FLIR ONE PERSONAL THERMAL IMAGING DEVICE FOR YOUR iPHONE5 & 5s

[mikeselectricstuff]

Teardown part 1

[mikeselectricstuff]

..and surprisingly it runs Linux....

[frenky]

..and surprisingly it runs Linux....

I can already see this been hacked into Flir One Ultra Pro Plus Extra , can't wait for part 2.

[Fraser]

Thanks Mike, great teardown 

FLIR will likely be watching their newest 'baby' being dissected..... interesting times for them 

Aurora

[ixfd64]

..and surprisingly it runs Linux....

Are you going to try to find out how to root it?

[mikeselectricstuff]

..and surprisingly it runs Linux....

Are you going to try to find out how to root it?

Not intending to - tried a few logins but no joy.
Can't see you'd be able to do much useful - I suppose you could have it spit images out of the debug UART.
Or maybe get it to talk USB to a PC...?

I've now pretty much figured out the protocol to the sensor, which is what I'm mostly interested in. Why have an I-phone sized board running Linux when you can talk to a sugarcube sized sensor direct?

[imau]

So, it seems max resolution could not exceed 160x120.

Is that right?

[HP-ILnerd]

Just FYI on the battery.  Unless there is a setting I haven't noticed, it doesn't share power with the phone.
I've only charged it once since I have it, and it still has 75%charge.  My iPhone, on the other hand draws
down pretty far every day.

[mikeselectricstuff]

Just FYI on the battery.  Unless there is a setting I haven't noticed, it doesn't share power with the phone.
I've only charged it once since I have it, and it still has 75%charge.  My iPhone, on the other hand draws
down pretty far every day.

I recall some mention of charging the phone when it was launched, and there is a DC/DC converter that I think may be there for the purpose - not looked closely yet. Maybe it only does it if plugged into a 2A charger, or once the F1 battery has charged.
A bit of a PITA to have to remove it to charge the phone seperately.

[HP-ILnerd]

I recall some mention of charging the phone when it was launched, and there is a DC/DC converter that I think may be there for the purpose - not looked closely yet. Maybe it only does it if plugged into a 2A charger, or once the F1 battery has charged.
A bit of a PITA to have to remove it to charge the phone seperately.

It's a minor annoyance, but it would be nice to charge the whole stack at once.  There's certainly no "plugged in" indicator on the phone.

[HP-ILnerd]

Some of the "other" apps mentioned by avvidclif are:

A closeup app, that allows you fine tune the visible image adjustment for parallax on close focus images (this works fine, but there are limits).

A panoramic app.

A time-lapse app that I am itching to try.  Apart from not having your phone on you, that could be really useful.

A "paint" app that allows you to superimpose thermal on visible, rather than other way around like it normally works.

[amyk]

Great teardown. That little sensor package is so cute... I wonder how long it'll be before smartphones start coming with them integrated as an option, with them being so easy to interface.

FYI the Apple connector ID chip is not much more than a 1-wire EEPROM that holds information about what the cable is and probably things like a serial number; the fact that their custom chip also has "2501" on the die reminds me of the DS2501 infamously used in Dell's laptop AC adapters for a similar purpose... and comparing the BQ2024/6 datasheets with the DS2501/2 ones, they are very similar.

(I may be the first one on the Internet to notice this similarity. )

It might be possible to connect a Flir One to an Android device via USB OTG...

[mikeselectricstuff]

Great teardown. That little sensor package is so cute... I wonder how long it'll be before smartphones start coming with them integrated as an option, with them being so easy to interface.

FYI the Apple connector ID chip is not much more than a 1-wire EEPROM that holds information about what the cable is and probably things like a serial number; the fact that their custom chip also has "2501" on the die reminds me of the DS2501 infamously used in Dell's laptop AC adapters for a similar purpose... and comparing the BQ2024/6 datasheets with the DS2501/2 ones, they are very similar.
I think the Apple chip does 2-way authentication - if it was just a memory the Chinese would have long since cloned it.


(I may be the first one on the Internet to notice this similarity. )

It might be possible to connect a Flir One to an Android device via USB OTG...

[amyk]

I think the Apple chip does 2-way authentication - if it was just a memory the Chinese would have long since cloned it.

As a matter of fact, they did. The links in that thread provide some more info including a basic pinout and a (very) high-level overview of the communication. If it shows up as a USB (OTG) device to the iPhone, like a standard webcam, then there's a good chance of it being hackable to work on any Android device without needing to disassemble it and work with the sensor directly.

[mikeselectricstuff]

Teardoen Part Two

[RobertHolcombe]

That custom breakout board is really impressive 

[miguelvp]

That custom breakout board is really impressive 

He does sell them:
http://www.whitewing.co.uk/protoboard.html

[RobertHolcombe]

I was referring to the video around 2:40 onwards, using the PLCC socket as a jig

[Rasz]

I am amazed by your proficiency with that scope Mike. At the same time I dont think I could work with scopes limitations (UI, buffer size, screen estate).
My first thought is slurping everything raw (Saleae clone), decoding on the pc and throwing it at wireshark, custom python script, or even spreadsheet.
No need to trigger when you can simply filter, run statistics, and test on the fly.


As always loved the video, especially the 'boring parts' at the end

[Fraser]

Mike,

Once again I am very impressed by your superb skillset....thanks for the excellent insight into the LEPTON 

[mikeselectricstuff]

I am amazed by your proficiency with that scope Mike. At the same time I dont think I could work with scopes limitations (UI, buffer size, screen estate).
My first thought is slurping everything raw (Saleae clone), decoding on the pc and throwing it at wireshark, custom python script, or even spreadsheet.
No need to trigger when you can simply filter, run statistics, and test on the fly.

Yes, you could do it that way, but with a scope you get a lot of info very quickly in real time, with good visibility of relationships between signals. Certainly way better for the earlier stages of reverse-engineering. If a protocol turns out to be more complex, then a dump & analyse process is probably the way to go.
Probably the biggest limitation on this scope is that it won't save lister data in all segments to a single CSV file. 
I probably ought to learn Python at some point....

[mikeselectricstuff]

I've been keeping an eye on ebay sales, and F1's have been selling for over $500 as recently as the 28th.
There have been a few going cheaper in the last week, but the avarage is still well over $400
Considering that they are also supposed to be in US Apple stores now this seems odd - I wonder if Flir's delivery rate has been rather slow.

Anyone here passing a US Apple store in the next few days? - might be interesting to drop in ask about availability.
 
Still no news of EU availability.

[Fraser]

I made a direct approach to FLIR regarding the purchase of a LEPTON sample. They did not even acknowledge my request. Considering I was asking ion behalf of a company, this was surprising. It would appear that FLIR are only interested in selling their new baby in their own products for the moment.

[firewalker]

Mike, are those cheap video glasses relative to the Flir One? Predator vision? 

Alexander.

[amyk]

Great work again!

That module looks amazingly easy to interface. As there is GPIO and provision for a shutter, that's what I would expect some of the register bits to control. There might also be some means for it to say "I need calibration", either a bit or some sort of "estimated error/badness" value.

Bit 0 of register 2 appears to be related to the die temperature sensor - if it's set it tries to read it again, waiting until it clears before reading the temperature from register 8. It also checks it before attempting to write to registers 6/4, so I'd guess it's some sort of "busy" bit for the temperature sensor. During calibration it actually writes what looks like a temperature to reg 8, but the sequence before that, to reg 4, is also different.

I was almost expecting the Linux bootup messages to contain a "Hello Mike :-)"...