio amplifier. Or HP made the 547A current tracing probe for just this purpose, part of a 'logic troubleshooting kit' which included a high-current pulser. The probe inductively picks up the signal from the pulser and lights a lamp when it is close to a PCB trace carrying the pulser current. They turn up on eBay from time to time, but the prices go up & down (a lot) in mysterious cycles. I got my kit for about £50 a few years ago
-
mmm yes, as someone who owns some Canon L series lenses, I know the pain all too well. ^-^
I noticed Flir has a model, the TG267, which would seem to "fit" within my budget (or really, consume most of it). Ive seen good things about Flir so they are kind of top of my list in terms of potential purcahses.
160x120 is OK (same as the images I posted above)
Field of view is wide (as suggested) so you will need to get close, or use a secondary lens. Using at the minimum focus of 300mm, the 'scene' will be 300x240mm, so that is 2mm sized pixels if perfectly focussed. May be a teardown on here with a simple refocus method.
TG267 has the pain of a visual overlay (MSX), not sure if it can be turned off (- well some tape over it will stop it !)
'Focus Free' is a lie. It just means 'fixed vaguely somewhere and you can't change it'. The visual image merged via MSX contributes most of the apparent sharpness to images, which with parallax from close-up use can be confusing.
Storage and upload is good as usual for FLIR, handy for other uses you might have.
8.7Hz is not great, but not a problem for faultfinding. You are stuck with that with most cameras as a limit both of processing and export control.
Bill
-
In the posted photo the temperature difference between the hot component and the background is very big, tens or hundreds of degrees. In a short circuit there will be almost no heating, probably less than a degree.
What is the smaller temperature difference that can be seen by the IR camera?
-
In the posted photo the temperature difference between the hot component and the background is very big, tens or hundreds of degrees. In a short circuit there will be almost no heating, probably less than a degree.
What is the smaller temperature difference that can be seen by the IR camera?
For video5.png the difference is probably a couple of degrees. Certainly barely noticeable to my finger.
The soldering iron is there as everyone knows what one is !
'Detection' starts around 10-20mK, but this test is where the good and mediocre differ. 100mK (0.1°C) would be easily visible to anything apart from old Pevicon tube and thermopile systems. For a small source (eg cooking a SOT23) you have to bear in mind that you see the average of a pixel sized box. So if the part only fills 20% it would need to be about 5 times hotter to see it or to conduct out to be a pixel size.
With a SOT23 thermal resistance of around 150°C/W you will see any 'improper' activity in seconds and well before magic smoke time. Beyond that a lot will depend on how the PCB conducts heat away.
I'll do a couple of videos of the standard 'party tricks', a phone book and rubbing a shoe across a carpet in silly shapes. What I have not got handy is a differential chart.
-
From what the OP has reported of this PCB’s fault, a particular PCB track has been identified as faulty with a short.
In such cases it is unwise to strain the power supply that normally feeds the PCB in case long duration over current is involved with negative effects on the regulators. I would approach such a fault with a Toneohm equipment but that is not the suggested MO of the OP. If using a thermal imaging based short tracing technique I would connect a lab power supply to the failed PCB track and the track to which it is shorted. I would apply a voltage of around 1V but set the current limit to something reasonable that will not damage the PCB track. 1A should be safe unless the PCB tracks are very thin. The lab power supply will continue to pass 1A through the PCB shorted track rather than fold back and shut down, as some equipment power supplies do. As such heat will be created in the PCB track and is easily seen with a thermal camera. Most thermal cameras can achieve a Delta T sensitivity Of better than 0.1C so the heating effect will be detectable. In some cases it may be necessary to carefully increase the current flow but that is often only needed for very challenging high layer count PCB’s and must be done with care. If the PCB gets too hot around the area of interest and thermal spread blurs the source, just cool the PCB in the fridge, freezer or using Freezer spray. Once cold, repeat the current induced heating test.
The PCB track path between the lab power supply input connections should be relatively obvious and any failed component or track insulation identifiable. It is not always so easy however, especially on a new PCB that failed production testing. A failed feed-thru under a VLSI chip or through hole connector can be a challenge ! Life is made a lot easier I’d a schematic is available for the PCB to identify components and connection nodes on the track of interest.
A thermal camera is quite an investment so I advise caution when shopping at the budget end of the market. Do not be fooled by some sellers offering deals that are too good to be true, as often they are ! Very low resolution plus masses of interpolation is not a wise camera choice for PCB work.
FLIR do offer reasonable cameras that use the Lepton but they are a compromise solution so suffer from relatively high image noise content and need a close-up lens for PCB work. Such a lens is just a $20 CO2 laser focus lens and is needed by many fixed focus cameras for PCB work at less than 30cm distance.
Now, given a choice, I would personally buy one of Bills Raytheon Thermal Eye 2000AS cores that produce lovely monochrome images of 160 x 120 pixels using a quality Germanium lens and Raytheon quality electronics with its built in image processing to provide good images. Such a core can be up and running in mere minutes and providing excellent images on a TV or dedicated composite video NTSC monitor. For £160 they are a bargain. The FLIR budget cameras are far more ‘sexy’ and a complete camera solution. The Lepton core is ‘OK’ but do not expect too much from it. The Lepton 3 160 x 120 pixel based cameras are the minimum specification I would recommend for PCB work at close range using a Supplemental close-up lens providing 100mm working distance.
Fraser
-
Take a look at this video:
https://www.youtube.com/watch?v=t5fICjcaJ3E (https://www.youtube.com/watch?v=t5fICjcaJ3E)
I'm in the same boat for cooling electronics though.. I hear the HTR-301 is pretty awesome.
-
A small piece of liquid crystal thermal color changing plastic like the kind made by Hallcrest is as good as an IR camera for the purpose of finding hot spots, and only costs a few dollars. they are non-conductive and (in my experience) last for years, and are extremely easy to simply place over the PCB where it will change color according to temperature, pointing out where the heat is and changing color instantaneously when it changes.
-
Thanks everyone for the tips and suggestions.
I think I will give the following approach a try.
First will try IPA because I already have a bottle of it and it will be quick to test. Louis Rossmanns video shows quite promising results. I do need to get myself a bench power supply though, so I can control the voltage and current, so I may just end up buying the unit I've been eyeing for some time now. Or I could hack something together with a smaller power supply and a resistor perhaps.
Failing that I will try this color changing plastic film that has been mentioned a couple of times. I will have to buy it, but it doesn't look too expensive.
Failing that then perhaps I will look at thermal cameras again.
Cheers!
-
+1 for toneohm technique described by Fraser.
also cheap alternative is to build yourself a "shorty" device.
http://kripton2035.free.fr/Continuity%20Meters/continuity-short.html (http://kripton2035.free.fr/Continuity%20Meters/continuity-short.html)
you can really spot tone differences if you move the probe from 2cm on a back plane !
-
Bit of an anti-climax here, the backplane does not infact have a short, it was simply low resistance between gnd and +5v due to 144 parallel 800ohm resistances (pairs of 330+470ohm dividers) used to terminate the bus at either end of the board.
Having removed those termination resistors any signs of resistance and continuity disappeared.
At least I didnt fork out for a thermal camera just yet. :-DD :phew:
-
Another way to find shorts (next time!) is to send A/C into the short, from your signal generator (e.g. 400Hz - 1KHz) and trace the short with a small sniffing coil (I use a head from an old cassette player).
-
a small sniffing coil (I use a head from an old cassette player).
Wow, that's a very cool trick to play with a cassette player head, nice hack! :-+
Thanks for the idea.
-
a small sniffing coil (I use a head from an old cassette player).
Wow, that's a very cool trick to play with a cassette player head, nice hack! :-+
Thanks for the idea.
The tape head is perfect for this, it resolves sub-mm distances on the PCB! And if you listen in stereo, you can hear when you cross the track... It is good fun! :D
-
Take a brush, and with it flood the board with IPA (alcohol). When powered up, the short circuited component will heat first, thus evaporating the alcohol in the area where the shorted component is.
That's what I was going to suggest as well. I saw Louis Rossman demonstrate it a while back, he claims it works better than a thermal camera. I did use it successfully once to locate a short.
I doubt the short is internal to the PCB, it's going to either be a shorted semiconductor or a power supply bypass capacitor most likely.
-
An old practice was a can of freeze spray. The bit of frost that disappears quickly is the source.
An inverted can of canned computer duster will also work.