Mµ Thermal Imager - real or fake?

[Corporate666]

I still have a hard time believing any company contacted them and expressed interest in a large volume purchase based on what has been shown so far.

I'm sure there are plenty of Corporate Suits as clueless as the punters who've waseted their money on this.

I don't believe any major corporations contacted them about "large volume orders", and requested samples.  If such communication took place, it would have been in the other direction, with the Mu guys soliciting distributors, and being told "sure, send us a sample, and we'll take a look", that they then misrepresent as interest in large volume purchases from distributors.  We sell stuff to a few international distributors and I've never had one refuse a free sample - but I've also never had one express interest in volume purchase of a product that doesn't exist and has no established sales behind it.  The Mu guys seem to have misrepresented the nature of PayPal's "oversight" (as well as IGG's), and they misrepresented the initial images (which were from a traditional TIC, not their device), so I'd bet this is just more of the same.  If a major corporation solicited samples from them as a precursor to large volume orders, I'd eat my hat.  And I don't wear a hat - so I'd go and buy one, then eat it 

[thermalguy]

I'm the senior developer/owner of a small thermal imaging company in the Kansas City area. I started this company in 2011 after about 12 years experience in thermal imaging. Our first product was definitely no cake-walk but we were able to complete it in about 7 months.

I have been watching the Mu Optics project with great interest. I have read the updates and comments and can't help but shake my head at some of the technical rational. I am not sure what detector technology they are using so it is difficult to estimate what kind of problems they are having so I will not comment either way. Unfortunately if they do not deliver, it will be the crowdsourcing concept that will suffer the most.

I do have a question though...is there a real interest from a commercial standpoint for an imaging head unit such as what they are trying to do?
What if it were an 80x60 with 70mK of sensitivity?

[eliocor]

What if it were an 80x60 with 70mK of sensitivity?

If not too expensive I would be interested in. But it should not have the typical limitation of USA made thermal imagers: 9fps!
I'm Italian and I do not need to follow such limitations, but...

[mikeselectricstuff]

I do have a question though...is there a real interest from a commercial standpoint for an imaging head unit such as what they are trying to do?
What if it were an 80x60 with 70mK of sensitivity?

I have no doubt there is - as soon as it gets down to 'cool toy' pricing people will buy it for fun. It would also be a valuable tool for anyone in the electrical or electronics field for rapid faultfinding, as well as the whole home energy thing that the mu people are targetting.
And of course the whole CCTV /security market.
80x60 is probably the lowest resolution that would be attractive, as with anything less it gets vary hard to figure out what you are actually looking at.
It's one of those chicken-egg situations - small market due to high cost, high cost due to small market. And the existing players in the market wanting to keep their healthy margins.

[MFX]

I think there is a large market out there for a cheap thermal camera (and by cheap I mean well under $1000), from a personal perspective one place is large live events with huge temporary power systems, it's very unlikely that you'd get someone in with several thousands of pounds of thermal imager on such a gig (the industry is slow to adopt new ideas particularly if they are expensive and often people need to buy their own kit)  but if it was something cheap enough to just have in your pocket then many tech crew would buy them personally and carry them for spot checking things like :-

Overloaded breakers
Overloaded cables
Arcing loose connections

80x60 would probably be fine for the above situations. Something for thermal checking of PCB's would interest me as well so would need a fairly close focus ability.

Martin.

[thermalguy]

We design, manufacture and sell high-end UFP thermal imaging systems. That includes 320x240, 640x480 and very soon a 1024x768 option. I would love to design a lower cost unit but from an imaging head unit perspective to target manufacturing/process control.

I envision a 35mmx35mmx45mm unit with an accessory port on the back. Then the user can purchase whatever port module they need as they need it. Maybe Bluetooth, WiFi, and USB for starters.

The unit would then stream to either a smart phone or pad.

With an 80x60 it would be tough to hit $350US. But $650 is possible if the market can take 1000 plus units a year.

In this industry, you pay for pixels and sensitivity. On the Mu Optics site, I see the pixel spec but there is not a reference to sensitivity. In thermal imaging, that is the spec you push. Since there is not a sensitivity specification, it leads me to believe there isn't sufficient statistical data from uniformity correction/calibration processes in place. Which in turn means there isn't a production process. I may be going to far but these are things that catch my attention.

[thermalguy]

@eliocor...I would have to check what ITAR restricts at 80x60. I don't remember.

[MFX]

$650 is still pushing it a fair bit for widespread personal purchase but I'd probably go for it, if you could get some demo units to large event lighting/sound hire companies then you may find there's a previously untapped market out there.

Martin.

[thermalguy]

This sounds like a fun little project. I will take one of our 320x240 cores and window it to 80x60. Then I will hook it up
to a Bluetooth EDR module and stream it to a laptop and display it in real time.

This is our YouTube Channel: https://www.youtube.com/user/CSTImaging?feature=watch

Give me 3-4 weeks and we will post a video of it working. If we get positive feedback we will do a USB hack.

If the feedback is still good, perhaps we will make it a product.

[mikeselectricstuff]

This sounds like a fun little project. I will take one of our 320x240 cores and window it to 80x60.

..but wouldn't that still cost 320x240 money? Maybe the optics would be cheaper.
The framerate issue is also significant - fast update rate makes it easier to see what's happening and what you're pointing at as the brain is good at putting the pieces together - limiting to, say 9fps would make this harder.

Of course I'm sure many cams that are limited to 9 fps are only done so by a software configuration setting - what a shame if someone figured out to hack this 

[thermalguy]

Mike, this is to prove the concept and show proof that it will work just fine. This would be a minimal cost on our part to play with.
The hard part, core design, is already done. Whether it is a 80x60 or 1024x768 makes no difference. You window the 320x240 to 80x60 and reduce the bandwidth as you are only sending pixel data not video data.

If this actually makes it to production, 80x60 is what will be installed.

[MFX]

I assume he means just for a demo to show what it would look like, actual product sensor would be 80x60?

Martin/

[MFX]

The response on the indegoogo page indicates there is definite interest in a low cost product, the only question is whether people would be happy with half the resolution for twice the cost (assuming when making their decision they believe the Mu imager has a real chance of coming to market of course).

Martin.

[EEVblog]

I think there is a large market out there for a cheap thermal camera (and by cheap I mean well under $1000)

Studies have show that $300 is the magical "toy" price point that Mike was talking about.
Above that and it becomes an "investment" in a tool.

80x60 would probably be fine for the above situations. Something for thermal checking of PCB's would interest me as well so would need a fairly close focus ability.

I agree such a resolution would be adequate for many uses.

Bare in mind also that educational institutions can buy the Fluke i7 thermal imager for $999.
http://triosmartcal.com.au/content/44-i7-education-offer
Yes, not everyone can get that of course, but it's something to think about, and Flir would not be losing money on that.

[mikeselectricstuff]

Out of curiosity, at the low end what are the approximate cost percentages of the sensor, the optics and 'everything else'?
Bearing in mind the fixed costs of vacuum packaging (presumably?), moving shutter and optics, would 80x60 actually be all that much cheaper than 160x120? Or does the higher res also increase the cost of the required optics?

[MFX]

Studies have show that $300 is the magical "toy" price point that Mike was talking about.
Above that and it becomes an "investment" in a tool.

There is an often ignored "no mans land" market of freelancers who wouldn't see such a device as a "toy" but also don't have the same sort of budget of a full blown company yet do still have a respectable disposable income for such equipment and there's a lot of them. Also if you can get useful kit into their hands they will get around, it will get seen being used and there will be interest in it. It would still be a substantial personal investment though so their decision window of to buy or not to buy will be much tighter than that of a company. A basic case study, a while back I bought an ACT SLA battery analyser, not cheap but they are good, purely from me using it on gigs I know of four other people that also bought them and also one company, doesn't sound a lot but those four people are going to be using them on other gigs with other techs ...

Martin.

[thermalguy]

I am very familiar with the FLIR i7. 100mK at 9Hz and the $999AU price is educational use only and great for students. I like the FLIR products in general.

In this category all microbolometers are vacuum packaged so it is a level playing field between array sizes. It doesn't cost any more to package a 80x60 than it does a 160x120.

The FPA and associated lens system are the #1 and #2 drivers of cost. Display system is #3 and its a toss up between processing electronics and shutter assembly. The magic of any core system is the flexibility of configuration and video processing capabilities.

There is definitely a cost advantage in going with an 80x60 versus a 160x120. However, the volumes drive everything.

FLIR moves close to 65K 320x240 detectors via Autoliv. Detectors that don't meet the grade are swept up off the floor and configured in array formats that provide the best performance and then stomped on by ALOT of recursive filtering.

What I am proposing to play with is an 80x60 at 30Hz with 70mK sensitivity with bluetooth, wifi, and USB connectivity to a smart device. The faster the volumes move to 3-4K per year, the closer to $400 we can get.

For me, first thing is first, I need to prove the prototype then talk to some other contacts to get the volume up.

[EEVblog]

What I am proposing to play with is an 80x60 at 30Hz with 70mK sensitivity with bluetooth, wifi, and USB connectivity to a smart device. The faster the volumes move to 3-4K per year, the closer to $400 we can get.

Why not do a Kickstarter 
Seriously. A commercial player with a proven track record showing a real prototype should be a winner.

[MFX]

A very unscientific straw poll of a group with about 3000 members (although a LOT are inactive lurkers), results after 12 hours :-

+25 I'm a freelance individual , Yes but only if it was much lower. (around $400)
+5   I wouldn't buy one but would routinely hire one for a gig.
+3   I own a company, Yes but only if it was much lower. (around $400)
+2   I work with event power but wouldn't buy, Hire or use one.
+1   I am building one in my spare time  (not sure this is entirely serious :-)
+1   I own a company, Yes if it was around $600 (£400)
+1   I'm a freelance individual , Yes if it was around $600 (£400)

Martin.

[eliocor]

I forgot to mention a really important feature of the thermal camera (especially for us electronics developers):
it should focus to NEAR distances (at least 15 cm) because it surely will be used to check our assemblies.

[MFX]

I'd like a near and far focus option but not sure if that's asking too much.

Martin.

[tom66]

Don't forget to publish the USB communication specification so we can write Linux/RaspberryPi drivers for it. Thermal imaging adds interesting ideas for certain robotics projects.

[mikeselectricstuff]

I'd like a near and far focus option but not sure if that's asking too much.

Martin.

I think that can be handled at least in a crude way by a simple lens distance adjustment - limitations on geometry etc, at close distances are probably not a huge deal, but it will need to be clear enough to figure out how the hot bit relates to its surroundings.

What I am proposing to play with is an 80x60 at 30Hz with 70mK sensitivity with bluetooth, wifi, and USB connectivity to a smart device. The faster the volumes move to 3-4K per year, the closer to $400 we can get.

Presumably this would have to be limited to 9fps for export - I had a quick scan of the regs and didn;t see any obvious exemptions for low-res.
Incidentally I wonder what the legalities would be of hacking an exported  9FPS camera outside the US....

[EEVblog]

Incidentally I wonder what the legalities would be of hacking an exported  9FPS camera outside the US....

Who cares.
It's yours, you can do what you want with it subject to local law.

[tom66]

Just put a 24C02 eeprom on there which has a byte for the frame rate limit. Make a USB "command" (??) called WriteEEPROMByte(address, data). You have only enabled people to hack their devices which shouldn't be prohibited.