Mµ Thermal Imager - real or fake?

[CanadianAvenger]

How can you not support JTAG?

You can not support it by not placing a JTAG header or pads on the board [and consequently pre-programming the chips before soldering them onto the board]  Beyond that, there is the "support" aspect, so while the board may be capable fo being programmed via JTAG, it can be reserved for factory use only, and not a supported method for customer use. [many consumer products are that way]

Having said that I agree with Hypernova, in that they are just throwing around keywords. Still not a single post of any substance.

[tom66]

I was presuming that there would be a JTAG header on board, but for various reasons, "we will likely only support USB with the firmware that ships on the camera" which would be fixed in "new firmwares that will enable the other update methods". Which doesn't make sense to me.

[Phroon]

How can you not support JTAG?

By not providing the files that could be flashed over JTAG? 

That's how read it with my 'The person writing this has no idea what he's actually talking about' glasses on.

[CanadianAvenger]

And yet another update posted today with no substance. Today was talking about using an external 12bit ADC, and how many samples per second they need for 30fps. Perhaps next week they'll do the math on how many bits per second they are sending over the wifi.

[firewalker]

Can you quote it?

Alexander.

[CanadianAvenger]

I didn't want to bore people with a repost, and assumd they could read it on IGG if they wanted. [or are they visible to backers only?]

Anyway, here it is:

Charles McGrath posted an announcement:
Hello All,

I saw some discussion in the comments about the thermal resolution of the camera. So I thought that I’d talk a little about that.

Our MCU does have an onboard Analog to Digital Converter(ADC) which does run at 1Msps(Million samples per second), but we won’t be using that due to a logic level discrepency with the microbolometer. Therefore we have a dedicated ADC on the board to convert the image data.

With an image resolution at 160×120, that gives us 19,200 pixels in each frame. At 30 fps, that gives us 576000 data points each second. For the best thermal resolution, we are sampling at 12bits. Giving us about 6.9Mbps of output.

Cheers!

[mikeselectricstuff]

I didn't want to bore people with a repost, and assumd they could read it on IGG if they wanted. [or are they visible to backers only?]

Anyway, here it is:

Charles McGrath posted an announcement:
Hello All,

I saw some discussion in the comments about the thermal resolution of the camera. So I thought that I’d talk a little about that.

Our MCU does have an onboard Analog to Digital Converter(ADC) which does run at 1Msps(Million samples per second), but we won’t be using that due to a logic level discrepency with the microbolometer. Therefore we have a dedicated ADC on the board to convert the image data.

With an image resolution at 160×120, that gives us 19,200 pixels in each frame. At 30 fps, that gives us 576000 data points each second. For the best thermal resolution, we are sampling at 12bits. Giving us about 6.9Mbps of output.

Cheers!

That's actually the most plausible thing I've seen from them to date, apart from the 'logic level discrepency', which could just be him not quite understanding the words the tech guy is saying.
Quite plausible you'd need 12 bits for full range from an imager.

[Corporate666]

I didn't want to bore people with a repost, and assumd they could read it on IGG if they wanted. [or are they visible to backers only?]

Anyway, here it is:

Charles McGrath posted an announcement:
Hello All,

I saw some discussion in the comments about the thermal resolution of the camera. So I thought that I’d talk a little about that.

Our MCU does have an onboard Analog to Digital Converter(ADC) which does run at 1Msps(Million samples per second), but we won’t be using that due to a logic level discrepency with the microbolometer. Therefore we have a dedicated ADC on the board to convert the image data.

With an image resolution at 160×120, that gives us 19,200 pixels in each frame. At 30 fps, that gives us 576000 data points each second. For the best thermal resolution, we are sampling at 12bits. Giving us about 6.9Mbps of output.

Cheers!

That's actually the most plausible thing I've seen from them to date, apart from the 'logic level discrepency', which could just be him not quite understanding the words the tech guy is saying.
Quite plausible you'd need 12 bits for full range from an imager.

Don't the SAM3X chips have a 12-bit ADC with 1M samples/sec built in?

http://www.atmel.com/devices/SAM3X4C.aspx?tab=parameters

I can't comprehend what "logic level discrepancy" means in terms of sampling with an ADC?!

[CanadianAvenger]

I'm giving him the benefit of the doubt in that I assume he meant signal level and not logic level. (He's a VFX guy and not an electrical engineer afterall) 

While the post is factually plausible, it still provides no proof that they are actually doing anything... no pictures or anything else substantive.  Everything he said there is something that one would have to figure out on paper BEFORE they start the design. [I personally would have gone with some analog conditioning and then using the internal ADC, instead of an external one, as it allows for more flexibility in tuning the input signal to get the best possible result - but it is a matter of choice, there's always more than one way to skin a cat - there may also be other extenuating details that demand an external ADC]

[mikeselectricstuff]

I didn't want to bore people with a repost, and assumd they could read it on IGG if they wanted. [or are they visible to backers only?]

Anyway, here it is:

Charles McGrath posted an announcement:
Hello All,

I saw some discussion in the comments about the thermal resolution of the camera. So I thought that I’d talk a little about that.

Our MCU does have an onboard Analog to Digital Converter(ADC) which does run at 1Msps(Million samples per second), but we won’t be using that due to a logic level discrepency with the microbolometer. Therefore we have a dedicated ADC on the board to convert the image data.

With an image resolution at 160×120, that gives us 19,200 pixels in each frame. At 30 fps, that gives us 576000 data points each second. For the best thermal resolution, we are sampling at 12bits. Giving us about 6.9Mbps of output.

Cheers!

That's actually the most plausible thing I've seen from them to date, apart from the 'logic level discrepency', which could just be him not quite understanding the words the tech guy is saying.
Quite plausible you'd need 12 bits for full range from an imager.

Don't the SAM3X chips have a 12-bit ADC with 1M samples/sec built in?

http://www.atmel.com/devices/SAM3X4C.aspx?tab=parameters

I can't comprehend what "logic level discrepancy" means in terms of sampling with an ADC?!

I can't think of anything you couldn't fix with some external circuitry - could be that the sensor output is 5V or bipolar, or needs buffering. But an external ADC could well have better performance.

[EEVblog]

I didn't want to bore people with a repost, and assumd they could read it on IGG if they wanted. [or are they visible to backers only?]
Anyway, here it is:

Hmm, does it even matter what they say technically any more?
The only thing that matters to the backers is seeing a prototype and/or real thermal image form the prototype.
They can shoot out a few sentences of technical detail every week until the cows come home... 

[CanadianAvenger]

I agree whole heartedly. The only thing that matters at this point is tangible proof of a working camera prototype, or notification of delivery. Any other discussion is just a waste of words. I would expect that a mass exodus is going to happen soon unless they produce something of substance.

[EEVblog]

I agree whole heartedly. The only thing that matters at this point is tangible proof of a working camera prototype, or notification of delivery. Any other discussion is just a waste of words. I would expect that a mass exodus is going to happen soon unless they produce something of substance.

If I was them would have been in serious PR panic mode a long time ago, and squashed the nay-sayers on the head. To let it have have gotten this far without showing anything of substance is just  project suicide. It's trivial given how easy it is to snap a photo of the prototype case(s), prototype board(s), test image or whatever.

[CanadianAvenger]

If I was them would have been in serious PR panic mode a long time ago, and squashed the nay-sayers on the head. To let it have have gotten this far without showing anything of substance is just  project suicide. It's trivial given how easy it is to snap a photo of the prototype case(s), prototype board(s), test image or whatever.

Absolutely! They might still have my money if they had.

[Jorge12]

P.S. Unlike what I would give to MuOptiBullSh*t  I heap tons of applause onto IR Blue for making an awesome, legit product. These guys really deserve it, so here's a link: http://www.kickstarter.com/projects/andyrawson/ir-blue-thermal-imaging-smartphone-accessory

Yes, your reconstruction of the Mu history is impeccable and the technical details you're sharing with us are full of meaning.
I also think that the IR Blue is an interesting project. As a confirmation of the validity of this kind of devices, here is another similar project that curiously uses the audio channels for data communication:

http://play.google.com/store/apps/details?id=com.optivelox.thermovisio

[Marco]

Can't a fresnel lens and a cold mirror work for the optics?

[Marco]

The transmittance of acrylic in deep IR is poor ... but fresnel lenses are thin. I did do a bit of googling and math in the past and transmittance of >50% should be possible ... might have made an error, but some companies market these lenses for low cost thermal cameras.

[CanadianAvenger]

P.S. Unlike what I would give to MuOptiBullSh*t  I heap tons of applause onto IR Blue for making an awesome, legit product. These guys really deserve it, so here's a link: http://www.kickstarter.com/projects/andyrawson/ir-blue-thermal-imaging-smartphone-accessory
Unfortunately I came too late to get one, but there'll probably be Eagle files etc. available soon to make one

I backed the IR-Blue as well, and received it the other day. The app is pretty good, and it works, but I would only classify it as a toy, rather than a tool. [Not to suggest that it's useless, far from it. It can be used to look for hot or cold spots, but the performance is pretty low, so scanning a large area would take some time.]

[mikeselectricstuff]

Can't a fresnel lens and a cold mirror work for the optics?

No. All I'm going to say is that plastic and glass completely block the passage of thermal wavelengths.

Polythene can be used - it's used for most PIR sensors, however purely on the basis that nobody's done it, I'd guess it isn't clear enough to be practical.  most, if not all  the low-res thermopile arrays available use (AFAIK) silicon rather than polythene. There is a company linked somewhere in this thread that claims to have some plastic fresnel lenses suitable for thermal imaging.
 

A germanium lens that size would cost around $450 (edmundoptics.com/optics/optical-lenses/ir-lenses/germanium-meniscus-lenses/3081).

If it costs that much from a specialist scientific distributor like Edmund, a high volume price between $50 and 75 is easily plausible. I know a company that uses that smaller lenses for a 16x16 sensor (approx 10mm dia) and pays something between $10 and $20 for them in 10K volumes. You probably do need something larger, or multiple elements for 160x120 to be useable though.

[fpga]

OK, y'all. As far as I can tell, this is how the whole thing has played out.

John McGrath, the lead guy, is damn rich. On a whim, he decided to buy a $6,000 FLIR E50 or E60 (other possibilities there) or just acquired one somehow. (flir. com/cs/emea/en/view/?id=41372) He found it pretty neat, and had the childish thought that he would make his own and mass produce it for less. (Kind of like what the guys in October Sky were thinking, but far more moronic and petty.)

This sounds very much like the better half of the projects / startups I've been involved with. They all have this very fancy presentation, pictures, 3D model, video, website, etc. During the interview you cannot help but think that if they have been able to accomplish all this, then what the heck do they need me for. Once I sign the NDA, they hand me a product or manual from a big established "competitor" and tell me that they want me to create a better and cheaper product than anything else out on the market. Sometimes they combine the manuals from several existing competitive products. When I ask about all the work that they had already done, they would say it was only for investors or marketing purposes only. As I dig in, I soon also realize that they had already sold X number of units and promised to ship in 3 months. So while the established "competition" spent millions in R&D and many years to develop their products, the startup founders expect to do better in 3 months with a single consultant. As for themselves knowing how the product is supposed to work or the requirements, they don't have a clue. They expect the hired consultant to figure everything out.

On a number of the projects, I've been able to succeed. But it took far longer than 3 months of dedicating my life to the project and with constant haggling over why the project is taking longer than 3 months, haggling over getting paid for the 80 hour weeks, constant feature creep because the competition added a new "must have" feature, finding excuses that they could tell their customers, fighting with vendors and distributors over parts cost and availability, and all of the associated stress.

For these guys, it is possible to find a brilliant consultant that could create the product for them. So there's hope, but they will also have to make sure they don't run out of money before they find the consultant or at least still be able to convince him/her that there will be worthy compensation for the effort.

I was just recently asked about exact same type of project idea, but for medical ultrasound imaging. The brilliant idea they have is that most of the cost of medical imaging is in the display, so if they replace the graphical display with an iPad, they could produce it for a fraction of the cost. Just design an ultrasound probe that we could plug into an iPhone or iPad -- it should be easy with an Arduino. What's next? CT Scan or MRI anyone?

[Bored@Work]

I was just recently asked about exact same type of project idea, but for medical ultrasound imaging. The brilliant idea they have is that most of the cost of medical imaging is in the display, so if they replace the graphical display with an iPad, they could produce it for a fraction of the cost. Just design an ultrasound probe that we could plug into an iPhone or iPad -- it should be easy with an Arduino. What's next? CT Scan or MRI anyone?

Were they knowledgeable about the requirements for developing medical devices? The mandatory processes and mandatory documentation I have seen for developing medical equipment is not for the faint of heart. If I got it right you can't even do some stuff retroactively. I.e. you can't just happily develop something in whatever way you like and in the end get it somehow certified. You need to follow the mandatory stuff right from the beginning of the development, otherwise you don't have a chance to get it certified.

[fpga]

Were they knowledgeable about the requirements for developing medical devices? The mandatory processes and mandatory documentation I have seen for developing medical equipment is not for the faint of heart. If I got it right you can't even do some stuff retroactively. I.e. you can't just happily develop something in whatever way you like and in the end get it somehow certified. You need to follow the mandatory stuff right from the beginning of the development, otherwise you don't have a chance to get it certified.

You're absolutely right! I worked for two real medical instrumentation companies in the past and it is hard not to get pulled into all of the procedural stuff like procedure training, training certification, reviews, document trails, dealing only with certified vendors, legal documents, etc. You can't touch anything before getting documented approval for which you need to document why you need to do what you need to do. Basically it all is for good intentions, but feels like you are dragging a boat anchor and everything takes many times longer than if you were to just do it. A medical product, even if it is non life support and non invasive, diagnostic use only, having a fully functional design is just the beginning in terms of development cost.

These guys were clueless. In fact the management at the real medical companies were clueless too, but at least they had whole departments dealing with the medical instrument certification, FDA, and QA stuff.

[geoffhamilton]

New update:

http://media.tumblr.com/42c707e2aaabf6d6ee9c09beff9f9b2c/tumblr_inline_mpmhfxKlgL1qz4rgp.gif

[MacAttak]

That would have more comedic impact if you posted it to the project comments along with accompanying EE word soup, to mimic the real updates that get posted there.

[Corporate666]

I had seen that the project creator (or one of them) yesterday posted in the comments, saying that an update will be coming today.

Then I noticed that the comment had been deleted. 

Strange.