Jenoptik IR-TCM-384 thermal camera - ‘Box camera’ Teardown

[Fraser]

Tim,

Excellent suggestion, I will investigate. It does have the appearance of up scaling mismatch.

There are two image stream display modes, RGB and IR-Data. RGB shows the cameras overlays and menus but IR-Data shows only the IR image data, no overlays or menus. The IR-Data mode is noticeably better imaging quality.

I will save an image to SD Card as well to see what quality is recorded. There are many options for image storage file type as well. PNG may be better than the default JPEG format.

Fraser

[Fraser]

I am uploading last nights images that I quickly captured from the camera feed.

The IC in the image measures 11.9mm across its plastic encapsulation and the torx head screw head measures 5mm across. Horizontal FOV appears to be approx 20mm at 50mm distance from lens. This gives approx 0.05mm per camera pixel, or put another way, 20 pixels per mm.

Distance of the PCB from the lens objective was approx 50mm.

I will do the diagonal hot wire pixel count and definition test later.

Note that some pictures are focused on the PCB tracks and some on the top surface of the IBM IC.

It was interesting to see that the hard disk IC imaged was running quite cool initially and whilst I was testing the camera, the IBM IC suddenly 'thermally erupted' in the image and shot up in temperature. This hard disk has been removed from service as it had a habit of becoming sluggish in its responses. I wonder if the sudden temperature escalation of this IC has something to do with that failure ?

Fraser

[Fraser]

Two more pics at the end of last nights test.

[Fraser]

A quick read of the manual has explained the two stream modes of RGB and IR-Data. I noted that IR-Data provides a better image quality.

RGB mode enables camera GUI menu overlays and image fusion to be seen on the control software image window. Saved RGB images contain no radiometric metadata. It is a mode used for remote control of the camera with a preview of the thermal image.

IR-Data mode displays only the pure thermal data stream from the camera with no camera generated overlays present. images are stored in an IRB format which contains full radiometric metadata.

It would appear that RGB mode slightly degrades the image in the window, and that saved, in order to accommodate the fusion of camera generated GUI displays. I will have to play with the IR-Data mode some more.

Fraser

[T3sl4co1l]

Really weird, those screenshots look like they've been pixel-resized multiple times.  If it were a single resize (with no filtering), the line doublings should be evenly spaced.  Instead I see patterns like, 1,1,1,1,2,1,1,1,2,2,1,1,1,1,2,1,...  There should never be two consecutive doubled lines in a single resize. 

And also, that.  Lack of filtering.  Makes things so ugly.  Oh well.

Tim

[Fraser]

I agree, there was immediately something about the images that looked weird to me. See above however. There seems to be some manipulation of the image to cope with the camera sourced GUI.

Also note that the saved images are 640 x 480 and the camera actually produces a 384 x 288 image ! Those two are not a good match in terms of upscaling !

The IR-Data mode is far more promising as it is presumably what is the 'pure' feed intended to be used in an industrial setting. I shall have to look at how the camera copes with generating a signal for the composite video, S-Video and VGA outputs. I suspect more mismatched upscaling or possibly the camera will drop the used FPA pixels to 320 x 240, then doubled for VGA output.

People sometimes like the extra pixels provided by a 384 x 288 pixel array but such can make thinks awkward when it comes to upscaling to a 'standard' resolution. This is why OEM's often use these 384 x 288 pixel arrays in the sensors 320 x 240 mode. ULIS accommodate these two modes in their ROIC as standard to assist in this matter.

Fraser

[Fraser]

Now for those readers who have patiently been waiting for me to release the cameras identity.........

The camera is a product of Jenoptik (Germany) so it is little surprise that it uses a European ULIS (France) microbolometer

The model ID is IR-TCM 384 that translates to "Infrared Camera - Thermal Camera Module 384 x 288 pixel imaging array".

The IR-TCM Industrial thermal camera module uses the same imaging core and main board as the Jenoptik Variocam handheld thermal cameras. Very good quality and in the same market segment as FLIR professional industrial cameras. Very expensive Germanium lenses are used to maintain good image quality. Radiometric accuracy is excellent. The IR-TCM is a 'hard core' industrial thermal camera similar to the FLIR A40 etc.

Datasheet is here:

https://www.jenoptik.com/cms/products.nsf/0/3922D918776B7BCEC1257775007F90C8/$File/ir-tcm_384_en_web.pdf?Open

These cameras squeeze the very best performance available from the ULIS  UL03-04-1 384 x 288 35um pixel A-Si FPA.

Fraser

[Fraser]

Pictures of the IR-TCM and handheld version Variocam. I also include a picture of the now familiar imaging core, as shown on the Jenoptik Core site.

Fraser

[Fraser]

These thermal cameras also appear on the market as Infratec products, a partner of Jenoptik.

http://www.infratec.co.uk/

http://www.infratec.eu/fileadmin/downloads/pdf/VC_HD/InfraTec-VarioCAM-HD-head-security-h-EN-mail.pdf

Fraser

[Fraser]

On the topic of image filtering, the camera shares a feature of other high end units in that it permits the level of image filtering to be set by the user. In the case of the IR-TCM module, it may be set to Off, Weak, Medium or Stong.

From memory, the stronger the filtering applied, the more likelihood of issues when imaging fast moving objects. In industry where the thermal data is key and pretty imGes are lowerpriority, the filtering is often set to Off or Weak in order to achieve the best effective frame rate from the 'sensor head' (the camera). This is likely why the images look poorly filtered. The filter was set to weak in the images I uploaded.

There is another program that I have for analysing the images captured in 'IR-Date' radiometric mode. It is called 'VarioAnalyze' and is intended for Close analysis of the images. The main capture and control program is called VarioCapture and it appears to be just an engineering program to set the camera up and capture images. It contains no report writing capability or niceties often found in consumer grade thermal camera software. It is more of a utility. Even setting the time and date must be done by activating the cameras on board GUI rather than by direct commands from the PC software. It is very much a remote control program for accessing the functions normally activated on a handheld cameras keypad. It does offer the advantage of image streaming and recording to the PC though.

Jenoptik now supply a new piece of software for their cameras. It is named IRBIS and may be more capable than VarioCapure. It's very expensive though so I will not be trying it out any time soon !

Variocapture does provide me with full remote control of all camera functions and I do have the option to build a RS232C remote controller for local control f the camera. I need to reverse engineer the command set first though.

So for those unfamiliar with the world of Industrial modular thermal cameras, I hope this thread has been enlightening. They can be useless without the host control software, built like battleships, very capable and are usually very expensive ! Buy them only if you know what you are doing. Mine cost me $250 and it was a risk. However it was a risk I was willing to take.

I will update this thread with further discoveries about these cameras, such as image quality, analysis software performance and the repair of the unit with a sticky aperture/FFC plate.

Not forgetting that little micro switch behind the SD Card port of course

Fraser

[Fraser]

Just so people are aware, I will not be reverse engineering these cameras beyond what is needed to get them operational and any areas that interest me. These are a controlled Dual Use technology camera and I respect that status.

Fraser

[Fraser]

Apologies for the slow progress on this thread. A health issue has recently heavily impacted on my ability to do much electronics related stuff. Normal service should resume shortly

Fraser

[Fraser]

For those interested in the control software aspects of industrial cameras, take a look at the InfraTec IRBIS options page found here:

http://www.infratec.co.uk/thermography/thermographic-software.html

This is not at all unusual in the world of Industrial thermal imaging. You buy a core image analysis program that actually contains all available options, but to activate the ones you want in your application, you have to pay for an upgraded licence. It can get VERY expensive ! $10K is not unusual.

So when someone complains abut the quality of free software provided with consumer grade and semi-pro cameras.... consider yourselves lucky to be given anything for free. Even FLIR TOOLS+ is a relative bargain at around $300 when compared to the software provided to industry at huge mark-ups. FLIR professional software is still available and still costs a small fortune. It is protected using the USB HASP HL dongle system due to its high value.

Third party software is sometimes also available. My NEC AVIO cameras can be remote controlled and their images analysed by either the OEM provided software, or some excellent third party programs. The NEC AVIO software has always been a bit clunky, much like this Jenoptik offering. Many NEC AVIO agents recommended a specific third party software that was very good. It was supplied by the 'White-Blue' company but sadly they discontinued it a few years ago. I am fortunate enough to own a licence for this excellent capture and analysis software. What is interesting is how many other thermal camera manufacturers actually used the software under their own branding. My licence works with them all so I managed to find the latest versions to use. The software cost several thousand pounds for a special, customer specific licence that unlocked certain features and support for certain cameras. Thankfully my licence is fully loaded so all features are unlocked 

Fraser

[WastelandTek]

get better soon Fraser

[Fraser]

Thank you

I spent a little more time on the camera and software today.

The control interface is a little clunky as no keyboard keys are used to replicate the cameras buttons and joystick interfaces. It is all mouse driven.

I have tested the camera with an SDCard and all is well. Images are stored on the SDCard so that option is enabled.

The camera has four levels of noise reduction as already stated. The best for everyday use is the 'Weak' filtering option as it does not noticing you impact upon imaging fast moving objects. The highest setting is definitely intended for static objects and such is stead in the manual. The filtering reduces the noise content of the images when small temperature spans are used. The camera is capable of 2C spans and with filtering switched off there is quite a lot of random noise content. This drops markedly when the filter is set to 'Weak' strength.

I noted in the imaging settings that there was an 'HQ interpolation' option that was selected. This may have been causing some of the weird artefacts in the original images. The cameras image output is upscaled by a factor of 2 before being displayed on the PC. This remains the case no matter what the condition of HQ interpolation. More investigation will be carried out in this area later.

The two streaming modes now make more sense to me. As I previously stated, the streaming is either 'RGB' or 'IR-Data'. In the RGB mode the images presented on the PC software window are literally the RGB video data that would normally be presented to the cameras LCD display or viewfinder (on the Variocam version). That is why the menu overlays are seen. This mode appears to provide a less pleasant image as it is literally an RGB video feed rather than raw, unadulterated radiometric data being read and displayed.
In the IR-Data mode of operation, the camera outputs all of the radiometric data needed to create an image on the PC. No camera generated GUI data is present. In this mode, the camera radiometric image data is used to create a clean and true thermal image in the PC window. The thermal span etc is shown as it is also supplied in the radiometric data stream. This data may be recorded as a data stream, video stream or as still images. The recorded data may be analysed using the VarioAnalyser program that is able to read the Radiometric file type directly. The VarioAnalyzer program also enables the user to save images in common file types such as JPEG, BMP and TIFF.
Various analytical tools are available n the camera and in the PC Analysis software. These include, but are not limited to Thermal profiling along a defined line, and an image Histogram. The analysis software is simp,e enough to use and even provides report writing tools. As true Radiometric data is used, the user may add temperature measurements to the images. Image palette, spans and centre temperature may also be changed to suit the uses needs.

The camera and software contains some interesting palettes and I must say I like some of the ones that I had not seen before. I do not see any option to create new custom palettes but it may be possible to add them to the analysis software. I will need to look into that.

The camera has behaved flawlessly throughout the testing and is a good performer. Clean images and very accurate temperature measurements. The Close-up lens is a bit specialist and limits the use of the camera to that of an inspection tool and thermal microscope. It images modern high density electronics with ease though.

I will hopefully find time to take a look at the combined aperture plate and FFC shutter tomorrow. It sounds like the stepper motor is having difficulty moving the aperture plate and so the drive current limit kicks in. The camera seems happy enough in other respects though.

I ordered another couple of Lemo connectors for the other camera today. Ouch that hurt in the wallet ! They are essential though so I just had to bite the bullet.

Later I shall set up the RS232 control link instead of using the Firewire. I can the. Send commands from the control software and capture them at the serial port. With some effort I should be able to capture many of the cameras control commands. Sadly not all though, as I feel sure there will be commands that the control software does not produce. The command set for FLIR PM series cameras are extensive and these cameras are likely similar.

I may, or may not, build a RS232C remote control for these cameras It would certainly be easier than using the mouse driven PC software GUI. I need to build a remote control for my PM series cameras first though as that is a higher priority.

I have not decided on the future of these cameras yet. I already own the FLIR A40 and recently purchased the genuine Close-up lens for that camera. I may keep one of these Jenoptik cameras and sell the other as you only need so many close-up cameras in the lab

I think I got quite a bargain with these Jenoptik cameras. Luck was on my side and good fortune smiled upon me. To find the proper remote control and analysis software on an agents web page, when such is not available from Jenoptik, was a real surprise and it was the difference between a good buy and a disaster !

Fraser

[Fraser]

I almost forgot...... the tiny microswitch behind the SDCard cover screw....

I still do not know what it is for. It does not reset the camera, enable any additional options or have anything to do with the SDCard operation.

If it is pressed in when power is applied, the camera will not start the boot sequence. I am wondering if the switch is used for firmware updates and forces the camera to look for a firmware file on the SDCard. This type of firmware upgrade process is not uncommon on thermal cameras. The firmware is loaded from the SDCard rather than from a Host PC. The firmware has a very specific file name that the upgrade routine looks for. If it does not find it, nothing happens.

For the moment, the microswitch function remains a mystery.

Fraser

[WastelandTek]

It sounds like the stepper motor is having difficulty moving the aperture plate and so the drive current limit kicks in.

I am probably revealing my ignorance here, is lubrication even an option in this application?  or is this a function of tolerances?

[Fraser]

WastelandTek,

A semi seized bearing or hardened grease are known issues in some camera bearings but this particular situation is different.

The Jenoptik camera uses a system that I have not met before. The large aperture plate/disc/wheel is like a giant gear wheel and driven on its outside edge. It has a central spindle that passes through a stainless steel sleeve. There is no oilite bearing as I would expect of a spinning shaft scenario. The reason is that the aperture plate only moves occasionally and at low speed. There is the lightest of lubrication on the bearing surface. The aperture wheel is unlikely to seize.

The way the camera uses the aperture wheel is novel. Different apertures may be selected by the camera to suit its needs. The same wheel has a blank high emissivity area that is used as the flat field correction reference. Most conventional designs use a simple flag as the reference. When the camera needs to carry out a FFC it drives the aperture wheel around to the blank section and after FFC completes, the wheel is driven back to the required aperture. All this happens at a relatively slow speed.

If I am honest, I prefer the 'flag and solenoid' design but Jenoptik must have had their reasons for combining the aperture plate with the FFC reference area.

The aperture plate wheel is driven by a small stepper motor via a wide nylon gear. It is possible that the gear is engaging too tightly with the aperture plates edge. If the camera has suffered an impact the weight of the stepper motor can bend its mount causing misalignment of the drive gear with the aperture plate teeth. I shall investigate this first. There is also the potential for an issue with the stepper motor driver.

Never be afraid to make comments or suggestions though..... all are welcome and none are 'silly' questions or suggestions as we learn by asking and discussing matters. Hence this rather long winded explanation of the system used in the camera.

Watch this space

Fraser

[Fraser]

The remaining camera fault has been resolved 

You will recall that the camera was failing to correctly position the aperture plate and that the plate is like a large gear wheel, driven on the edge by a stepper motor.

I suspected misalignment of the stepper motor due to the force of the impact to which the camera was exposed. Whenever inspecting a mechanical or electronic device that has been dropped or exposed to high impact energies, you should look at all heavy items to see whether they have moved or the chassis has bent. In the case of the Jenoptik camera, it is very well built and should withstand reasonable abuse. It does use standard mechanical components in the form of the stepper motors though.These COTS stepper motors/actuators are made with a standard sheet metal mounting plate that secures them to the very rigid JenoptiK chassis.

When the camera experienced the high forces travelling through it during the impact event all parts will have become far heavier due to G forces upon deceleration. As suspected, these forces traveled through the two stepper motors and their mounting plates. The focus stepper linear actuator is very well supported by its bracket and the actuator shaft. It is therefore quite stable. The aperture plate stepper motor is supported by only its front face mounting plate that is made from mild steel. The screws had not moved but the engagement of the nylon stepper motor gear with the aperture plate was far too tight for easy 360 degree rotation of the plate.

During the impact event the aperture plate stepper motor had tilted outwards on its thin mounting plate causing the gear to effectively dive into the aperture plate gear teeth. The mounting plate had not recovered its correct shape after the impact event. This left the gear engagement so tight that no tolerance existed for radius variation in the aperture plate. The assembly then jammed at certain points in the rotation sequence causing an over-current situation in the stepper motor driver.

Upon close inspection, the distortion in the stepper motor mounting plate was very small indeed. the slight tilt in the mounting plate was corrected and the motor refitted. The gear mesh and backlash was carefully checked and adjusted to optimum.

Upon re-assembling the camera it was tested and the aperture plate now rotates perfectly and normal operation has been restored.

I attach pictures of the parts discussed in this post fro better insight into the design.

Fraser

[Fraser]

Some details of the close-up lens that is fitted to these cameras.

The lens is marked Jenoptik IR 2.0/50mm LW CLOSE-UP LENS 707033

This is not a standard lens with either a close-up auxiliary lens or mounting extension tube. It is a dedicated close-up lens.

The lens is a heavyweight at 500g and the lens elements are constructed from pure Germanium. The number of lens elements is not known at this time.

The weight of the lens lead to the mounting lugs being distorted during the impact event. Thankfully the damage is not to the lens elements and the lenses are otherwise undamaged. These would have been very expensive lenses so I am surprised that they were not salvaged by the original owner. Maybe the lens mount lug damage caused them to be condemned ?

The good news is that the damage to the lens mounting lugs is repairable if required, but the lenses seem to seat correctly and securely as they are for the moment. The lens mount locking ring uses a helical lug retainer which explains why the lens is still retained well.

The lens has a pair of gold contacts that mate with a pair of gold spring loaded pins in the cameras lens mount. I believe this is the Lens ID chip and maybe more. Measurements on the two contacts show 2 Meg Ohms so it is not a simple thermistor, RTD or thermocouple as might be expected. The camera has to know the lens identity in order to set its calibration table offsets. The camera is also capable of reporting whether it has been calibrated with any given lens attached (identifed by serial number).  It could be that the lens contains a circuit that reports lens type, serial number and temperature via the two contacts.

All in all, very nice quality professional close up lenses.

Fraser

[Fraser]

An interesting discovery regarding the Jenoptik close-up lens ....

Yes it IS a close-up lens but is it a 'ground up' close up lens/macro design ? I think not.

From the first time I saw this lens, I thought it looked unusual. The front element looked like it was 'tacked on' rather than being an integral part of the lens design. Could it be that the lens is just a standard type with a close-up element attached to the front ?

Now there is nothing particularly wrong with using an auxiliary close-up lens, but to supply a "close-up" lens that is in fact a standard lens with a close-up element 'bolted' on the front struck me as a bit unusual. Most OEM's offer the standard lens options plus a number of auxiliary single element close-up lenses to mount on the front of the standard lenses. In fact Jenoptik do offer this exact option with their lenses. The exceptions are usually Macro lenses and microscope lenses. These are designed to fulfill their specialist roles from first design, and not just an adapted standard lens. I can find no listing for my exact lens so it is a bit of a mystery. It was also unusual to see that no mount thread was provided fro a lens protector. This can be an important accessory for a close-up lens to prevent liquid splashes or impacts on the expensive AR coated Germanium lens element.

I was making an assumption about the lens design though so needed to prove if I was correct. A quick look at the area around the 'objective' element further raised my suspicions. There was no standard internal lens retaining ring but there was a very familiar looking slightly concave lens element that looked to be part of a lens ring/sleeve that could unscrew. I tried to unscrew the front ring but it has been locked in place with lacquer or glue on the threads...... damn! Oh well, maybe applying some heat and more force will budge it ?  Before going off in that direction, I checked the other camera lens to see if it was also 'thread locked'. I was in luck.... with very little force, the front ring unscrewed  Only a very small amount of lacquer/glue had been uses at the rear of the threads.
Upon removing the front lens retaining ring, I could see that the lens was not part of it and was sat in a recess in the lens barrel. Not what I had expected to be honest. I expected the front lens ring to have the front lens element secured to it. I carefully removed the front lens and to my joy I could see a standard looking Objective lens sitting behind it  The bulbous objective lens appears very similar in appearance to the standard Jenoptik 50mm lens objective element. The close-up lens element and its retaining ring is not needed in terms of the other lens elements and their retention in the lens barrel.

It is too late here in the UK to start doing tests on the lens without the front (close-up) lens element fitted, but I very much suspect that I will find the camera will focus fine without the close-up lens and that it is just a standard 50mm Jenoptik Lens in a modified lens barrel to accommodate the close-up element. If this is in fact the case, I will be very pleased as the close-up lens somewhat limited the versatility of these cameras. If they can be set-up as standard OR close-up lenses they will be far more useful to me 

There will be a very slight change in the transmission characteristics of the lens without the Germanium close-up element fitted, but any difference is likely to be minor and of no consequence to me. It can be tested and quantified easily enough using my Black Body test sources anyway.

Well dear readers, it is almost 4am here in Blighty so time for bed. I go to my bed a happy chap in light of tonights discovery. Hopefully I will prove that the lens works like any other standard lens once the close-up element is removed.

Just enough time to upload the pictures......


Fraser

[Fraser]

OK, I have tested the camera with the front 'close-up' lens removed.

Well I was wrong, the lens does work well without the 'close-up' element ..... but this lens is by no means standard. It has a lengthened lens barrel compared to the standard lens. This means that it is still a close focus lens, just not as close as with the close-up lens fitted !

The improvement in focus distance is worth while though. The lens is far more useful in general PCB work when the close-up lens is removed.

The camera has adjustable focus so I have listed the optimum focus distance for its closest and furthest focus settings.


Focus distance test.....

With close-up lens

Set to max Far = 40mm
Set to max Close = 20mm


Without close-up lens lens

Set to max Far = 320mm
Set to max Close = 50mm

I took some quick pictures of the ubiquitous hot coffee cup during testing to show the different colour palettes. I tried both RGB and IR-Data modes and IR-Data images are much cleaner than RGB, as suspected. I will upload some of the IR-Data pictures later but this was not a formal test. In the mean time I am uploading the RGB images as they show the colour palettes. The vertical bar on the front of the cup is caused by my radiated heat as I was in front of it.

Fraser

[Fraser]

More of the RGB mode pictures

Fraser

[Fraser]

OK, some RGB images Vs IR-Data images that I have loaded into VarioAnalyzer and then saved as JPEG files. Some loss of quality may occur due to the compression but without VarioAnalyzer .IRB files cannot be read.

I am still playing with the software so also tried the auto 'Clean', 'Soften' and 'Sharpen' image improvement tools. I set maximum strength on them to see how severe they are. They will follow the RGB IR-Data(converted to JPG) comparison images.

I decided to leave the HQ interpolation mode switched on as I assume it is supposed to improve the image quality ? I shall have to do some test on that later.

In the following images, the RGB images contain the cameras GUI but the IR-Data images are clean except for the temperature scale.

Fraser

[Fraser]

Clean, Soften and Sharpen filters applied using the VarioAnalyzer edit mode set to the highest effect setting.

First image is the untouched version, followed by Cleaned, then Soften, then Sharpen. The edits are individually applied to the untouched image and not sequentially to one image as that would cause weird effects !

Fraser