FLIR Thermal Camera for teardown - my latest patient, the E2

[robrenz]

Very interesting and well documented thread, keep up the good work! 

[Fraser]

@robrenz,

Thanks.

This is just me having fun.... you just don't usually get a blow by blow account on such things from me.    I just hoped it might make an interesting little thread for some. It also helps my grey brain cells to write about the camera..... helps focus the mind and to spot errors in my thought processes.

I have been asked to document fault finding on thermal cameras before but it is quite a task to do properly. I may yet do a more formal document on the topic.

Aurora

[Fraser]

Cleaning of the microbolometer capsule has been completed 

It took a little more aggressive cleaning action than just the TePe brushes but I am happy with the result. There was significant red rust between several pins which can't have been good for performance.

I ended up using a combination of IPA washing, TePe brushing and surface abrasion using a blunt tipped dental pik. These removed all but the most stubborn corrosion and that is nowhere near the leads so is of little concern. I will be happy with 90% on this task  'Before' and 'After' pictures attached.

I then evaluated the risk of lifting the corner of the microbolometer that had been pressed closer to the PCB. There was the risk of lifting pads that were placed under tension, cracking the glass vacuum seals through torque action and the lesser risk of distorting the capsule case. In the end I elected to correct the lead distortion and lifted the capsule using careful leverage. All went well and the capsule now sits parallel to the PCB and the leads are correctly shaped 

Another step closer to the retesting of the unit.

Aurora

[Fraser]

I have spoken a lot about the microbolometer in the E2 but have not explained its internal design or the purpose of its various external connections. Time to remedy that for those unfamiliar with this type of device. Don't worry, I am not going to go into any great depth here, just a high level overview.

The microbolometer used in a staring array such as fitted in the E2 is basically a large matrix array of bolometers or thermistors in common terms. These bolometers are individually read by read out electronics and their values are used to build a thermal map of what the array is seeing. The thermal energy of a scene is projected onto the array by the thermal cameras optics and so the thermal scene is recreated on the surface of the bolometer array. All sounds simple until you consider how such an array will behave in the real world.

Each bolometer in the array will have its own unique thermal response characteristics and the whole array will be subject to thermal drift due to localised heating of the die and ambient temperature changes. Untamed such an array would be next to useless for a thermal camera application.

In order to tame the array of bolometers several techniques are often employed. Note that there are many ways to achieve the desired result, and this is just one. It is however the way the E2 operates.
First of all the microbolometer die temperature is stabilised at a temperature usually around 30 Degrees Celcius. This is done using active heating and cooling in the form of a Peltier Element. Note that this is NOT a cooled array in thermal camera terms. They are very different beasts. This is a temperature stabilised array. In order to stabilise the die temperature a closed loop controller is employed. This has thermal sensors attached to the die feeding it with temperature  readings. The controller then takes the appropriate action to heat or cool the die. This temperature stabilisation of the die does not counteract the changes in bolometer temperature due to scene illumination.

Non Uniformity correction and flat field correction are also needed to tame the, now temperature stabilised, microbolometer, but that is a topic for another day.

Sticking with the microbolometer, what would we find if we took one apart ?

Well the first obvious component is the capsule in which the array is housed.  The capsule is hermetically sealed and has a thermally transparent window at its front. The area inside the capsule is a vacuum which helps to prevent internal thermal currents and impurities that could degrade the arrays performance over time. Inside the capsule we would first see the staring array of bolometers mounted on the die. These are extraordinarily fragile and cannot be touched or cleaned with compressed air. On the rear of the die we would find the single stage Peltier element for temperature stabilisation. The Peltier element operates using the thermal mass of the metal capsule as its counterpoise. On the die we would normally find two temperature sensors that serve the closed loop temperature stabilisation controller. Some microbolometers also have highly complex processing power on the same die as the bolometers, others do not. The array of bolometers is read by read out electronics and either passed to the on board processing or sent to the capsules external I/O pins.

When looking at the microbolometer capsule from the outside you will see that there are a large number of pins around its side. In the case of the E2 capsule, the pins are present on three sides. These pins are for input/output functionality and have glass seals insulating them from the metal capsule and retaining the vacuum within. On the fourth side of the E2 microbolometer you will see two larger pins and a crimped tube with paint at its tip. The two larger pins are the supply rails to the Peltier element and their size relates to the current that can be involved for temperature stabilisation. Low conductor losses are important here. There is a grounding pin on the capsule case. The crimped tube is the vacuum port through which the air inside the capsule was removed. It is crimped, soldered and then coated with a gas sealant paint to maintain the vacuum over decades. Microbolometers do not require renewal of the vacuum during their service life.

The window that sits in front of the staring bolometer array is usually manufactured from Germanium. A very expensive material (metal) that has properties desirable in thermal imaging. The window seal to the capsule is achieved using specialist epoxy that is gas tight.

Well that just about covers the basics of the microbolometer that you have seen in the E2 pictures. This type of microbolometer is expensive to manufacture and so some consumer grade modern thermal cameras have moved to cheaper designs that may be manufactured more easily. These do not require the metal vacuum capsule. The same basic design principles apply however, just in a different format. 

Aurora

[Fraser]

The FLIR patent covering the Early Ex series provides some nice pictures of teh optical block and core.

I have extracted one and attached it here.

The key items of interest are:

3b,3d & 3e are the DS1624S thermal sensors
10 is the microbolometer array
10A is the Peltier element
10B is the microbolometer capsule mounting clamp
11 is the lens shell
11a & 11b are the Germanium lens elements
13 is the diaphragm
19 is the FFC shutter

For those wishing to read the Patent document. Search for Patent US6707044

Patents US7157705 and US20040124359 also apply to the early Ex series cameras.

Aurora

[CatalinaWOW]

Appreciate your work on this Aurora.  Thorough methodology and simple explanation. 

For those that are curious, thermal stabilization partially solves some of the problems in getting usable information out of these arrays.  Each bolometer sits in a delicate thermal balance.  Heat comes in from the external scene, and goes out by radiating to the back of the array and by conduction through the leads that connect each bolometer to the readout circuitry.  In a typical design these leads are the only mechanical support for each bolometer element, and are made as thin as possible to minimize conduction losses.  When a uniform temperature scene is viewed each bolometer ideally would stabilize at the same temperature, somewhere between the scene temperature and the readout temperature.  In reality the temperatures vary due to a variety of manufacturing variations in the individual bolometers, including lead dimensions, emissivity, bolometer dimensions and so on (in addition to variations from flux through the optics and other factors outside the bolometer array).  Since the radiation balance part of the equation is a power law and the conduction portion is linear with temperature, operating at an unstabilized temperature changes the balance between conduction and radiation dramatically.   Stabilization makes compensation for all of these variations between individual bolometers more tractable. 

In addition the slope of the thermistor response to temperature varies significantly with temperature, so the response of the array can be improved by heating it.  The amount of heating is a trade between this response gain and a number of negative factors whose importance varies with application.

[JBaughb]

This is a great read. Getting into the mind of someone performing equipment repair is always fascinating to me. I guess I am very much interested in other though process and how it differs from my own. Bravo, good sir.

[Fraser]

@CatalinaWOW,

Nice description 

I was surprised when I met my first non temperature stabilised microbolometer.....that found in the modern FLIR E4. When FLIR told me that they just let it stabilise at its own natural operating temperature I was puzzled as to how they tamed the thermal behaviour of the  bolometer array.
I understand that they did not bother    It relies upon die temperature monitoring and a generic compensation table that FLIR produced. I say generic as FLIR have clearly stated that the camera is calibrated at a set ambient temperature and some cameras will work better than others at different ambient temperatures, even when in the same batch, depending upon how well the compensation table matches the microbolometer characteristics     I believe this comes under the heading of 'good enough' for the intended market.

With regard to the pros and cons of the Peltier temperature stabilisation design, it may be considered a necessary evil in some cases. We already know the pros's but as to the cons......

It is hungry on power at initial start up, but this diminishes as the operating temperature is reached. If a camera is sat in a very cold environment or operated in such, the Peltier element has to do a lot of work and so draws significant power form the hosts power supply.. In battery operated equipment this can be an issue. A more serious issue with the Peltier element is out-gassing. Peltier elements contain chemicals that can out-gas during the operational life of the microbolometer capsule. The capsule contains a vacuum and would normally be considered relatively inert in terms of an environment for the bolometers. The out-gassing of the Peltier element effectively contaminates the environment inside the capsule with some potentially negative effects in high performance microbolometers. I believe manufacturers do their best to deal with this issue but it remains a negative attribute of 'on die' Peltier temperature stabilisation that should be considered.

Some may ask why the Peltier element is not external to the capsule. The answer is simple....thermal efficiency. When inside the capsule it is in a vacuum and is in close physical contact with the die that it is heating or cooling. It uses the metal capsule as its counterpoise against which to heat or cool and this is effectively sitting in free air to permit radiation and convection cooling if needed.

Aurora

[Fraser]

@JBaughB,

You need to be careful getting into my mind....it can be a scary place full of all manner of weird and wonderful technical trivia 

On a more serious note, I should warn that I am writing up this investigation purely for fun and hopefully the interest of the reader. I suffer from chronic fatigue syndrome at the moment and that can definitely blunt my thought processes when needing to concentrate for any significant length of time. It is one of the awful symptoms of CFS (M.E.) that upsets me most. I used to have a pin sharp analytical mind, but these days its a bit more like a rusty old 6 inch used nail    Sharp it is not !

This is the first thermal camera investigation and repair that I have done in some time. I do not claim that my approach to this one has been the best or most efficient. I am just feeling may away around a camera that I have not worked on before. Its just a fun little adventure for me and I decided that I would extend the teardown thread to share my progress with the diagnostics and repair work. You could consider this thread to have become my 'notebook' or doodle pad for this repair. I am hopeful that those who read it may learn something, anything, from my meandering commentary 

If nothing else, readers can have a laugh at mistakes that I may/will make along the way 

Aurora

[mikeselectricstuff]

There is an apparently unused pin  in the image - I imagine this is probably for a getter to remove the last traces of gases from the envelope after pumping and sealing.

[Fraser]

@Mike,

I am honestly not sure about a Getter in microbolometers. The capsule is not a high voltage environment like a thermionic valve and is decidedly more delicate. a getter pan erupting inside a microbolometer capsule sounds like a nightmare ! I will take a close look at my photographs again. I thought that pin was a ground connection but I may be wrong.

The E2 is now reassembled for its first test since cleaning.

The good news is that it still works and is producing a thermal image.
The bad news is that IMHO it has too much horizontal line noise at low spans and the spot temperature measurement seems to have serious issues with accuracy !

The FFC correctly flattens the image so at least that is coping with any gradient across the microbolometer array. I did a quick and dirt test using a mug of boiling water. Pictures attached.

Note in the pictures of the thermal core assembly the close proximity of the microbolometer pcb unused cream connector to the black sub-assembly that sits vertically behind it. One corner of that sub assembly points forwards and is almost touching the microbolometer connector. The sub-assembly is very poorly mounted and is only held in place by an aluminium 'arm' coming from the sub assembly heatsink ! Very poor design that allows the sub-assembly to rock on that single pivot point.

It is my belief that this E2 was dropped, landed on its lens end and the sub-assembly travelled forward into the microbolometer pcb. The location of the cream connector on that PCB correlates with where the microbolometer leads were most distorted and that is where I applied leverage to correct the distortion.

She's alive, but not out of the woods yet. I want to do some tests on the DC-DC converters looking for unexpected noise on the microbolometer supply rails. Some monitoring of the Peltier element drive is also warranted. I will also need to investigate the inaccuracy of the spot readings. In full AUTO mode the camera tends to select a narrow span and that produces noise in the image, which is to be expected. Manual mode allowed to to produce a less noisy image but I remain concerned that this camera still has some issues to be found and resolved. This could get interesting 

The bad news is that I am knackered and will need to take a short break from this project. I also have chores to complete to keep my wife happy. Watch this space fr the next stage in diagnosing this patients ailments 

I'll be back !

Aurora

[Fraser]

The images that I was seeing when the camera was first fired up after re-assembly and being aimed at a cup of boiling water. Note the temperature spans that were in operation in the pictures. Spot temperature measurement seems way off the mark. Emmissivity was not the issue here.

Tests with the lens cap on got really weird as the camera selected narrow spans and there was temperature drifting in the images. The FFC event did correct this though. Damaged Peltier thermal coupling to the microbolometer  die ? Not sure yet. Tests using a lens cap are never a great idea though as there is thermal energy passing out of the microbolometer through the optical path. I will do a proper test with a Black Body thermal source later.

Aurora

[Fraser]

@Mike,

I can confirm that the large unused pin on the microbolometer is not a ground connection to the capsule as it passes through a glass seal like the Peltier element supply pins adjacent to it.

Hmmm some more study is needed of the original Honeywell Microbolometer Patents on which this microbolometer is based. I have them filed somewhere.

Time to rest now though

Aurora

[Fraser]

One last comment from me today.....

I was told by a service agent for FLIR that the cameras that followed the PM Series were not as well built or mechanically robust. I consider the E2 to be proof of this statement. It really is a pretty poor design inside.

Ignoring the 'bodge' wires, which I might add have been executed very well, it is the whole feel of the case and PCB retention mechanisms. They all feel very much like a prototype rather than a production unit. The main PCB is secured to one half of the case using quality screws and plastic embedded brass ferrules. The microbolometer sub-assembly that sits vertically behind it is quite frankly an appalling piece of engineering. Truly a joke. The plastic chassis is supposed to sandwich a heat-sink and thermal pads between two PCB's (Main and sub)....it doesn't do this well and the PCB's pop out of  their tiny retaining clips. The thermal pads are not able to do their job as they are no longer in contact with the IC's that they are supposed to help cool.

The sub-assembly has not a single screw on it. The heat-sink is used as the mounting for the sub assembly and a single aluminium arm extends down to the cameras tripod mounting where it is retained by a single screw. It is fair to say that apart from this single mounting point and the connector that joins the assembly to the main PCB, this thing is floating  Crazy when you realise that the heat-sink cools the main processor in the camera ! The heat-sink should be screwed down onto the main PCB to ensure good thermal transfer.

More poor design is to be found in the display section. The LCD panel is placed into a carrier and little plastic clips are glued on either side of it to roughly retain it. The whole LCD panel is floating with no proper impact absorption material present. It is eventually held in place by fingering around the clear plastic screen   ... read on, it gets better (or should that be worse?)......  the LCD display carrier also contains the rubber keypad. The LCD carrier and keypad are clipped onto the display PCB. The display PCB is free to move up and down in the clips by at least 2mm    That is to say, if you try to press a rubber keypad button, the PCB moves down and away from the panel by around 2mm ! So how is the display PCB kept in place ?......  not a retaining screw to be seen. The display PCB mates with a large connector on the top edge of the main PCB. The display case retaining screws sandwich the display board between the   display case and the main PCB edge connector. Great design ..... NOT !

What can I say ? It is certainly not one of FLIR's finest designs, and considering how expensive they were when new, I would go as far as to say it is shoddy workmanship by the mechanical design team.

TTFN

Aurora

[CatalinaWOW]

Line noise of the type you are seeing seems unlikely to be a thermal contact to the Peltier issue.  Seems more likely to be an electronics issue, perhaps a dodgy ground or unconnected shield due to the construction shortcuts you have noted. 

Get your rest and keep enjoying this.  I sure am enjoying following along.   While I don't have CFS, I have noted that as I get older I can't keep up the intense thinking sessions I once did.  It's OK, just takes longer to get to the end.

[Fraser]

@CatalinaWOW,

I completely concur, the line noise is not what I would expect of a Peltier issue. It is the images that show a cool middle with graduated halo's radiating out to a warmer border to the image. But even this does not resemble what I would expect of a separated Peltier element.

 I have witnessed the cold finger in a cooled AGEMA 550 separating from the die and causing a similar graduated halo effect, but that is a completely different technology. I am not even certain that the Peltier element can separate or whether it is an integral part of the microbolometer die. 

I will be focussing on the noise lines for now as they are something I can investigate further.

Time to break out the oscilloscopes 

Aurora

[mikeselectricstuff]

I'm sure I've seen reference to a getter in some microbolometer info somewhere.
Of course the chemistry may be different to that used in valves, for one it's a direct connection as opposed to an induction coupling, so temp could be well controlled.

A quick search for microbometer getter found this very detailed article on packaging:
 
http://accessengineeringlibrary.com/pdfdownload/?contentId=advanced-mems-packaging%2Fp200199219970327001

 

[SeanB]

Phillips developed some non sputtering getters a while ago, mostly used in sodium and MH lamps as a getter that does not cause a thermal stress area on the outer glass. They are small, reasonably active and will not sputter material. That would be perfect in this application in a pocket inside the metal envelope provided you can get them hot enough during pump down to activate them.

[Fraser]

@Mike,

Great find, Thank you 

That document is very helpful in understanding the internal construction of the microbolometer and the out-gassing issues of the Peltier element.

They show the getter in the vacuum pump down port. I have now learnt about modern getters as well. Excellent 

Having seen that the Peltier element is soldered to the microbolometer, it is highly unlikely that my E2 has suffered Peltier separation  Happy days.

Many Thanks for finding this and sharing.

Aurora

[Fraser]

@SeanB,

I was unaware of getters in microbolometers or in proximity to sensitive electronics. I studied thermionic valves where the getters are brutal things 

Today was a good day.......I learnt something new. 

Aurora

[SeanB]

PIc here, with a little info.

http://www.lamptech.co.uk/Spec%20Sheets/Philips%20SOXPSG35.htm

Close up of the PSG pellet.

http://www.lighting-gallery.net/gallery/displayimage.php?album=2854&pos=21&pid=80275