Electrophysics Micronviewer 7290A heat imaging test and inherent display lag

[Fraser]

The Electrophysics Micronviewer 7290A is a vidicon tube based science camera that is capable of imaging from 0.4 to 1.9um. This makes the camera capable of work in the SWIR band where it is used to observe targets that have features revealed in SWIR or emit in that band. Silicon based cameras do not normally provide imaging beyond 1.1um so this Vidicon Tube based camera has an advantage over the common CCD and CMOS high sensitivity technologies. The Vidicon tube used in this camera is a specialist type from Hamamatsu with the identity N2606.

My Electrophysics 7290A camera is going to a new home with a fellow forum member and I wanted to test the camera to ensure that it is still in excellent condition. I discuss the 7290A further in this thread .....  https://www.eevblog.com/forum/thermal-imaging/swir-electrophysics-micronviewer-7290a-user-manual/

My tests are simple and not intended to go beyond showing that the target of the vidicon tube is not burnt and that typical vidicon tube type imaging is produced. At the request of other 7290 camera owners I include tests showing the Tube lag that is a feature of this particular type of vidicon tube. Such is unavoidable in these cameras and must be tolerated as part of their inherent characteristics.

The camera was set up looking at the simple scene of a soldering iron resting in its stand and the cameras lens focused on the matt surface of the soldering cartridge.

The soldering iron is a PACE Intelliheat TD-100 fast heating type with the element in the tip section for a fast response and efficiency of heating.  The camera is equipped with a Fujinon 25mm 1:1.4 TV lens that is designed to illuminate the 1"size Vidicon tube. These lenses are an excellent choice for the 7290 series of cameras.

The test comprised setting the camera up on a table in a dark room and selecting its Auto AGC mode to obtain the best exposure possible in the situation. The PACE soldering iron was then placed into the cameras field of view and stepped through a range of temperatures from 200C to 450C in 10C steps. The cameras ability to image heat was thus tested to check that the Vidicon tube was working as expected, and it was. See the attached pictures for the results. A high temperature Blackbody source could have been used but I kept it simple for the requirements of these function tests. The soldering iron also produced a nice, well defined thermal target in the scene.

To test the tube lag the camera was again set up viewing the soldering iron but the iron was not heated and remained a passive part of the scene being observed. A white piece of card was placed between the camera and scene and the time that the latent image remained on the cameras image output was observed. The white card was then removed to observe how quickly the soldering iron scene was restored to the cameras output. A hand was then used to observe the effects of the tube lag on a scene that contained movement. The results were recorded as a demonstration of the effect. The Vidicon Tube in the tested 7290A performed as expected and showed no signs of damage or failure.  As such the results of this very simple set of tests may be used to test the relative performance of other 7290 cameras for thermal energy response and tube image lag.

Fraser

[Fraser]

Pictures continued 310C onwards...... This is the range in which I expected to see some obvious thermal emissions from the soldering iron on the 7290A output

[Fraser]

400C onwards .... the 7290A can easily image some thermal energy.....

The final image in this set is of the test scene with the lights on and the soldering iron set to 455C. Its emitted thermal energy is clearly visible even with the lights on.

Fraser

[Fraser]

And now for some videos demonstrating the lag.

I have had to ZIP them in order to attach them here.

Fraser

[Fraser]

This video has a longer period with the scene obscured to allow the latent image to dissipate more.

Fraser

[Fraser]

Test repeated.....

[Fraser]

The 'hand lag test'!

Fraser

[Ultrapurple]

Thanks Fraser - absolutely fascinating.

Back in the day the first video cameras I played with were the vidicon-based Pye Lynx. Had a lot of fun back when camcorders were only just being invented (and very expensive).

One of the things I do remember from the Lynx was the image lag, although I dimly recall this depended very much on the intensity of the scene illumination. Brighter scenes had (much) less lag. But I definitely don't recall the lag being as significant as what you're seeing here. I shall contact some of my TV camera historian friends and see if they can offer any insight. It may of course just be that the Hamamatsu N2606 tube is (perhaps intentionally) laggy as a trade-off against something else.

The only other vidicons I have any experience with are those in the Argus 1 and I'll defer to Bill on those!

[Fraser]

Ultrapurple,

The SWIR tube has greater lag than a conventional visible light Vidicon tube due to the target material and the time it takes to dissipate the latent image. I was operating the Camera in low light to avoid any risk of burning the target with the fully open aperture. It might be possible to alter the lag through better lighting or adjustment of the biases but I would not wish to mess up the factory settings.

Regarding the thermal energy imaging, we must remember that a standard visible light lens is fitted so we will not see the full SWIR capability of the SWIR tube. Apparently older TV lenses transmit better further into the SWIR band but I have not done any tests. The Fujinon lens should be good to 1.6um though.

Fraser

[Fraser]

Regarding the target scene FOCUS setting ......

I forgot to say, I did not disturb the camera during the temperature test so the lens focus will have been incorrect as it was focussed in visible light and was imaging an SWIR emission. I should really have refocused the lens on the soldering iron tip whilst it was emitting heat. My mistake. The IRIS in the lens was wide open so depth of field was also at its minimum.

As readers can tell, this was a ‘quick and dirty’ test rather than a carefully scripted and configured laboratory analysis of the cameras capabilities  I should have done better. Maybe next time when I have more time to setup the test etc.

Fraser

[Fraser]

For anyone interested in what equipment I used to capture the images from the camera. I used a EZCap USB dongle and it’s included image capture software. I captured video as MPEG2 and used the provided file conversion software to translate the videos to MP4. File size was significantly reduced  An originally 14MB MPEG2 file was reduced to something under 2MB. Hence why I uploaded the videos to the forum rather than another platform.

A word of caution regarding the capture dongle I used ...... my version of the EZCap checks the video input before allowing the recording to be started. The fact that I was feeding it a monochrome signal caused it to declare that no video was present at the input so recording could not start. I suspect it looks fir the colour burst signal.

I tricked the EZCap into recording by applying a colour bar signal to its input from a CCTV tester that I have and initiating the first part of the recording function. This convinced the EZCap that a valid signal was present at its input and it presented the “confirm record’ dialogue box in readiness to start the recording. I changed the input to that from the 7290A camera and confirmed the start of recording. All then went well A bit of a faf but this is the challenge of some modern consumer electronics..... they do not expect people to use unusual or obsolete technology with them so it is not supported.

There are many USB video capture dongles that have been named “EZCap” and it is hard to know how they differ. Maybe a later version would accept a monochrome input? I have posted a picture of the EZCap unit that I used. It is a relatively early model but was said to be pretty stable so I sourced a used unit as it was no longer made. The software supplied certainly seemed stable on my HP 5870 i5 laptop that is running Win7 x64. This particular unit predates Win10 so driver compatibility and availability is not known. I find USB Video capture dongles to be a very mixed bag. Sure they are small and cheap, but I seem to always have issues with driver or software application stability plus the potential to cause weird errors on my operating system. They are the single most unstable devices I have used on my computers so I tend to only install them when needed and remove their software afterwards. Video capture is a significant task for a computer, especially older models, so I suppose I should expect some challenges but I found the internal PCI capture cards to be easier to integrate into a PC. Maybe they were just a better design or better drivers ? I have mainly moved to laptops now though so have to use a USB capture device. I did buy an Expresscard capture card to play with though.

Fraser

[Ultrapurple]

Many, many years ago I obtained a GTH Electronics 'ACE' Advanced Digital Convertor video Enhancer' that can do wonderful things with ropey video signals. Apart from the very obvious functions such as adjusting white balance, colour, contrast and so in, it's capable of moving the colour information horizontally and vertically to correct for videotape copying where the colour information can drop down a line or two with every generation. It's also a pretty amazing standards converter, does aspect ratio conversion, and more. Mine has a version of the software that can cope with extremely poor input signals that are incredibly noisy and have poor/missing syncs, and it'll output a rock-steady picture. (It can't do much about excessive noise). Unfortunately they recently went out of production but if you're ever likely to need to play with analogue video (PAL/NTSC/SECAM) then I recommend you keep your eyes open for a secondhand unit. There was even a version that would convert any video input to true 24 frames/sec so that monitors could be (movie) filmed without frame rate problems - and the comprehensive colour controls meant that any monitor could be made to match any colour temperature to suit the film.

Magic.

Here's a link to the manuals for about 20 versions of the converter / software.

I have absolutely no doubt that the ACE could take the output of the Electrophysics Micronviewer 7290A and deliver rock-steady analogue video to anything you like.

The photo is borrowed from the GTH Electronics website.

[Fraser]

A great pity that they shut down production but they helpfully provide guidance on possible alternatives to the ACE hardware..... an,Ely “Virtualdub” that they say is free and can do all the functions of the ACE in software rather than hardware. I shall have to look at Virtualdub and the other software they recommended.

http://www.gthelectronics.com/gthplans.htm

http://www.virtualdub.org/

http://www.infognition.com/VirtualDubFilters/

https://www.neatvideo.com/

I do have a very high quality video capture unit that I really should hook up and play with. It is the Matrox MX02 Mini and not as portable as the dongles but it apparently produced excellent captures. It takes Component, Y/C, composite and HDMI inputs

Fraser

[DaJMasta]

I'll get some similar tests up with the 7290 and a 7290A as a comparison, but I can speak a bit to analog capture on modern hardware - I've been capturing a number of sources for live streaming and recording and have tried a few options.

For hardware, something like what you've got seems to be the easiest to work with.  There are cheaper dongles marked "Easycap" or similar (the brands are fairly fluid, they're usually just rebadges of a handful of products, though I think ezcap is either a maker or a primary brand name) which interface fine but have lower image quality, but the one you mention by Ezcap (not the same) and UCEC and others seem to be better laid out and give somewhat better noise performance (not a huge difference, but better).  On the "high end" of these capture dongles is the elgato video capture units, which are quite a bit more expensive, but which offer at least as good capture as the Ezcap/UCEC units, maybe better, and everything's got its own quirks.

If possible, look for a capture device that supports UVC.  The USB video standard ensures that it's recognized by lots of software and that the software has some control over its resolution and other features, and the way it plays with software is the usual main concern.  The more rectangular Easycap dongles (and other brands) will work fine and support UVC, but have the lower video quality, but the only reason I can see for going beyond a UVC Ezcap/UCEC unit is that they don't seem to play nice when there are multiple on a single system.  That means if you want to record multiple sources, you can go back to the lower video quality units, or shell out for the elgato units - which are not UVC and while they work great, there will be some software (microsoft's camera app, discord video chat, others) that doesn't support them.

On the software side, if your input device is UVC, the default camera app works in windows but limits the settings you can access most of the time (I'm sure there are equivalents for other operating systems).  Manufacturer specific recording apps often work but lack features or are tailored entirely around recording so they have a set autoshutoff or extra buffering or other "features" that make them ill suited to live work.  My favorite program, and what's been most versatile so far, is OBS studio, a program designed for livestreaming.  It takes more setup to use than most (have to setup a scene, a source, arrange them on the canvas), but it also will take multiple (even varied) input sources, gives you access to the hardware settings they offer, and allows easy application for text or graphic overlays, scene transitions, and similar - meaning you can avoid the need for anything more than transcoding after you've captured.  It is worth mentioning that the elgato capture cards only seem to support the widescreen SD output resolutions for some reason, so especially in OBS the image can appear stretched on default settings, but the software also allows you to manually transform the captured image so you can get to 4:3 or 5:4 as needed for your source.  Virtualdub will generally capture well, but I find the setup process for streaming capture devices more tedious and have run into issues with non DirectShow capture.

There are also a number of PCI and PCIe based capture devices that are supposed to be what people doing this professionally have used (not sure if that statement is current), but I have no experience using them (small form factor PCs don't have slots to spare!)

[Fraser]

Thanks DajMasta 

I believe the Matrox MX series units were used in professional video capturing circles but they have some drawbacks......

1. The computer interface is limited to PCIe card or Expresscard so no native USB or FireWire option. Thunderbolt was available if an expensive interface was purchased.
2. They are relatively old.... circa 2010
3. Matrox stopped support after Windows 7 so forget using one on Win10. I think MAC new OS support also ended at the same time.

Matrox effectively dumped their MX series capture devices and there was outcry from the user base with some swearing to never buy another Matrox product. The units can be bought relatively cheaply on EBay but are often missing the power supply, host cable and host interface (PCIe or Expresscard). Without the bespoke host cable and interface they are paperweights. Complete kits are to be found but more expensive.

It is a pity that the MXO2 Mini support was discontinued as it is a nice unit. Likely more complex to use than a simple USB dongle device but the capture quality at 1080P was said to be excellent. Some clever technology resided inside those Matrox units.

Fraser

[_Wim_]

Thanks for posting this. It is very nice to know what kind of lag a slightly used camera has. I will repeat the same test this weekend also, as I do not exactly remember how good I got mine in the end, but if I remember correctly, it was quite similar.

For the computer interface I use a DFG/USB2pro DFG/USB2-lt from '"the imaging source" (https://www.theimagingsource.com/products/converters-grabbers/video-to-usb-2.0-converters/dfgusb2pro/).

I also own a very low cost Chinese EasyCap device, but quality was notably better the DFG device (which I did not expect). On other analog camera's I could not see any difference between the two USB grabber devices, but the output signal of my 7290 is a bit weak and not fully according to spec, so that probably explains the difference.

Edit: just discovered my device was not a DFG/USB2PRO but an older DFG/USB2-It

[_Wim_]

Regarding the thermal energy imaging, we must remember that a standard visible light lens is fitted so we will not see the full SWIR capability of the SWIR tube. Apparently older TV lenses transmit better further into the SWIR band but I have not done any tests. The Fujinon lens should be good to 1.6um though.

In this datasheet from Sofradir they claim spectral transmission up to 2.2µm for their electro-physics lenses, but only show a transmission plot until 1.2µm, so not sure what is still getting through this long.

[DaJMasta]

I ended up getting an older 25mm electrophysics TV lens for mine for the 1" format, but I tried a few 2/3" format C mount lenses with it first and despite focusing problems, they did seem to transmit SWIR alright.  I get the impression that normal optical glass generally transmits into this band, where the only thing you have to watch out for is the IR cut coatings or otherwise integrated filters/AR coatings that block transmission.

I didn't realize The Imaging Source made a capture device as well, but I've got a few of their USB3 cameras and they've got fairly straightforward drivers, good free software, and well specified dimensions/sensors/framerates/etc. - nice to use.  Doesn't get recognized by the Windows camera app, but it works with literally everything else I've tried.

[Fraser]

I just gad a look at the Imaging Source USB2PRO capture unit. Sadly at $180 it is beyond my budget for such a unit. I found a Linux TV page that may be of interest as it details the USB2PRO chipset

https://www.linuxtv.org/wiki/index.php/The_Imaging_Source_DFG-USB2pro

It uses a CX23102 all in one video to USB capture IC and a 24C02H eeprom. Maybe another manufacturer uses the same CX23102 IC in a more affordable unit ?

Update: The Linux TV page lists other video capture units that use the CX23102 

https://www.linuxtv.org/wiki/index.php/Conexant_CX2310x

I attach the pictures from the above referenced Linux TV page

Fraser

[Fraser]

Regarding lenses and transmission. I understand that the more modern lenses often present the issue of coatings that effect performance in the SWIR band. If maximum transmission bandwidth is required then maybe a simple lens made from a common ZnSe lens element, or multiple elements, could be used ? The transmission plot certainly covers SWIR.

Some ZnSe lenses for CO2 laser use are coated for 10.6um so they may, or may not work well in the SWIR band. There is some fun to be had experimenting with CCTV lenses and lens materials 

As has been said, common optical glass CCTV lenses without special coatings are really good enough for many tasks, but the transmission does drop off as you approach 2um. I bought several vintage CCTV lenses in order to find those that could illuminate the Vidicon target fully and also provide good transmission. I settled on the Fujinon 25mm 1:1.4 lens that has a good reputation in the visible light imaging world. I found that anything less than a 25mm lens was unlikely to fully illuminate the Vidicon target and avoid corner shading/vignetting.

Also note that the 7290 camera should produce a relatively flat image when looking at a plain surface. If shading towards the edges is visible, the camera needs shading adjustment to flatten the image.

Fraser

[_Wim_]

I just had a look at the Imaging Source USB2PRO capture unit. Sadly at $180 it is beyond my budget for such a unit. I found a Linux TV page that may be of interest as it details the USB2PRO chipset
 

I also would spend $180 for it, but I got it for free  .  Today I used the recording function for the first time, and it was no good (jittery images). I had to result to a screen capture program to record the screen instead. I must say, my USB2PRO DFG/USB2-lt device is quite old (I have it at least for 4-5 years, and was used industrially before), so windows 10 was for sure not foreseen

Edit: just discovered my device was not a DFG/USB2PRO but an older DFG/USB2-It. This was only supported until windows XP, it is a wonder it still works on windows 10

[_Wim_]

Today I have done some more tests with both the easier cap device and the .USB2PRO USB2-It
First video’s show a very big difference between the two:



Later I discovered that the signal on the easier cap is clipping when I tried to connect the camera to both capture devices simultaneously (so loading the camera output more resulting in a lower output signal).  If other have problems with these cards, a simple attenuator might improve image quality drastically (left: USB2PRO USB2-It right: easier cap)


Just for fun I did some side by side comparisons with the Flir E60 (always with the USB2PRO capture device)
=>   side by side with soldering iron heating up to 450°C


=>   side by side soldering iron heat-up with 7290 camera fitted with an IR filter (Schott RG715)



=>   side by side hand test with 7290 camera fitted with an IR filter (Schott RG715)


Test setup:
-   7290 Camera I repaired (see https://www.eevblog.com/forum/thermal-imaging/swir-electrophysics-micronviewer-7290a-user-manual/msg3265658/#msg3265658)
-   Electrophysics 25mm f/1.4 lens with fully open aperature
-   Led lighting from above
-   JBC - BT-2BWA
-   Did my best to adjust carefully, so this is the best result I can get for the 7290
-   Screen recording via ShareX screen capture (https://getsharex.com/) which immediately makes compact movies ideal for posting here (edit: but still sometimes too big  )

EDIT: uploaded to you-tube

[Ultrapurple]

Regarding lenses and transmission. I understand that the more modern lenses often present the issue of coatings that effect performance in the SWIR band.

'We' have the same issue with near-IR and UV photography, where some lenses are markedly better than others outside of the immediate visible range. Bjørn Rørslett has compiled probably the world's best investigation of (mostly Nikon F-mount) photographic lenses used for near-IR and UV; you'll find the page here - the long-ish introduction is well worth reading but you'll find links to the lens assessments near the bottom of the page, grouped into different focal lengths (fisheye, wide angle, normal, medium long, telephoto, zoom, special purpose) plus a few other categories. There is also a smaller survey of other brands and lens types.

It's well worth a look in my opinion if you're looking for a lens for just outside the visible band without breaking the bank. Bjørn Rørslett really knows their stuff (I'm not sure what personal pronoun they use these days; read around a bit if you want to know more).

I have found the reports reliable when I've compared them to my own lens collection. The 'gold standard' remains the UV-Nikkor 105mm (which contains no glass and is amazingly well-corrected for a huge wavelength range) but if you're only looking at near-IR there are plenty of affordable possibilities.

[DaJMasta]

I think you're going to find a lot of C mount lenses that are fine in NIR or SWIR just because of the nature of their consumer use - they get used in security systems where NIR performance for night vision monitoring is a common feature, so they will usually be lacking a built in IR cut filter, whereas it's more likely to be built into the lens (or into the camera body itself) on a DSLR or similar because of the image quality reductions in the visible band.  Of all things, you actually see some discussion of this in ghost hunting circles - there is a somewhat standard modification to DSLR camera bodies (or point and shoots) that involves removing the filter in front of the sensor to widen the range of spectral sensitivity for.......... seeing ghosts, presumably.


Anyway, I've been using some low light monochrome cameras for NIR and near UV (to 300nm or so where the sensor sensitivity is basically zero) video and have been able to use many standard C mount lenses from major manufacturers.  A lot of AR coatings don't seem to have a substantial impact on out of band transmittance, so they are at least usable in the normal responsivity of a silicon sensor.

[Ultrapurple]

In my limited experience, low light monochrome cameras don't have (or need) IR filters - in fact, having one would reduce performance because it would crop out a useful part of the spectrum. A silicon sensor is intrinsically sensitive down to nearly 1100nm, so chopping 1100-700nm and leaving only 700-400nm makes no sense.

Likewise, few lenses I'm aware of have IR filters per se. The AR coatings may however have variable performance at the extremes: they don't need to work outside the visible band, so why sweat if they go wacky?

The bigger problem with using lenses outside their design wavelength is that they are (and can only be) optically corrected over a certain range and outside that band the performance gets bad very quickly. This image from Wikipedia shows roughly the focus errors for different corrections (roughly, different numbers of correcting elements) and you can see that away from the design centre frequency - in these cases in the visible band - things go screwy pretty fast. And that's without considering the also-significant issues such as spherical aberration that will likely also get bad fast. It may be that an indifferent lens will work well enough for a relatively low resolution video sensor, but once you start looking at HD and above (eg photography) it really starts to matter.