Hi just a follow-up video to the teardown of this flirty G 165 visual thermometer or thermal-imaging camera and you can see it's still in pieces because it was a real bastard to get apart and I just want to cover a few things which have come to light since the previous video was shot. so let's take a look at them now. The first one, of course was that I basically completely destroyed this thing, taken it apart and there were several reasons that led me down that path. There were the two screws here of course and I sort of guess that they might have been screws on the front here.

So I went down that path to try and take off the front bezel and all that sort of stuff and that didn't work and then it looked like I could prise the front off and that didn't work and everything else. and one of the first things I did try was to actually pry out the LCD you can probably see a couple of marks in here from where I probably had a go at that and it didn't seem to budge at all. So it you know, a few design things led me down the path to basically just ripping the whole thing apart, but as it turns out, the four screw holes in there that the screws actually do go through the front. So this screen and problem and maybe the rubber is applied after the fact, but it looks like maybe it comes out through the plastic there.

So I just wanted to actually poke through those holes and see if that's a way to get this thing open. So let's see if I can poke here. We go poke through a hole here. Maybe poking through I can feel that I'm hitting something? Is it the screen I think it is seems to.

oh yeah. yep, yep. Oh geez. I'm putting a lot of force on that yeah look I could feel yet it's coming off.

it's coming off. the screen is coming off so it looks like it is glued on there. So probably it looks like the way to get that off if you want to play around and modify your one is to I'm heated up I probably would actually release the glue on there most likely and allow you to get a suction cup in there. Or maybe you know, a knife or something in there.

a little plastic pry bar to lift up that I'd probably don't think you're going to be able to do it without heating it up. although I haven't tested that. Obviously this one still has a screw in it. so several people were right in.

That cheese. It's not easy there. We go there, we go. It is there we go.

It is stuck down there. We go. it's out. its out.

so that is the way. Yep, Yep. Nice. Ok, so there you go.

There's the four screws, screw holes right there. So if you want to do this, that's the way to do it. And it was actually yeah, there we go. It's got an adhesive back on that side.

that should have been obvious, but yep, I missed it. Oops. Well, at least nobody else has to destroy there. AHA tg165 if you want to take it apart.

Yep, you can actually get that screen out. Well, you might be able to get a knife under there and pry it out, but I'd recommend. Probably trying to heat it up and loosen the glue on there. perhaps.

but just be careful you don't want to ruin your LCD or anything like that. And also when I was up playing around with that trying to prise that screen out to begin with, it did seem as though the rubber was like our overmolded over the top of that screen. That doesn't seem to be the case. although if maybe is a bit.

I mean it's got a lip I Think it does have a lip, so it is. It is sort of does over mould it a bit. so that's sort of what fooled me into thinking that you know that wasn't the solution to actually get that screen out. So yeah it is, but still at your pain in the ass, they should have made this battery much easier to replace.

That's just stupid. Now the other thing I missed of course is the shutter because I was expecting an external shutter on this thing over the inside of the lens or something like that. There we go. You just saw the shutter closed there.

It turns out this actually does have a micro shutter attachment built onto the top of this and is not. It's not really like a proper external shutter, but it's actually a purpose-designed clip-on attachment. and if I actually looked at the photo of the module in the datasheet, it would have been fairly obvious that it was a clip-on attachment. I Noticed that it did have something clipped on the top, but I just thought that was a natural part of the Lepton package here.

But as it turns out, no, this has an optional which the datasheet doesn't tell you about. it just says you know it supports an external shutter. It doesn't say that it's a micro shutter actually on this thing and that's what that coil was on the top that we saw and there's that. there's a little shutter inside there and I'll get a close-up on this in a minute, but occasionally it will actually close over and you must assert if you were watching you would have seen it before.

close over and I don't know. I'm probably not going to be able to make it do it because it does seem to be quite random. But anyway, even if I try and heat it up there I'm not sure it's going to do it out there. We go.

it just closed, it just blinked, and yet it does the compensation and then dit. So I can completely miss that. It does have a shutter because I was expecting like a huge external shutter like they've got on the floor one and other thermal imaging cameras. But this thing.

It's so cute. little tiny micro shutter. Now of course I wasn't expecting it to have a shutter because the datasheet actually says that it has automatic temperature compensation and other stuff. and really, because it's not actually measuring the data coming out of this thing.

I didn't think it would need it. So I'm quite surprised that it does actually have a shutter, but they've gone for this little arm presumably cheaper and simpler integrated clip-on shutter on the top and you can see that little thing over there will when it shutters close. Hopefully I'll get it. I'll be out.

like we might have to wait until it does it, but anyway, it'll blink and a shuttle will come across and we'll do some. It'll take about you know, half a second to do a thermal correction. There There we go it. Just there we go.

It's a beautiful so that's got an internal surface on it that has a nice even thermal temperature gradient on it, like a black body surface so to speak, so that it covers it over and it can do some compensation. So it has uniform temperature across all of the sensor. So what it's basically doing here is what's called a nook correction or non-uniformity correction or flat field correction because the sensor itself is factory calibrated at temperature to give annex well as flat a field as possible. So if you, if you're measuring a blackbody for example, or a surface that's all at the one nice uniform temperature, then you should get a nice uniform temperature mapping out of the device.

But due to the inherent nature of the physics of it and everything else, they individual pixels tend to drift around and do all sorts of weird and wonderful things. So you can end up with grainy images and blotchy blotchy images depending on the thermal gradients. So if you cycle the thing through thermal temperatures and things like that, it can affect the performance of these sensors. And so periodically, it's got an internal temperature sensor in there and it knows when it's starting to drift and there is actually a status output in the status stream.

The serial status stream of this sensor that actually tells the microcontroller that hey, I think you should do a nut correction or a flat field, a flat field correction and then the microcontroller can command the shutter to come in and I'm not sure if the shutter is actually like it is probably to set what must be two separate pins on that on the actual on the physical connector down in there. So I believe that the microcontroller then in structured instructs it under software control to do that shudder Compensations: So I don't think that the actual Lepton sensor is itself actually commanding that nuk compensation or flat field compensation there. I think that it's sending the command to the microcontroller and it's saying hey, I think you do this and then the micro switches on that solenoid which then I can't make it, do it again, makes it blink and re compensates the thing and then updates the internal calibration tables because once that window comes over the top there, then it knows it's got a nice flat surface, even temperature surface to actually calibrate against. And it updates the internal calibration tables.

and everything's hunky-dory so it determines that it has to do that once in a while. Now the electron sensor itself actually has what's called our scene based non uniformity correction or nook correction in it. Where if you're moving around in your image is moving like this, then it can do some software stuff to actually compensate on the fly. and it's not really, you know, so it doesn't really need that external shutter even field compensation to actually do that.

So it can pretty much do that in in software. But when you are measuring and looking at a static field that doesn't move, then like if you've got the thing mounted on a tripod for example, this thing does have a tripod mount. I'm not sure why because I thought that you know the major use case scenario for this thing was actually turn it on. You know take a scan around, take a measurement, boom, turn it off and that's it all.

Sort of handheld you. so not sure why this thing has a tripod mount and the proper thermal camera. The flurry for an D 8. They don't have a tripod mount, so it's real and they have a really, you know, a proper big huge external shutter.

Oops. Time outs 2 switched on. So I'm actually very surprised that it you know they went to the effort to have this shutter, especially when they're they're not doing it. they're not taking that temperature as I said isn't coming from the Lepton sensor.

The scale. There's no scale over here, so I'm surprised that they bothered to do that when this thing is just going to be used. you know, switch it on for a couple of minutes, take a few measurements to switch it off, so it's It's rather unusual I don't know, but they decided that they needed it. Maybe there.

Maybe there'll be a firmware update in the future, but in any case, if you wanted to maybe hack the firmware on the thing and you could maybe add some scale functions and read the data out of the Lepton sensor, that's certainly possible and with a shudder in that may not up there it goes. Just blinked may not be as good in performance as a proper external shutter, but you know, hey, it's probably good enough for the low-end market that they're aiming for here with this thing. and if we turn the power on here, you'll notice that a couple of seconds after prignon, it should blink. There we go, and that is.

It does the calibration flat field compensation and that's pretty consistent. If I turn it, turn it on. Boom. Here it goes.

There we go. That's pretty consistent, so it does at least one at a fixed time after power on. That could be the Lepton sensor actually instructing send in the command out, or the micro might just go. Well, let's just do one.

Cameras been turned on and everything's fine, but after that, it does seem pretty random. It's not like a time-based thing at all. So one thing I wanted to test is does it actually do the calibration more frequently if it's going through a fairly quick ambient thermal ramp? So you know I'm here in the lab and it's you know, 23 degrees or something like that and it's pretty done constant. And if I turn the aircon on it, you know it's constant within its cycles within half a degree.

Celsius or something like that. So maybe it's not drifting all that much and you wouldn't expect it to just an ambient temperature like this. So what I'm going to do is actually stick it in the thermal chamber. I've already got the thermal chamber heated up to sixty degrees, so I'm basically moving it from the lab environment into a warmer air environment.

And you know I don't obviously have a temperature sensor on this or inside this. I'm not sort of reading a temperature out of the lepton sensor itself. You can actually get the temperature data out of this. It does actually come out of the serial stream on the thing.

but I'm not going to think go to that amount of trouble here. I'm just going to put it in and see if it does actually do that compensation and calibration more frequently when it's ramping through a temperatures. And let's give that a go. So here's my little lab thermal Chamber which is seen in a video a long, long time ago.

It's a 89 degrees Celsius Inside, it's had time to warm up, so I'm going to I just opened the door, whack it inside and then actually watch it. It'll take time for the heat to penetrate the lepton sensor of course to actually build up thermal equilibrium inside and change that sort of thing. but I just want to see if then it actually you know it calibrates more often just by having the same measuring the same field image. It'll be like measuring and well outside here.

So I'll be able to wave my hands around and and but see if just ramping that temperature up makes a difference. There we go. 55 Yep, so let me there we go. and I can No I can't wave my hand.

it's too far in. Sorry, but we'll be able to see the shutter there. We'll be able to see it blink so it's just got a uniform field. Instead, there we go at blinked.

So let's see if it blinks again, you can see the image there change in. That's interesting. Oh, there we go. Yeah, it only took 17 seconds.

So I think it might. That's certainly more frequently than more frequent than while I was getting before. But we've only got a couple of data points there, so that doesn't help. Come on.

Yeah, there we go. Yes, 17 seconds again. Enter it. Ah, okay.

well, it's a bit more than 17 actually. My stopwatch is uh, I have to reset it. But come on, you can blink. You can do it.

Yep, there we go. 20 again. So it seems to be doing it fairly consistently reset. so it's like 25 seconds by the time I actually start it, reset my stopwatch and start it again.

another 10 seconds. I Think it might do it again. So that's interesting data that that temperature inside it would be ramping up inside the lepton sensor. It should do it right about now.

Come on, come on, don't make a fool out of me. Although, you know it's not going to be a linear ramp. There we go. That one took a bit longer I think, but it's certainly much more frequent than before.

I Think that's pretty much proven that that it's a certainly based on the ambient temperature is causing it to compensate more frequently because I can go for several minutes easily. when it was at ambient room temperature and it didn't operate the shutter, there we go. that one took longer again so it's giving progressive taking progressively longer and once it reaches thermal equilibrium in there like inside, the Lepton sensor is actually bought up to the same temperature as inside the thermal chamber, then well, yeah, I'd expect um, it to get back to just being the same update rate is what we got in the ambience. So I think we've proven there that putting it in a rampion ambient environment does well.

Pretty much exactly what you'd expect. You'd expect it to do more frequent compensation because it's got an internal temperature censoring that Lepton. it knows there could be another like little slot 23 temperature sensor on the PCB inside the mainboard. I Don't know whether or not that, but I Basically, the Lepton sensor itself is capable of knowing when it needs that compensation as I said, can send out a serial command telling the micro that it needs to do it and the micro will decide if it wants to command that.

So if you didn't want the thing to do that we could actually disable that. You can probably hack the software to disable. or of course you could do a hardware mod to disable that shutter if you found it. really annoying.

And I've taken it back out here. and I expect it to update as it cools back down. It should update probably just as frequently as what it did before there we go. That was our first one.

I'm going to time it. There you go. That took 46 seconds there to activate that shutter. There you go.

That second one took a minute and three seconds. so it's getting progressively longer. Alright, it's been going for three and a half minute now and still nothing. It's uh, yep.

so it's well and truly. ah, happy little camper sitting there thinking that I need to do any correction whatsoever. Not a problem. I'm happy with this ambient non changing ambient temperature.

Nothing's drifting, everything's just hunky-dory inside my array. Yeah, I'm not gonna wait any longer. This is ridiculous. So I thought we'd just take a look at see if we can actually see that shutter and unclip that top part of it.

So I'll take it out of its lovely little socket there and try and unclip it because you can see it just clips on the clips on the top there. So we should be able to do that. This is trickier, but eh, that's going to so happy. don't break it because I'm looking at the camera screen here when I'm doing this sort of stuff.

night. there we go. there's the entire assembly are very nice. very nice.

I Thought you know is that my Springs it all come out. but no, it's just making contact with those two pads there. So there's the proper. There's the actual Lepton sensor which is what you see in the datasheet and this thing, of course, um, has only showing its ugly head on this Tg165.

It's certainly not in the floor one. As I said, it uses a huge external shutter. This is still an external shutter, but it's you know, a microbe shutter designed to purpose-designed to clip on the front. so that is.

that's very nice. I Like that. well engineered. And the other thing to note too is that it's performance can't be that great Because for this thing to be effective and a truly good calibration shutter is that it needs to have a temperature sensor directly on the shutter which comes over so you know precisely what that value is.

And if you look inside a proper thermal camera they have like a big aluminium plate or something like that that comes over energy and I painted matte black and all that sort of stuff. and it's got a temperature sensor directly embedded on that plate so it's a uniform temperature. But this thing of course is just nothing. It's just probably a plastic I don't know what material is in there Anyway, it doesn't matter.

there are the only connections going over are the coil of course to actually actuate the shutter there. So it's got no temperature sensor. so you've got to rely on the internal temperature sensor inside the Lepton sensor. and yeah, well, you know it's not going to be as good as a proper external calibration shutter, so it might work fine for non-uniformity correction.

of course, because the plate will be at the one temperature, but it you know it probably is for absolute calibration, it's you know, to a measured ambient temperature, it's not going to be that great. Now on here. it looks like the lines for the actuation coil actually go down to the corresponding pins down in here, but they don't. I've actually checked I've measured it and also check the pin out of the device and know that pin and that pin are part of the Mipi interface.

So they're nothing to do with this thing. but obviously there is no pin out on the device. that actually because I've got the pin out for the thing from the datasheet. so I don't there is no pin which says you know shadow actuation.

There's no separate pins for that. So what I think's going on is it must be doing it through. The chip has our various GPIO just general purpose I/o that can be configured for anything so it must be driving that so the microcontroller could still do it. Microcontroller could still get the command because this as I said, this will send out a command when it thinks at neiiore status bit when it thinks it needs what, when it knows it needs to do a compensation and then because it's drifted.

It's got an internal temp sensor and it can actually detect noise and or do all sorts of probably some tricky software involved in that. but it knows that needs a compensation so it can send that out via the serial interface and then the micro can still read that and go Okay, yeah, I Agree, Let's do a shudder compensation. Let's activate the shutter. So then it commands one of the GPI to sweet one of the GPIO pins inside the unit which then drives the coil.

It must be it because it's not on the pin out. Now, if you are going to probe a flat flex connector like this, then don't go down and probe these pins down here like this because you probe can easily slip off and then short out two pins together. What you need to do is probe the back in there and that's actually recessed like that so it's impossible to short out your pin so you can probe round. Oh I Just saw some data there so it's got the plastic down separating them in there so you can really get in there and probe.

Along with that, you can always drag the thing along without actually shorting anything out. So there's little tip and I'm trying to buzz out the pin outs on this and trust me, it's a real pain in the ass. You got to take the Lepton out of the socket. You've got to hold one probe in there, use the right tongue angle, get on the other side and I it's it's horrible.

but I'm getting there. Well, that was a real pain. But anyway, I confirmed a few things here. Is that out the SCL and SDA the I Squared C command database is actually connected so we can probe that.

but I have probed these pins before and there doesn't seem to be anything on there. Anyway, they are connected through to the module. by the way, this is the pin out on the flat. flex over here and this is the Lepton Module over here and I can also confirm that they're using video over SPI They're not using the MIDI interface, definitely the SPI bus.

They've got Spi chip select connected, they've got Spi clock of course, and they've got Spi Mozi the output pin so that ie. output from the Lepton module into the microcontroller. but there is no my So pin so there's no input on that coming back into the module. Ok, so I've got a powered up and it works.

works just a treat and I can control some things here. So let's have a look at the I Squid Sea line 1 2, 3, 4. That's the fourth pin along that's supposed to be the SDA pin and you can see there's no activity there whatsoever. I'm triggering right in the middle of that and there's just nothing on the SDA line at all.

I Haven't waited long enough to see if it actually sends a command on that when the shudder changes, but it's got a read data back out of that to get that shutter information. I Thought Anyway, well to get the information that it requires a nook events. So yeah, there's nothing on there. and and the SDA and the I squared C clock as well.

That was just me changing pins there. It's white triggered and there's absolutely nothing on the clock either. It's just sitting high. So zip and if I actually turn the power off okay and I turn the power back on, let's hang on.

Oh turn the power on. Oh, there we go. We triggered something. Well, it went high when I actually pushed there we go.

If I just push the power button once it actually goes high, then let's have a look at that. There we go. So if I press a hold down that power button, it actually goes. The clock line on the I squared C goes high.

But what? I Want to do? Hang on now. I've got to have three hands here. so turn it on. Boom and ah, what have I got something? It went low there, but it's the clock line.

You know it's not like it's clock in I squared C data at all. So we're getting nothing on that I squared C bus zip. And that includes when the shutter goes actually because the shutter is going to go as we saw before, like two seconds after Buddha. So there's just there's just nothing there and that's the data out of the module there on the SPI.

So that's that's video over SPI and you can see see the packets there, start sending awaiting. Actually, we should be able to get the update right on that should we. So if we capture that, we're getting packet bursts of video here with a spacing of 38.5 milliseconds works out to about 26 are times per second week. We're getting that it looks like are we get in 26 frames per second from that Lepton sensor? I'm not entirely sure.

and if we have a look at the waveform here that we've captured one of those big packets those happen every 20s. those happen 26 times per second then we've got some higher frequency content here. So if we zoom in on that and what here we go, zoom in on that. You can see that we've got some SPI stuff happening in here and if we go along, scoot along.

we can see that, then it changes over at that point to double. You know, larger amounts of most likely the video data. And of course, this is an 80 by 60 sensor. and well, what do we get if we look at that high frequency content there? I've counted those packets in there and bingo, we get 60 of them.

So each one of those little bursts in there is one complete line of data. And of course, if we go into there will no doubt find there's 80 bits of information in there. So too easy. Yeah, it is confirmed.

It's not giving out any more information than they claimed 80 by 60. And of course, there's not going to be a single bit in there. There's going to be 80 pieces of information for each individual sensor dot in there. Of course, it's not just an on-off type thing.

there's going to be a value in there. so there's going to be 80 words of information within that. and then each one of those is repeated as we said, 60 times across. Here for each line of information, and all this stuff at the start is some other sort of data.

Actually, there's some really high frequency content in there. There's a lot of information within each one of those. Look at that. There's as an absolute ton of information.

So look, I don't have access to the there's a separate document apparently from Fleur that is the like, the information like the serial protocol interface and all that sort of stuff. and I don't have access to that. So yeah, I'm not going to speculate exactly what the data is, but as we saw, we definitely did get 60 packets of information there. so that is, and this one's actually shorter.

so there's you know. There's no surprise to find that that would be 60 lines of information in there. That's pretty certain. I Think So there you go.

That's just some more playing around with this thing. I Was very surprised that there's basically nothing on that I Squared C Box I Couldn't find anything when the shutter activates on there when it powers up. just nothing. but it is actually connected through.

so maybe they have, you know, got a future use for that thing I Don't know, but it's basically just spitting everything out from the Lepton sensor. It doesn't seem to be actually getting anything in, so the shutter in there is over I Believe it looks like it is actually controlled by the Lepton sensor itself rather than you know sending the command as I said, back to the microcontroller and then the microcontroller. deciding that's going to activate the shutter. Now it looks like it's done in there because it doesn't seem to be any information going back to the Lepton sensor at all, which is really interesting Anyway, aren't yet requires.

You know if you really wanted to go into, it requires a lot more information and having access to those serial protocol documents would be very nice as well. Anyway, that was just a follow up. Hasn't been a quick one. It's probably like 30 minutes or something anyway.

Hope you enjoyed it. Catch you next time. Bye you you.