EEVblog® Electronics Community Forum
Products => Thermal Imaging => Topic started by: 龟苓膏 on February 13, 2025, 03:39:29 am
-
Brothers, I have an E75 here, but the temperature measurement is inaccurate.
It differs by 2-3 ℃ from the other one. The other one is a brand new and unused thermal imager, so the temperature should be accurate.
May I ask how to access the E75 temperature calibration page?
Connect to WIFI and log in to the http://192.168.5.104/ (http://192.168.5.104/) 。The operation interface of the thermal imager inside does not have a temperature calibration interface.
I'm going to borrow an E76 and a blackbody from a friend to compare and test
If you can tell me how to enter the calibration liquid level, I would be very grateful
-
You are comparing a professional thermal cameras measurements to those of a budget thermal camera. Which is likely to be most accurate ? Just because a budget thermal camera is new, it does not mean that it is accurate enough to diagnose a calibration error in a high specification E75. You need to check accuracy against a source of known accuracy and ensure that the camera is configured with the correct Emissivity values, distance to target and ambient temperature. This is the minimum required for an accuracy check.
The calibration utility on FLIR cameras was removed many years ago (back in the days of the E4 hack). Calibration requires that the camera is returned to a FLIR calibration centre and it is very expensive. It is very unlikely that your E75 has drifted out of calibration.
Borrow a decent Black body temperature reference and put your mind at ease on this matter.
Note that a measurement tolerance of “+/- 2 Celsius or +/-2%, whichever is greater” is the standard specification for many decent thermal cameras. Most are actually more accurate than this stated specification. Budget thermal cameras can have a specification of +/-3 Celsius and +/-3% or even worse in some cases. I have found several budget cameras that barely meet this accuracy specification. When comparing two thermal cameras, you have the obvious potential situation where they can deliver divergent temperature measurement readings.
Fraser
-
Thanks to Fraser,My friend's E76 also arrived today.
All parameters are consistent, testing the position of the same point. The temperature difference measured is within 1 ℃.
After another friend's Blackbody arrived, I tested it again.
-
Good news :-+
Fraser
-
Hi，Fraser
After comparison and disassembly inspection, it was found that there was a problem with the AutoCal link of the lens, which caused the shutter refresh function to malfunction.
When using the AutoCal function manually, you will hear a clicking sound with a delay.Causes the screen refresh time to be too long.
I searched all major websites but couldn't find this component.
Fraser, do you know where to find it? Or where to find this shutter motor?
-
Sadly I do not know of a source for that part outside of a FLIR service centre. Last time I checked there was a $400 fixed price repair fee for such repairs :( This assembly is just a mechanical mechanism though. If part of it has developed a fault, it should be possible to repair it, just like fixing a mechanical clock. The rotary solenoid is a relatively simple device comprising a fixed field coil with a magnetic core that rotates through a fixed angle within the field coil on a shaft. If there is a problem with the shaft or its bearings, this should be repairable. The rotary solenoid rotates through a fixed angle and acts upon a linkage to the FFC flag. The linkage and flag arm have pivot shafts that can be damaged but should be repairable. The actual FFC flag arm can become sticky in its operation for several reasons. These include oil contamination of sliding surfaces, distortion of the flag arm causing undesirable contact with its surrounding chassis or the presence of foreign body contamination between the flag arm and its surrounding chassis. Always inspect the rotary solenoid to make sure that no magnetic metal contamination is trapped between the field coil and magnetic armature. All of these issues can be diagnosed and repaired through careful disassembly of the FFC flag assembly. It is much like working on an old mechanical clock mechanism. A methodical approach to disassembly and reassembly is needed and take LOTS of pictures to aid reassembly ! There are very small small pivots, bearings and drive parts. With care the assembly can be refurbished to full operation again. FLIR simply replace the whole assembly as the time needed to refurbish it exceeds it value. If you, or someone experienced in working on small mechanical mechanisms, has the free time to work on it, you will save yourself a lot of money. The worst case scenario with rotary solenoid driven mechanisms is a failure within the field coil winding. Whilst repairing detached end points where soldered to the terminals is easier, repairing the coil would involve rewinding it and that is not a minor undertaking. If a camera is dropped, the heavy components can distort their plastic mounts during the event and this can tear the field coil away from its end point attachment. This is not a fault compatible with your symptoms though :-+
A sluggish or sticking FFC flag is normally due to pivot damage or contamination of the mechanism with debris or oil. I hope that your assembly can be repaired. Trying to replace the a rotary solenoid will be challenging. Finding something close to the size and specification of the FLIR part will be the first hurdle. Actually fitting it and making it mechanically compatible with the FLIR FFC flag mechanism would be the second hurdle. I am not saying that such is not possible (I would certainly undertake such in the case where no other option existed) but it could be quite a challenge. Rotary solenoids are not rare but it is finding the correct size, correct shaft design, operating voltage/current and rotation angle that is critical to correct operation in the FFC flag assembly.
Good luck with the investigation of this fault. Much can be learnt before disassembly by carefully watching the FFC flag arm operation and noting any contact points or strange noises coming from the mechanism. As a simple fix, you could try using compressed air to blow through the assembly in case there is contamination within (plastic debris or parts that have broken off of the assembly) Make sure to catch anything that flies out under air pressure as you may need it if it is a pivot or other part that has broken off the assemblies internal parts. The risk with using compressed air is the unintentional distribution of any lubricant within the assembly and this can just add to your problems if it gets where it should not be. Using a sensible air pressure can avoid this problem though.
Fraser
-
Have you observed the voltage applied to the rotary solenoid to ensure that this is not a drive issue rather than a mechanical problem ? Whilst rare, it is worth checking. The drive usually fails rather than causing excessive FFC event duration, but if the rotary solenoid is bistable it needs a reversing of drive polarity to disengage the FFC flag from the optical path. If the disengagement drive is intermittent it could cause your symptoms. This is somewhat unlikely though.
Fraser
-
Thank Fraser. :-+ :-+
I'm glad to see your reply. Unfortunately, I am not good at mechanical repairs.
So I still plan to see if I can find the same components.
I saw damaged lenses on a second-hand platform in China. I plan to take a chance and see if I can buy a component with a normal shutter.
After confirming that there are normal components, I will challenge myself and try mechanical repair.
-
I have tried several directions to test the opening and closing of the shutter.
The first type is vertical and inverted testing, where the shutter opens and closes smoothly without any delay. I can hear a coherent sound of 'click' and 'click'.
The second type is the positive and negative horizontal test, where the shutter sound is a "click" or "click" sound, but after about half a second, a slight "click" sound can still be heard.
Take out the shutter motor separately and manually toggle the switch, only to find that a part of the shutter will eventually get stuck and then return to its original position to hide.
And I found that during the positive and negative horizontal tests, automatic calibration will calibrate a temperature with a very large deviation. After replacing my friend's normal camera, there has been no such malfunction.
So I plan to solve this problem first.
-
Thank you for the update.
If you are not confident in your own mechanical abilities in this particular case, you could visit a local clock and watch repair shop and ask them to look at the assembly. They are used to working with solenoid operated mechanisms that contain small parts and pivots. If a plastic pivot is broken, they can drill a hole in rather assembly and fit a new brass pivot. If the rotary solenoid is the fault, they may be able to see where the fault lies and fix it. In the West the cost of engaging the services of a watchmaker would be very expensive but you may have better luck locally.
If I were approaching this camera fresh in for repair, I would first check that all was well with the voltages being sent to the FFC flag assembly. In the case on monostable rotary solenoids, the FFC flag would remain in the optical path only as long as the voltage was present on its terminals. A spring would pull the Rotary solenoid back to its ‘parked’ position once the drive voltage disappeared. The other type of rotary solenoid uses a bi-stable mode of operation where a voltage of one polarity is applied to make it move the FFC flag into the optical path, and the opposite polarity of voltage is used to return the rotary solenoid to its ‘parked’ position. The duration of the applied voltages in a bi-stable system can be relatively short as it is system that is stable in both of its operating positions.
Once I was certain that the camera electronics or firmware were not causing the unusual behaviour of the FFC flag(s) I would investigate the FFC flag assembly, as you have clearly done. I see two rotary solenoids in that assembly. Is this a two part FFC flag with both solenoids operating during an FFC event ? Sometimes a second solenoid is used to introduce an optical aperture for high temperature ranges so I am unsure of the configuration in you case. A split FFC flag looks likely in the confined space though. If you have split FFC flag (leaves) it is a well known issue that if pivot lubricating oil or lens thread grease gets on the flag leaves, they stick together when an FFC event occurs. This contamination can slow FFC flag disengagement or even prevent it. The FFC flag leaves may be cleaned of contaminants with Isopropyl alcohol but care is needs as they are fragile. Again a clock and watch repair shop could do this operation for you.
Fraser
-
I just saw your last post. So it looks like the flag mechanism itself is the cause of the problem and not the rotary solenoid ? If so, this is good news. You just need to find someone who is used to repairing small mechanical assemblies. Get them to dismantle, inspect and clean the flag mechanism to stop the flame sticking. It could be something as simple as a mis-positioned spring in the mechanism jamming a pivot. Hence why I suggested a local clock and watch repairer to help you :-+
Fraser
-
With regard to calibration errors….note that the camera uses the FFC shutter flag as a temperature reference to fine tune measurement calibration. If the FFC event is having issues, this can effect the calibration of the camera.
Fraser
-
Hi Fraser,
Communicating with you has gained me a lot of experience.
Yesterday, I made up my mind to disassemble the mechanical part. It turned out that the structure of the mechanical part was also very simple.
After reassembling, the problem has been solved.
I bought a blackbody on a second-hand platform. I calibrated it with my friend's new E76.
The blackbody temperature can stably reach the set temperature value within an emissivity of 0.96 and a distance of 1 meter.
Then I used this blackbody to calibrate E75. Without E75 Auto Cal, the measured blackbody temperature value will differ by about 0.5℃. Once Auto Cal is turned on, the difference will remain stable at 0.3℃。
Fraser . Can you help determine where the problem is?
-
What do you think is still wrong with your camera ? The camera is producing measurements that are well within its specification. Chasing small variations in a measurement will likely waste your time. Thermography is not a precision measurement technology. For precision measurements you need to be using a contact measurement sensor, such as PT100 RTD’s.
Fraser
-
For readers confused by the term “Auto Cal” in relation to the FLIR E75 camera (and most E7x cameras)…..
It is normally necessary to have a thermal camera and lens calibrated together to achieve good measurement accuracy. Attaching a lens to a camera with which it has not been calibrated can lead to measurement errors due to the differing transmission characteristics of different lenses. FLIR Auto Cal addresses this issue by loading calibration data into the camera for whatever lens is attached. It is also possible to save the correct MSX alignment for that particular lens. As a result FLIR offer various interchangeable lenses for the E75 that do not require the camera to be returned to FLIR for a calibration run to capture the lenses characteristics in the cameras lens calibration table. This is a positive development as calibrating a new lens to a camera used to be quite expensive. That said, the lenses for the E75 are very expensive so FLIR likely still make their normal profit from lens sales but buyers avoid the inconvenience of sending their camera away for lens calibration.
See the FLIR Auto Cal document for details of the procedure and the calibration target used during the process.
https://flir.custhelp.com/app/answers/detail/a_id/1570/~/flir---exx-and-t5xx-cameras-changing-and-calibrating-lenses-%28autocal%29
Fraser
-
Just in case anyone is wondering about the claimed measurement accuracy of thermal imaging cameras that were used during the Covid-19 pandemic…..
Thermal imaging was seen as a fast and effective way to check individuals or groups of humans for evidence of fever (elevated body temperature). Whilst this whole approach was deeply flawed and not particularly reliable, efforts were made to improve the accuracy of thermal imaging systems to make their measurement of humans more accurate. I will not detail the challenges of estimating a humans core temperature using extremities and “hot spots” on the face (tear duct region of eyes) but needless to say, the thermal imaging of humans approach faced serious challenges if it were to be effective.
In order to improve upon the nominal +/-2 Celsius or +/-2% (whichever is greater) measurement tolerance of a good quality thermal imaging camera, manufacturers had to significantly improve their camera calibration routines, camera temperature stability and FFC event accuracy. Even after making these improvements, the measurement tolerance of even the better human temperature measurement thermal imaging cameras was +/-0.5 Celsius. The Chinese Government had set a tolerance target of just +/-0.3 Celsius for their domestic thermal camera manufacturers to meet. In order to improve the measurement accuracy to that required by the Government, manufacturers introduced Blackbody thermal references to their human temperature measurement solutions. The temperature and Emissivity of the Blackbody emission plate was known and the Blackbody was positioned in the cameras field of view. The required measurement corrections could then be calculated in software to ensure the best possible accuracy of thermal camera measurement. There will always be an error introduced by the varying Emissivity of the target but the +/-0.3 Celsius measurement tolerance was achieved.
Before anyone gets too excited about this. The cameras that were adapted from general use models to accurate human temperature measurement models were modified in both hardware, firmware and calibration. To achieve the +/-0.5 Celsius measurement accuracy, most used improved FFC hardware/routines to better self calibrate the camera during use. The calibration at the factory was greatly enhanced with more reference points over a narrow range of temperatures. A human temperature measurement camera usually has a restricted measurement range of around 20 Celsius to 45 Celsius over which its specifications for measurement accuracy are valid. The early attempts to improve the accuracy of standard general purpose thermal cameras failed miserably and were rejected by the Chinese Government. The models that actually passed the Government testing were greatly enhanced over the standard general purpose models. This is one of the reasons why these human temperature measurement cameras cost significantly more than general purpose models (there were other, profit orientated, reasons of course!) Some thermal camera manufacturers tried to “cash-in” on the Pandemic by releasing general purpose thermal cameras that used a Blackbody” reference source to improve measurement accuracy via software. Some of those cameras will pass the +/-0.3 Celsius measurement tolerance specification, whilst others will not. YMMV.
So in Précis, do not expect the measurement accuracy of a high quality human temperature measurement thermal camera from a general purpose camera, even if it’s menus include a human temperature measurement mode ! Read the manufacturers measurement accuracy specification and work with that when using the camera.
Fraser
-
Thank you Fraser.
We tested again in Blackbody tonight, Keep a good distance between E75 and Blackbody.
The test results are satisfactory, and the E75 can maintain an accuracy of ± 0.2 ℃.
Thanks to Fraser for popularizing the functions of AutoCal. I always thought AutoCal was the calibration function of the E75 shutter
-
Camera “self calibration” is normally called Flat Field Correction (FFC) and this uses the FFC flag positioned in front of the microbolometer to provide a thermally flat reference for both pixel level equalisation (flat field) and measurement fine tuning. The temperature of the FFC flag is known as a calibrated temperature sensor near to it reports the air temperature around the FFC flag. It is this “known flag temperature” that can be used to fine tune the cameras measurement temperature but it does not check for any changes in the transmission of the lens system (optical path). Lens calibration characteristics are used by the camera to determine the effect of a particular lens on the temperature measurement and offsets applied in the calculations.
Thermal cameras normally carry out an FFC event regularly at initial start (as the microbolometer warms up) but then reduce the period between events to around 120 Seconds during normal use. Some cameras monitor the internal temperature of the unit and instigate additional FFC events as and when needed to maintain measurement accuracy. A thermal camera is usually at its most accurate for temperature measurements immediately after an FFC event (assuming a sensible camera warm up/temperature stabilisation period has elapsed). This is also when the pixels appear at their best in terms of individual pixel thermal drift. Those wanting the best possible accuracy from a thermal camera will often instigate a FFC event manually just before the measurement. This is why thermal cameras often have a manual FFC function via a key press. If the image begins to show significant pixel drift, the user can also use the manual FFC event to clean it up.
Fraser