Yet another cheap thermal imager incoming.. Seek Thermal

[Fraser]

@miguelvp,

Many thanks for all the testing you have done, especially as you do not have the benefit of live updates. If SEEK now look at this thread they will be aware that we are aware of an issue with the shutter and maybe they will come up with a solution ?

In the mean time, the user community can continue tests and consider their own solutions.

You comment regarding the localised temperature build up within the lens housing is interesting and Mike should be able to se this with his E4 when he has time. It will be interesting to see whether his stripped down SEEK suffers from this problem. An hardware option may be to perforate the lens housing to allow better ventilation but that is as nasty as it sounds and not a great first step IMHO. The regular FFC shutter operation is very unusual and maybe SEEK have good reasons for this or maybe its is a result of the poor equalisation of the microbolometer pixels as you have suggested. It is certainly a poor FFC result compared to other cameras that I have worked on. The FFC shutter appears to be a negative influence on the equalisation at the moment....not great !

If SEEK made it possible to set the FFC event minimum interval it might enable some fine tuning for best trade off between pixel drift and shutter thermal contamination. Can the 12um  microbolometer really be so unstable that it needs such regular FFC events ? Helpful comment from SEEK would be a blessing here as we are trying to improve their product performance which is good for everyone, including their sales and support team. I will write to them and see what they say.

Its 04:12 in the UK so its back to bed for me 

Aurora

[Fraser]

Friendly email sent to SEEK and this thread referenced for their information. It will be interesting to hear their thoughts on this matter.

Aurora

[callipso]

Friendly email sent to SEEK and this thread referenced for their information. It will be interesting to hear their thoughts on this matter.

Aurora

Thanks, I was just gonna do that

[miguelvp]

Friendly email sent to SEEK and this thread referenced for their information. It will be interesting to hear their thoughts on this matter.

Aurora

Thanks, I was just gonna do that

might not hurt to do it more than once.

[callipso]

Then I shall.

[Fraser]

I don't think there is any harm in my including a copy of my message to SEEK here as a record of events for others to read so here it is:

From Fraser to SEEK....................

==============================

Dear Sir/Madam,

I have been an industrial thermal camera user for the past 27 years and presently own 29 such cameras in my collection. I collect and repair thermal camera technology so have component level repair experience with microbolometer based thermal imaging cores. My reference designs on which I cut my teeth were the venerable FLIR PM570 and PM695 cameras

I have ordered a SEEK camera from you via a friend in the USA and I am currently waiting for its arrival as part of your order fulfilment plan.

In the mean time I have been working with friends on the testing of your product.Please see the postings by myself and others on the EEVBlog forum going backwards from this point:

https://www.eevblog.com/forum/testgear/yet-another-cheap-thermal-imager-incoming/msg537479/#msg537479

My user ID on EEVBlog is "Aurora"

I spent my evening and much of the night working with "Miguelvp" and others to determine the cause of a Flat Field Correction issue that appears to be present in your SEEK camera design. As you will see the camera appears to be suffering from thermal contamination of the FFC shutter and this causes a temperature gradient in the FFC results table that is clearly visible to the user. This is unfortunate as it degrades performance and could impact upon the SEEK cameras reputation in the user community.

Due to the limits of testing to date, I have not been able to determine (using Miguelvp's tests) whether the issue only occurs with small temperature span settings, but it is pretty obvious from the images that the FFC on your camera has some thermal gradient issues.

It has been suggested that the thermal gradient on the FFC shutter flag may be induced by localised heating within the lens assembly and the frequent operation of the FFC shutter, that is in itself unusual in my experience. Such frequent FFC events have a negative impact on the user experience as even the more common >2 minute interval attracts criticism from the unenlightened consumer thermal camera user community. Many do not know the purpose of the FFC event and they just see it as a freezing of the image that caused annoyance.

I should be grateful for any comment that you are able to make on this matter and hope that your R&D team may have some comment on the frequent FFC events and shutter flag thermal contamination that is present in your camera.

I hope that you will see the positive attitude that members of the EEVBlog forum have towards you product and our desire to help you make it as good as it can be for our mutual benefit. When I receive my SEEK camera I will be happy to offer any UAT comment that may be helpful to you.

Kindest Regards

Fraser
(Surname and email address removed)

================================

Please bear in mind this was written at 04:00(ish) in the morning so no critiques on the writing style or grammer please 

Fraser

[mikeselectricstuff]

A quick warning to anyone playing with an open unit - the flex, where it meets the USB connector, is prone to a sharp bend which fractures the track , so I'd strongly suggest you add something to stiffen it here - a blob of hotmelt would probably do the trick.

[Fraser]

@Mike,

Thanks for the advice.... I hate it when flex cables fracture. A major power supply ribbon in one of my PM570's did that and it was a PITA to repair to an acceptable standard.

OEM's sometimes help by adding a thin piece of flexible plastic like a cable strain relief buy even that can sometimes fail to prevent a track fracture.

I hope you found the work that Miguelvp and others did yesterday/last night interesting. Any input from you gratefully received when you have the time. I am hopeful that SEEK will read this thread (if they are not already doing so) and see that they have a good community of technical minded people supporting their new 'baby' 

Aurora

[mikeselectricstuff]

Quick observations :

Mine shows a gradient at power-up, so it isn't thermal.

Removing the shutter  and pointing at a flat surface, there is no gradient, but noise increases pretty rapidly between cal times - presumably this is a warm-up thing.

The surface of the shutter, though nominally flat, has machining marks from the mould. Although in a photo these make it look like the surface is curved, this is not the case.

[callipso]

Seems like I was a bit briefer...

---

Hello,

a possible flaw in the design of the SEEK imager has been discovered (yet not
fully investigated as of now) by some members of the EEVblog forum. I'd
like to inform you about this discovery and to point you where the issue
is discussed. (link:
https://www.eevblog.com/forum/testgear/yet-another-cheap-thermal-imager-incoming/285/ )

The community of dedicated and highly skilled electronics engineering
proffessionals there would be eager to help you address the issue..

Regards,

Matyas Zrnovsky
PGP signed

---

Also is THAT the shutter? Isn't that supposed to be a precision, thermally conductive metal part with an absolute temperature sensor somewhere nearby?

[Fraser]

Callipso,

Your email is concise and to the point. Mine is filled with waffle 

Well done

Aurora

[Fraser]

@Mike,

Interesting comments, thank you.

At least if the shutter is the cause of the issue it is a part that may be redesigned quite easily as a sub assembly. It could even be a retro fit for owners willing to fit it.

Mike, could you look at the FFC shutter face using your E4(E8) and a close up lens ? It would be interesting to see how 'flat' it looks at thermal wavelengths. I am not impressed with the surface finish shown in your picture.

@Callipso,

FFC shutters can be made from many materials, often coated metal or very thin hard plastic with a matt surface. Gloss plastic surfaces can act like mirrors at thermal wavelengths (a bad situation for a FFC shutter). Some cameras monitor the temperature of their metal shutters using a thermistor or thermocouple bonded to the shutter but this can be a reliability issue due to cable flexing. Others use a temperature sensor adjacent to the shutter to establish the ambient temperature in which the shutter is operating.

Aurora

[Fraser]

If SEEK were willing to supply a spare shutter sub assembly we could play with the design and see what works best. I would start by carefully polishing the shutter flags surface with plastic polish and one flat and shiny I would matt it with an abrasive or appropriate paint layer. If that still does not solve the issue I would remove the shutter flag and graft a thin metal shutter in its place. The metal shutter would be pre-painted with a suitable thermal paint to provide good emissivity.

Oh if only I had my SEEK to experiment with !

Aurora

[mikeselectricstuff]

Need to go out now -  but I think it may be a timing issue. Video later!

[Fraser]

A timing issue  Now that is a new viewpoint to the discussion. Excellent. I look forward to seeing what you have found. Thanks Mike.

At least if it is a timing issue that is a software function that can be changed.

Aurora

[ricksastro]

Quick observations :

Mine shows a gradient at power-up, so it isn't thermal.

Removing the shutter  and pointing at a flat surface, there is no gradient, but noise increases pretty rapidly between cal times - presumably this is a warm-up thing.

The surface of the shutter, though nominally flat, has machining marks from the mould. Although in a photo these make it look like the surface is curved, this is not the case.

Interesting.   I suppose it could still be thermal, but like Aurora intimated, could be a reflection of the thermal profile of the sensor itself on the plastic shutter resulting in the same effect.

[ricksastro]

My note to Seek:

I've attached an image taken with my Seek thermal camera taken of a thermally flat surface.   It shows a full 7 degree thermal gradient across the image (using the hi/low feature).   This gradient develops over a very short period of time (about a minute to reach the 7 degrees, maxing out at about 9 degrees after several minutes).

In the eevblog community (user ricksastro), there have been many postulations as to the source, with possibilities ranging from thermal profile of the shutter itself to the heat reflection of the sensor on the shutter to many others.
https://www.eevblog.com/forum/testgear/yet-another-cheap-thermal-imager-incoming/9999

I would hope you would make fixing this a top priority, since a 7-9 degree swing in measured temperatures is certainly not what I thought I signed up for.

Also, your algorithm to determine color pallet differentiators is fairly ineffective when small differences are seen (like looking at a wall with studs behind).    If there is a heat source in the image, the contrast between wall and studs is seen easily and clearly.   When the heat source is out of the image, the contrast is decreased dramatically to the point that the studs are not easily seen.  This is completely counterintuitive and I assume is the result of trying to minimize the appearance of the aforementioned gradient or noise.

Please let me know how Seek plans to address these findings of the community.   Participating in the blog above would go a long way to secure a following in the community as well as save you many hours of engineering time tracking down these issues.   A fast and positive response will make this one of those iconic devices that started it all.    Ignoring the community and it will likely be relegated it to the archives of devices with great potential but no support.

Thanks much…you’ve made a good device, and if it weren’t just so close to a great device, I wouldn't have bothered writing this.

Best of luck!

[efahrenholz]

Hello everyone, I've been a lurker for awhile but I just had to register because this is a fascinating thread.

As mike pointed out, it could be a timing issue.

Perhaps the gradient is due to a rolling shutter effect...

[ricksastro]

Hello everyone, I've been a lurker for awhile but I just had to register because this is a fascinating thread.

As mike pointed out, it could be a timing issue.

Perhaps the gradient is due to a rolling shutter effect...

If it were a rolling shutter effect, it wouldn't be an ovoid blob like it is.   But it could be something related I suppose.

On mine, I turned on high/low and watched the progression.   When first on, there was a small gradient (2-4 deg).  It grew to 7 after a minute and peaked at about 9 after a few minutes.   That points to an accumulation, whether it be error accumulation (magnifying the effect, like stacking Flat fields to reduce noise) or thermal accumulation.

[Fraser]

No matter what the cause turns out to be, it sadly points to SEEK possibly releasing their product before it was fully debugged. Not an issue with the application but more of an issue if it is hardware related. UAT surely revealed this issue to SEEK prior to release ?

Of course we have to consider the possibility that the pre-prod units did not have this issue and the large shipment from China or another supplier is not identical to the pre-prod units.....how many times has that happened to a design ? ....... I know plenty of cases of such including one where a unit used a GBP20 microwave SMT capacitor and the manufacturer decided to use a cheaper capacitor in the first production run as it looked good enough..... it wasn't and the performance of the unit was ruined. A lot of expensive rework on the production units ensued. I was the one who discovered the problem as I never trust specs until I have personally tested a unit and confirmed them. The manufacturer was mortified !

This is why I am very interested to hear SEEK's response to our comments.

Aurora

[mikeselectricstuff]

By timing I meant I thought it might be sampling when shutter wasn't fully closed, but have now ruled that out.
Still investigating...

[mikeselectricstuff]

Meanwhile a bug report - if you leave it in the "please plug the camera in" state, it leaves the phone screen on and never times out, so ends up flattenning the battery

[efahrenholz]

@mike,

I indeed meant a literal rolling shutter, as in the flat field is capturing frames as the shutter is closing which means the averaged field is not even. But could you elaborate on how you ruled that out?

[Fraser]

I have been looking at the thumbnail images of the temperature gradient issue that several users have provided. I may be seeing what I want to see but from a distance I believe I can see the shape of the angled end of the FFC flag in the darker coloured area. It isn't a 'blob' or oval, the shape looks like a trapezoid.

Maybe I am seeing things that are not there but I welcome other peoples opinion on this.

I have attached the clearest image and added a line that I believe shows the edges of the anomaly and potentially the shape of the shutter flag ?

Note that I deliberately drew the line on the thermal image before looking closely at the flag shape so as to not program myself to draw a certain shape. After drawing on the thermal image I drew a red line around the flag on Mikes shutter picture. The angles on the two pictures do look similar to me. Mikes comment on a potential timing issue may still be a possibility ?

Aurora

[efahrenholz]

Yes it does appear that way. What if its a single frame on the tail end of the flatfield group of averaged frames, for instance, as the flag is moving out of the way too early before all the frames have been captured? I can't tell what the orientation should be, but you might be right about the ghost of the flag showing up in the image.