Keysight U5855A TrueIR thermal imager

[DaJMasta]

I think I'll try my hand at it once I get a chance, it is a welcome addition.  Wanted to mention the other thing I noticed in my short time using it  - the relatively narrow field of view.  It's not as narrow as some, but I noticed that the Flir E series generally had much wider fields of view than other brands, and you can see it in using the unit.  It won't get in the way of spot checking, and it effectively gives you more detail using the same resolution at the same distance, but it means you have to scan around somewhat more and it feels very different in some respects when taking pictures vs. a regular camera.

To give you a rough idea, It's going to be near impossible to take a selfie with this thermal camera, with arm fully outstretched, I was unable to get my entire head in frame (and no selfies are not a thing I usually do  )

[marshallh]

For anybody stuck using the Keysight TrueIR software -- I just loaded the sample images into NEC AVIO Infrec Analyzer and it displays the images perfectly. Since these camera lines are both designed by NEC it makes sense.

I was extremely unimpressed by the Keysight software. It is inferior in a variety of ways to the NEC software, albeit having a slightly more streamlined UI.

For example, the NEC software is able to perform high quality resampling so you can properly resize thermal imagery to any resolution:


Keysight TrueIR - notice how pixelated it is. It doesn't even perform any processing beyond palette mapping.


TrueIR also can only save to a tiny 320x240 JPEG with lossy compression. If you export a report to PDF the resulting image is riddled with dct artifacts and nearly unpresentable.
NEC software can export to an uncompressed BMP with any report material included. However, it does not support custom palettes.

[Fraser]

Thanks for the tip on the software marshallh. I own some professional NEC AVIO software and will have try that as well. I will also use the Infrec software now that I have seen the superior resaampling performance.

Fraser

[Fraser]

Dajmasta,

The common field of view on industrial thermal cameras was 24 Degrees on a 320 x 240 pixel sensor array. The manufacturers considered that optimum for the resolution. Wide angle and telephoto lenses were made available for applications that required such.

FLIR use wider angle lenses than nominal on the Ex series, which is a mixed blessing. You can lose detail in the image but you can use the camera in more confined environs.

Fraser

[adranp]

Hi :Fraser,

If possible, I want to know your opinion regarding Flir E40 (upgraded to E60) vs this Keysight U5855A. I know you have both.. Was looking for a E40 but this caught my eye and it seems very similar in performance..

Thanks in advance..

[Fraser]

The FLIR E40 upgraded to E60 specification outperforms the U5855A in all areas. They are not really equals (Apples and Oranges scenario) as the U5855A contains a 160 x 120 pixel microbolometer, has no Wireless connectivity, no video recording, no interchangeable lens and tops out at 350C. It also uses the slightly more noisy Amorphous Silicon (as opposed to VOx)  microbolometer technology.

The U5855A is an excellent camera in its class and costs far less than a used FLIR Exx series camera if bought from the Keysight dispodpsal store. That makes it an excellent choice for those not wishing to spend too much, yet still have an excellent quality thermal camera with the many features that the U5855A camera offers.

Be advised, Fine resolution mode (Super resolution through image sticking) does not provide a 320 x 240 pixel image equal to that of a camera equipped with a physical 320 x 240 pixel microbolometer array.

Fraser

[Fraser]

Initial non invasive tests on the poorly U5855A.....

1. Main system Processor is operating normally, even after fault appears
2. Microbolometer and its integral ROIC appears to be operating normally
3. Pixel Noise presence in image when fault occurs suggests signal path from microbolometer to video processor is unbroken.
4. Thermal image contrast is totally lost during fault. Visible light image is present.
5. Cameras auto ranging is ramping during normal operation which is not correct.
6. Temperature measurement is reading high (38C) during fault but correctly at other times.
7. Video processor appears to crash at times causing image corruption and freeze.
8. Camera image processor appears to reset during normal use
9. The fault is thermally responsive as it persists until the camera has remained off for a period.

All in all an interesting fault I already suspect a potential DC-DC converter issue. Either noise on the rail or varying voltage outside of spec. Such causes havoc in systems that contain ADC's and video processors. The symptoms can be quite unpredictable and chaotic in appearance.

The first check will be to substitute known good boards into the faulty unit to isolate the board that contains the fault. The risk to the good boards is low in this particular case. Once the failed board is identified I will download all the IC datasheets and identify all power supply rails. All power supply rails will be monitored for unexpected activity.

Without a set of schematics I could reverse engineer the failed board but that takes time I do not have spare at the moment. I will most likely carry out thermal profiling of the board to check for components in thermal distress and then compare key test points with those of a known good unit whilst both are running.

More on this repair will appear in a separate thread.

Fraser

[adranp]

:Fraser Thanks a lot for your reply.. Just one more thing.. If the E40 would be stuck (not upgradable to E60) would you take it vs U5855A ?  Price-wise the U5855A is unbeatable..

[onesixright]

For anybody stuck using the Keysight TrueIR software -- I just loaded the sample images into NEC AVIO Infrec Analyzer and it displays the images perfectly. Since these camera lines are both designed by NEC it makes sense.

Looks awesome! I'm not truly familiar with their software. But very interesting !

Is their software free or only paid? If free, any place for download (did google, but didn't get much wiser)?

Thanks!

[Fraser]

The situation does not exist..... All versions of E40 may be upgraded to 320 x 240 resolution.

Given the choice of a standard non upgraded E40 Vs the U5855A, with both at the same price point, I would choose the E40 as it is still the better camera even when not upgraded....... BUT you will not find a decent fully operational E40 for $600 any time soon. A good used E40 generally costs at least three times as much as Keysight are asking for their U5855A. For some people that is too much to pay for the E40's advantages.

I forgot to mention, the Exx series are 60fps image update rate, the U5855A is <9fps. Some applications benefit from the higher frame rate. International shipping of an Exx officially requires a DUT Export licence, the U5855A does not require such.

Fraser

[adranp]

:Fraser Thanks for all clarifications.. I was under the impression the newer versions of e40 are un-upgradable.. Really thankful for your replies..

[Fraser]

The new E40 needs a different approach but may be upgraded. Take a look at the E40 upgrade thread for further detail.

The latest version of E4 is not currently upgradeable.

Fraser

[Fraser]

For those interested in the NEC AVIO NS9500LT-B image analyser software, I have just gone through the ordering process for free download.

The download appears to require you to be a current owner of a NEC AVIO camera so I just input details of my TH7102WH camera. The download is provided after these details are verified. I am not sure whether the software is available without going through this process. The download is 43MB.

So whilst this software is 'Free', it may well require you to already be a customer in order to download it.

http://www.infrared.avio.co.jp/en/support/thermo/download/ns9500lt-dl/index.html

I own the full (non lite) PRO version of the software already and the more advanced 3rd party R&D analysis software that cost an arm and a leg. I suspect the Lite version will be fine for most though as the more capable software packages just provide more R&D type tools and remote control of the cameras that they support.

The NEC AVIO software download page provides many programs to download and use with their cameras. Some are free, others are upgrades to licensed programs. The NEC AVIO software naming convention and download pages leaves something to be desired. I suggest downloading the free programs to see if any work with the U5855A and appeal to you.

http://www.infrared.avio.co.jp/en/support/thermo/download/software.html

Fraser

[exor]

As I don’t have any NEC AVIO camera I just input my Keysight camera type and serial number. I also told that I would like to test if their program (free NS9500LT-B) it is better than Keysight program for picture handling and they were kind and sent the link with ID and password. Excellent service!

Not tested the program yet, currently downloading..

[Fraser]

Excellent news 

The U5855A is partly their product, but they need not be so friendly so Kudos to them 

[DaJMasta]

Got my charger today and ended up spending a couple hours with the camera.  Battery life is quite good, maybe 3h or so in total running and a chunk of it with max display brightness and I'm down one bar on the indicator.  Also, the 1.30 firmware adds the palate I was thinking of, it's called "Iron" though there was a "Hot Iron" variant that was even on the old firmware.

Only time I notice the sensor noise is really low dynamic range where it tends to be warmer, but there's occasionally a sort of crosshatch pattern in that situation, otherwise the fidelity has been great, hard to believe the sensor is only 160x120.  I'm getting a hang of figuring out when things are reflections or what effects are the result of the sun or other things too, it's a new world to get used to seeing in some respects.

I hooked up the RCA output to a cheap capture card ($6 or so) and got video on the PC with no issues, the screen goes off when it's sending video, but the video is an exact copy of what's normally on the display.  It may be my imagination,  but the framerate almost looks higher on the video connection - if it is, it's not by much, but it seemed smoother than I'd expect from 9fps.  My unit had a broken strap tab (making it a good chunk cheaper!), but I got a small piece of rope around the base and through the tab on the other side and it works just fine.

For some reason, my readings seem to be a little high - 3C or so at body temperature and a bit more at the high end of the range.  I think this is probably due to my settings, I'm running all defaults and while the ambient temp and humidity are about right, I'm not sure everything is set properly, will have to look into it, the emissivity settings were correct at least for the skin tests, and the relative differences seem to be right, best I can tell.  One notable exception to running hot is foliage - it seems to read very cold most of the time.  Probably partly because of the water transfer out the bottom of leaves (this was in the evening) and partly because of the material itself, but it was interesting to see negative number readings on a warm day outside.

[Fraser]

I am pleased that you are pleased with your camera

They are very nice and the fine resolution mode (image layering to 320 x 240 pixels) does help improve the images produced.

Measured temperatures can be effected by camera settings, with emissivity being the greatest influence on such. Be aware that skin is the exterior temperature of the body so it is generally cooler than the core. Checking the temperature at your temples or inside your mouth may provide a more accurate reading.

A thermal camera temperature measurement is normally specified as +-2C or 2%. It is not the most accurate temperature measuring device as there are so many influences on IR non contact temperature measurement.

Fraser

[HighVoltage]

Foll those that bought a U5855A with broken handle holder, like me, here is a good working fix.
I have used a 2mm stainless steel rod and mad it fit with one single bend.
The holes I drilled carefully with a 2mm drill bit and then glued the rod in place.
This solution works really well and is much stronger that the original plastic holder.

I just painted the rod black with a sharpie, otherwise it looked a little weird.

[ollihd]

Finally got my U5866A. Came with a working strap and a new battery. Compared the calibration to my just calibrated E4 and it seems to be spot on. The ability to manually focus for me is extremely important (a must have in the future). My question is: when taking a really close up photo, is it normal for the non-IR image to be misaligned / not the same with the IR-image? This seems to happen on very close up images. This is probably due to the different lenses? The same happens with the laser also (of course?).

As I didn't have a charger for the unit I went to my local flea market and got one for 3€. For the case I bought a Pelican -case replica for 35€ and it works like a charm.

[marshallh]

He has arrived. Pictures taken with both U5857A and F30S which are similar in many ways






"Fine resolution" does work sometimes, it makes the image a little better. Definitely not a replacement for a real 320x240 sensor. But it is a slight step up.

[marshallh]

[marshallh]

Easily comes apart. Mind the gasket



Unlike the other NEC Avio-designed camera I have (F30S), there is no sign of design/plug compromises to fit space.



Back of acquisition board. LTC2245 is a 10-bit, 10MSps ADC.



Very similar TMS320 used as on F30S. Has 128MB DDR2. They sure love the TI bus buffers. On F30S they use a X3S100 fpga. Not sure what he AVATAR bga is, probably a CPLD. NEC Avio don't seem to have a vendor preference for Altera or Xilinx. BQ24171 battery management.



Interesting that they used the Industrial temp grade version. 7 = Mid-range speed grade. Wired to 64MB DDR2. AVIO logo is visible on the board.
All instances of "PB-FREE" are crossed out.



Microbolometer is not shielded from dust, so rubber gasket on case is justified.



Interestingly, same package but much smaller window than F30S' sensor which is also 160x120. Take care to clean off dust if any appears.



Geared down focus pot. Shutter actuator is identical to F30S and sounds the same.

[TheSteve]

Great tear down - thank you for the excellent pics. When I opened mine I had a quick look for flash/eeprom but didn't see much. In your pics I still don't see too much that is obvious. I see the EPCx16N which is storage for the FPGA. The chip marked "AVATAR" doesn't look too standard for a flash part but I guess it is possible. More then anything I was wondering if they stored the camera unique information in the main flash or in a smaller eeprom. If we wanted to change the model # a small eeprom would be very useful.

[marshallh]

AVATAR may be an odd aspect ratio NOR flash. Usually they are never square (but they do exist). I agree that it is the only real place firmware could be hiding. The EPCS16 is good for 2 megabytes but the bitstream size for the FPGA is at least 1/3 of that.

I've confirmed that the F30S does actually run the exact same DM644 processor. It uses a BGA Spansion NOR flash with 64mbyte capacity. So I think you are right.

However, I've spotted at least 2 more chips of interest - the  Microchip (probalby 24FC series eeprom) and the chip whose I2C lines are shared next to it.





Here is the board inside the F30S - designed by the same team pretty obviously:

[TheSteve]

It looks like U22 is a Microchip 24FC128 and U25 is a TI PCA9306 bidirectional i2C bus voltage level translator.

Who feels like removing U22 and dumping its contents?