en-AU.vtt

WEBVTT
Kind: captions
Language: en

00:00:00.000 --> 00:00:00.399
English (AU) (Spoken) [Manually Transcribed Captions]
github.com/WizardTim/WizardTim-captions

00:00:00.399 --> 00:00:05.800
So today we're going to be taking a quick
look at this here PIR sensor manufactured

00:00:05.800 --> 00:00:11.170
by HPM, specifically a XL632 model.

00:00:11.170 --> 00:00:16.510
This PIR sensor failed several years ago so
unfortunately I no longer have the plastic

00:00:16.510 --> 00:00:21.840
case it goes in although I was able to find
this here picture on the internet although

00:00:21.840 --> 00:00:25.490
rather pixelated it shows what it used to
look like.

00:00:25.490 --> 00:00:31.170
In addition I also found the original manual
that it came with and it's filled with all

00:00:31.170 --> 00:00:36.469
sorts of fantastic stuff such as this here
usage artwork, as well as some quite useful

00:00:36.469 --> 00:00:42.200
information about how to use PIR sensors as
well as how to place them and what their sort

00:00:42.200 --> 00:00:46.830
of limitations are and what sort of
loads you can and can't connect to it.

00:00:46.830 --> 00:00:52.100
So here's all of the internal circuitry from
it that I kept and we'll have a look at how

00:00:52.100 --> 00:00:56.730
it failed and how it failed in a not so nice
way that it came a little too close to maybe

00:00:56.730 --> 00:00:58.370
causing a fire.

00:00:58.370 --> 00:01:04.150
So first of all we can see that the PCB is
split into two separate PCBs the top one has

00:01:04.150 --> 00:01:11.380
all of the low voltage sensing circuitry including
the actual PIR pyro-element and the other

00:01:11.380 --> 00:01:17.400
PCB has all of the power electronics including
the power supply and whatever is switching

00:01:17.400 --> 00:01:19.020
the load.

00:01:19.020 --> 00:01:24.409
And of course clear as day we can see this
here three pin cable between the two of them

00:01:24.409 --> 00:01:30.400
that certainly seen better days and was on
the verge of catching fire as well as this

00:01:30.400 --> 00:01:34.869
here power PCB that is heavily discolored.

00:01:34.869 --> 00:01:40.450
So taking a look at this here power PCB we
can see that there's all sorts of power devices

00:01:40.450 --> 00:01:49.160
that you see in this sort of thing we have
this here TO-220 device which is a BTB08 this

00:01:49.160 --> 00:01:55.400
is a triac which is what switches the load
it's a triac instead of a relay because these

00:01:55.400 --> 00:02:01.150
PIR sensors quite often turn on and off so
you want it to have a very long lifetime and

00:02:01.150 --> 00:02:07.979
a relay which is mechanical doesn't last as
many cycles as a triac.

00:02:07.979 --> 00:02:15.379
We also have a bridge rectifier here as well
as a rather large maybe two/three watt resistor

00:02:15.379 --> 00:02:21.940
next to it so we can assume that this device
uses a resistive dropper there is an inductor

00:02:21.940 --> 00:02:28.720
as well as a capacitor however these are just
across mains as well as in series with the

00:02:28.720 --> 00:02:35.090
load presumably to just come in compliance
with EMI and RF emission requirements.

00:02:35.090 --> 00:02:42.450
There's also a whole heap of transistors as
well as capacitors and resistors splattered

00:02:42.450 --> 00:02:48.720
across the PCB these are mostly just to drive
the triac by the looks of it although I couldn't

00:02:48.720 --> 00:02:54.030
really be bothered to fully reverse engineer
this thing that's probably what they do.

00:02:54.030 --> 00:02:59.250
So taking a closer look at the discolored
area of the PCB we can see it's around the

00:02:59.250 --> 00:03:05.660
dropper resistor and one of the leads has
completely delaminated from the single-sided

00:03:05.660 --> 00:03:12.230
PCB just completely lifting the pad and this
solder joint looks more like a cowpat than

00:03:12.230 --> 00:03:17.400
a solder joint so I think it's fair enough
to say that this is where the failure happened

00:03:17.400 --> 00:03:23.310
and it stopped working because it's certainly
been quite hot around here on this PCB and

00:03:23.310 --> 00:03:27.980
we can also see that the other leg of this
dropper resistor has also started to lift

00:03:27.980 --> 00:03:35.409
off the PCB as well as this here transistor
as well as all the way up the PCB over to

00:03:35.409 --> 00:03:41.519
here between these two pins where another
power resistor is I'm not quite sure what

00:03:41.519 --> 00:03:47.370
this power resistor does but it looks to be
a one watt resistor.

00:03:47.370 --> 00:03:52.650
By the way these two resistors still measure
their correct value as marked on their colour

00:03:52.650 --> 00:04:00.870
codes so we can assume that this high temperature
high dissipation is not a result of long-term

00:04:00.870 --> 00:04:08.840
use causing the resistance to drift but is
instead a should I call it a design flaw that's

00:04:08.840 --> 00:04:14.730
caused too much dissipation which this resistor
really hasn't been able to dissipate correctly

00:04:14.730 --> 00:04:18.560
without damaging the PCB and surrounding components.

00:04:18.560 --> 00:04:23.760
Now of course I'm not going to attempt to
repair on this sensor it's really just not

00:04:23.760 --> 00:04:30.479
worth it the PCB is quite badly damaged with
all of this heat exposure as well as the fact

00:04:30.479 --> 00:04:35.690
this happened several years ago this sensor
has already been replaced however we can still

00:04:35.690 --> 00:04:38.540
tear it down and learn something from it.

00:04:38.540 --> 00:04:43.289
So let's take a look at the other PCB this
here is the cable that connects between the

00:04:43.289 --> 00:04:50.350
two of them it's three conductors inside and
as you can see it's quite heavily burnt presumably

00:04:50.350 --> 00:04:55.479
this is just going to be ground, power and
the triac on off this thing doesn't do any

00:04:55.479 --> 00:04:57.789
sort of dimming.

00:04:57.789 --> 00:05:02.710
And here is of course the actual sensor PCB
there's actually quite a lot on here we've

00:05:02.710 --> 00:05:11.889
got three SOIC devices first of all we have
a 27L2C which is a low power CMOS dual operational

00:05:11.889 --> 00:05:19.960
amplifier we also have the jelly bean LM393
dual comparator and of course no design is

00:05:19.960 --> 00:05:26.230
complete without the ubiquitous 555 general
purpose timer and this PCB of course has a

00:05:26.230 --> 00:05:31.900
smattering of passives such as resistance
capacitors as well as some active transistors

00:05:31.900 --> 00:05:37.870
some diodes and well that's it there's no
fancy microcontrollers on here however that's

00:05:37.870 --> 00:05:42.889
a rather surprising number of components for
a PIR sensor these days, these days you'll

00:05:42.889 --> 00:05:48.480
just see everything implemented in a single
custom chip as well as significantly fewer

00:05:48.480 --> 00:05:49.480
passives.

00:05:49.480 --> 00:05:55.870
I'll also mention that these are custom BISS
chips meant for PIR sensors have also been

00:05:55.870 --> 00:06:03.690
reused for these fancy radar sensors that
sort of replace PIR sensors in certain applications

00:06:03.690 --> 00:06:08.509
these BISS chips certainly weren't intended
for this sort of application but some rather

00:06:08.509 --> 00:06:13.740
fancy magic and wizardry has made this possible
these things are rather impressive how they

00:06:13.740 --> 00:06:16.880
managed to get this all done for such a low
cost.

00:06:16.880 --> 00:06:21.669
And of course on the other side of this PCB
is not much special this the electrolytic

00:06:21.669 --> 00:06:28.169
capacitors for timing and power supply regulation
there's these two potentiometers to set the

00:06:28.169 --> 00:06:33.630
time delay as well as the sensitivity there's
one of these cadmium sulphide (CdS) photosensors

00:06:33.630 --> 00:06:38.889
for day night sensing but of course there's
also the pyroelectric sensor right here in

00:06:38.889 --> 00:06:40.370
this metal can.

00:06:40.370 --> 00:06:44.819
And that's actually why I'm making this video
today to take a closer look at these metal

00:06:44.819 --> 00:06:49.599
cans there is actually quite a lot of information
on the internet about these plenty of people

00:06:49.599 --> 00:06:55.039
will tell you how to use them, like Dave Jones
from the EEVblog has a great view about how

00:06:55.039 --> 00:07:01.539
to use them and how they work as well as bigclive
who quickly brushes on how they work inside

00:07:01.539 --> 00:07:07.410
however I haven't found any good pictures
or any good materials about exactly what's

00:07:07.410 --> 00:07:13.230
inside them and how it's all sort of constructed
inside so I want to take a closer look and

00:07:13.230 --> 00:07:16.040
actually open one of these up.

00:07:16.040 --> 00:07:21.139
First of all just cut off the metal can from
the PCB we can just rip it off with some side

00:07:21.139 --> 00:07:28.139
cutters don't need to be careful about this
PCB and here is the actual sensor it's marked

00:07:28.139 --> 00:07:35.250
KDS9 [5E1] and it has this nice little window
which is probably made of silicon to pass

00:07:35.250 --> 00:07:43.400
the long wave infrared now unfortunately these
markings didn't really return anything however

00:07:43.400 --> 00:07:49.509
after looking quite a while I found this data
sheet from Murata I suspect that this is the

00:07:49.509 --> 00:07:53.490
device just based on the wording that they
use when they talk about the manufacturing

00:07:53.490 --> 00:08:00.750
technique but we can see either way the sort
of stuff that should be inside there is down

00:08:00.750 --> 00:08:06.970
here a ‘pyroelectric element’ which has
two pieces and it's in this sort of configuration

00:08:06.970 --> 00:08:12.039
which is exactly what other people say is
inside these on the internet so taking an

00:08:12.039 --> 00:08:17.449
old pair of side cutters to the silicon window
we can just break that and get one of the

00:08:17.449 --> 00:08:23.860
blades in and start cutting the metal can
and just clear out the shrapnel and soon enough

00:08:23.860 --> 00:08:29.759
peel all of that back and here's the actual
internals and we can see that this is actually

00:08:29.759 --> 00:08:35.070
a ceramic hybrid module which is exactly what
that data sheet mentioned before about construction

00:08:35.070 --> 00:08:40.600
technology so we can see here there is this
thick film resistor here which is the same

00:08:40.600 --> 00:08:47.300
as in the datasheet schematic as well as the
pyroelectric element here and of course it

00:08:47.300 --> 00:08:53.690
has three pins and it's got a sort of matte
finish on it and it's rather interesting uh

00:08:53.690 --> 00:09:00.779
it's sort of like a PTC device in its appearance
I guess because well it is a piezoelectric

00:09:00.779 --> 00:09:06.860
device at the end of the day umm it's still
rather interesting because we'll it's quite

00:09:06.860 --> 00:09:14.180
thick umm… actually it's *exhale* I don't
really understand this though it's doesn't

00:09:14.180 --> 00:09:18.970
really make any sense it's really thick I
guess it must have really low thermal connectivity

00:09:18.970 --> 00:09:25.370
because of course there's very little energy
coming in from the infrared umm… actually

00:09:25.370 --> 00:09:31.220
looking at this this really doesn't make sense
umm… actually where's the transistor in

00:09:31.220 --> 00:09:34.780
this um that was meant to be a trans… have
I been lied to?

00:09:34.780 --> 00:09:37.500
Is what, what is going on here?

00:09:37.500 --> 00:09:41.970
What's with these two pads here there definitely
was something…

00:09:41.970 --> 00:09:42.970
Wait a minute!

00:09:42.970 --> 00:09:50.470
This looks suspiciously like a upside down
SOT-23 transistor uh let me get my spudger

00:09:50.470 --> 00:10:00.420
in here and brute force it open and sure enough
we can see that there's a lead frame inside.

00:10:00.420 --> 00:10:05.340
And if we look a little bit closer we can
also see that yeah that's a silicon die, yeah

00:10:05.340 --> 00:10:11.230
this is an upside down transistor that's uh
definitely not a pyroelectric element and

00:10:11.230 --> 00:10:17.320
these two solder joints are presumably where
the pyroelectric device once was I presumably

00:10:17.320 --> 00:10:22.910
broke it when I rammed that pair of side cutters
right through the window and it like in neanderthal

00:10:22.910 --> 00:10:29.829
so uh yeah taking a look at the video and
yeah whatever this white thing is here that's

00:10:29.829 --> 00:10:36.320
definitely broken and here again yeah that's
broken and I presumably uh emptied that out

00:10:36.320 --> 00:10:37.810
into the bin.

00:10:37.810 --> 00:10:42.959
But fear not because through the magic of
buying another one from Digi-Key that's probably

00:10:42.959 --> 00:10:48.500
about the same we have a second and third
chance for me to screw it up again.

00:10:48.500 --> 00:10:54.010
But before we cut open this one a quick intermission
instead I want to mention this this is the

00:10:54.010 --> 00:10:58.589
smartphone that's recording this video and
it's also the one that recorded the last video

00:10:58.589 --> 00:11:03.490
about the Simpson washing machine this is
actually my everyday smartphone that I bought

00:11:03.490 --> 00:11:11.060
last year and it's been pretty good for making
these videos is it takes 4K well uh Sony advertises

00:11:11.060 --> 00:11:17.760
4K but it's actually not DCI 4K it's some
weird sort of not really just barely 4K that

00:11:17.760 --> 00:11:24.380
uh some weird 21x9 aspect ratio so I just
crop it to 1080p but regardless of that it

00:11:24.380 --> 00:11:30.660
was still a pretty good upgrade from my old
smartphone from 2013 which was this here Nokia

00:11:30.660 --> 00:11:36.630
Lumia 1020 which has an absolutely fantastic
camera module I mean just look at the size

00:11:36.630 --> 00:11:42.910
of this sensor module it is enormous especially
for a smartphone it's 41 megapixel so it takes

00:11:42.910 --> 00:11:50.360
fantastic photos but because it's from 2013
the video quality is 1080p, 30 frames per

00:11:50.360 --> 00:11:57.210
second, just barely and uh it drops most of
those frames however this and the new Sony

00:11:57.210 --> 00:12:03.910
phone have a number of limitations for example
the aperture is fixed as well as there's certain

00:12:03.910 --> 00:12:09.230
software limitations like you can't really
properly adjust the white balance as well

00:12:09.230 --> 00:12:15.449
as a number of software bugs that Sony does
not want to fix for some reason, so I decided

00:12:15.449 --> 00:12:21.199
to go out and buy a little camera that I could
properly adjust the settings on and this is

00:12:21.199 --> 00:12:26.720
it here, well okay it's, it's not a small
camera it's actually a professional full-frame

00:12:26.720 --> 00:12:31.570
camera but trust me I've got a good reason
for this and of course this is just the body

00:12:31.570 --> 00:12:37.110
of the camera I also went out and bought a
lens for it, well okay I bought seven lenses

00:12:37.110 --> 00:12:43.450
for it however they take fantastic pictures
so it was worth it in my book now I've actually

00:12:43.450 --> 00:12:48.720
already shown you some of the pictures from
this camera the previous PCB shots are actually

00:12:48.720 --> 00:12:55.560
from this here very fancy 300 mm telephoto
lens which takes a fantastic PCB shots as

00:12:55.560 --> 00:13:02.740
you can see here on these PIR modules and
here's a very nice PCB shot of an old GeForce

00:13:02.740 --> 00:13:09.240
4 card that I took for something else, but
this here 90 mm macro lens is the one that

00:13:09.240 --> 00:13:14.670
took all of the pictures of the PIR sensor
that you're about to see this lens is actually

00:13:14.670 --> 00:13:21.880
capable of 1:1 macro photos so they're quite
large and very high quality it's far greater

00:13:21.880 --> 00:13:26.670
than anything I would have been able to do
on the smartphone however I decided that that

00:13:26.670 --> 00:13:31.180
macro lens part way through the video wasn't
really going to cut it for some of the other

00:13:31.180 --> 00:13:36.019
pictures you're going to see so one of the
fantastic things about having a full-frame

00:13:36.019 --> 00:13:41.850
interchangeable lens camera is that I can
jerry-rig all sorts of adapters together to

00:13:41.850 --> 00:13:48.840
make this which is a microscope objective
on a tube extension as well as an E-mount

00:13:48.840 --> 00:13:53.769
so this is what took all of the micro photos
at the end of the video that you'll see so

00:13:53.769 --> 00:13:58.440
I hope you enjoyed the improved quality of
these photos I certainly enjoyed them quite

00:13:58.440 --> 00:13:59.930
a lot.

00:13:59.930 --> 00:14:04.860
And another cool thing is that this channel
actually regained its monetization so YouTube's

00:14:04.860 --> 00:14:09.930
actually been giving me a cut of the ads you've
been seeing on these videos now and with that

00:14:09.930 --> 00:14:15.170
money I actually went out and bought this
SD card and this lens filter so thank you

00:14:15.170 --> 00:14:19.500
to those who have been watching the ads although
I don't recommend it because the ads are not

00:14:19.500 --> 00:14:23.990
worth your time ,the money has actually been
going somewhere other than YouTube's back

00:14:23.990 --> 00:14:30.220
pocket and presumably influencing my P-score,
so enough with all of that talk let's get

00:14:30.220 --> 00:14:39.449
this macro lens on now and take a look at
that PIR sensor module.

00:14:39.449 --> 00:14:46.769
So this here is the replacement PIR sensor
that I bought off Digi-Key it's made by Murata

00:14:46.769 --> 00:14:51.910
which is the same as the previous sensor however
it's a different model just because this is

00:14:51.910 --> 00:14:57.149
the one that I was able to buy um and well
it's pretty much exactly the same it's about

00:14:57.149 --> 00:15:03.920
the same size it's a same two element thing
so taking a rotary tool grinder to this one

00:15:03.920 --> 00:15:08.940
this time that way I don't have to break through
the window I can just grind off the top of

00:15:08.940 --> 00:15:16.019
the metal can and we can see that it's glued
in with this black glue and then cutting into

00:15:16.019 --> 00:15:21.490
that even more we can take all of that glue
out and we can actually see now that piece

00:15:21.490 --> 00:15:28.019
of uh whatever it is ceramic piezoelectric
material as well as a FR4 PCB by looks of

00:15:28.019 --> 00:15:33.769
it in this one rather than a ceramic hybrid,
so cutting the rest of that off now and we

00:15:33.769 --> 00:15:39.259
can see Oooo, now this, this is the picture
that I got the camera for, check this out

00:15:39.259 --> 00:15:44.100
this is actually not a single picture this
is actually many pictures stitched together

00:15:44.100 --> 00:15:50.000
in a composite called a focus stack that way
it's really sharp really nice but I can make

00:15:50.000 --> 00:15:55.170
this picture even better check this out *pop*
there we go now it's good.

00:15:55.170 --> 00:16:00.470
So if I bring back that application note from
before we can have a read of it we can see

00:16:00.470 --> 00:16:05.970
that the sensor uses a pyroelectric effect
of a pyroelectric ceramic which is a kind

00:16:05.970 --> 00:16:13.740
of piezoelectric ceramic which is what this
thingy here on the top is the way this works

00:16:13.740 --> 00:16:19.760
is that when a human walks past the thermal
infrared energy heats up these little cells

00:16:19.760 --> 00:16:26.380
here which causes a ‘spontaneous polarization
of the ceramic’ which you can see here in

00:16:26.380 --> 00:16:31.699
this figure pretty much that just means that
it creates a voltage differential between

00:16:31.699 --> 00:16:38.380
the two electrodes of the sensor pretty much
like you'd see in the piezoelectric effect

00:16:38.380 --> 00:16:45.100
of certain ceramic capacitors as well as piezoelectric
speakers and vibration sensors it's pretty

00:16:45.100 --> 00:16:51.970
much the same thing however it's based on
temperature so it's pyroelectric.

00:16:51.970 --> 00:16:57.660
And as you'd expect in order to be so sensitive
to something as minuscule as the thermal energy

00:16:57.660 --> 00:17:03.350
from a human walking past the thermal mass
of this piece of ceramic has to be extremely

00:17:03.350 --> 00:17:10.620
low so if we take a look here at this here
telecentric photo side-on we can see that

00:17:10.620 --> 00:17:18.110
the ceramic element on the top is extremely
thin it's actually 0.1 mm thick so it has

00:17:18.110 --> 00:17:26.130
a decently good response to any sort of thermal
energy but it's still not as good as the micro-balances

00:17:26.130 --> 00:17:32.090
in something like a microbolometer that you'll
see in like a thermal camera but that's so

00:17:32.090 --> 00:17:36.900
much more expensive something like this is
amazingly cheap and can be put into something

00:17:36.900 --> 00:17:44.770
like a consumer product we can also see here
that the element has two black rectangles

00:17:44.770 --> 00:17:50.600
and that's because there's actually two separate
pyroelectric elements here and on the next

00:17:50.600 --> 00:17:55.980
page of the application note we can see that
it starts to talk about why the optics are

00:17:55.980 --> 00:18:00.610
so necessary for this thing and that's because
the lens at the front of this thing actually

00:18:00.610 --> 00:18:07.230
focuses onto two separate points which is
these two separate pyroelectric elements and

00:18:07.230 --> 00:18:13.080
that allows it to sense movement much more
easily at a much lower cost it doesn't have

00:18:13.080 --> 00:18:19.409
to measure absolute temperature it just has
to recognize changes in the thermal energy

00:18:19.409 --> 00:18:25.960
when someone walks through the field of view
in specific ways and that's why before in

00:18:25.960 --> 00:18:34.490
the HPM product information we saw that they
took the time to explain to the customer how

00:18:34.490 --> 00:18:39.490
you should place it and in what ways it will
be triggered it won't be triggered when you

00:18:39.490 --> 00:18:46.190
walk away from the sensor but when you walk
across its field of view it will trigger and

00:18:46.190 --> 00:18:52.760
if we cut this PCB out of the PIR sensor we
can see on the other side that there's a transistor

00:18:52.760 --> 00:18:58.210
and two capacitors and that's exactly what
they said would be there in the product data

00:18:58.210 --> 00:19:03.760
sheet and also as you'd expect because this
thing is so thin if you just take a pair of

00:19:03.760 --> 00:19:08.650
tweezers and poke at it pretty much breaks
immediately.

00:19:08.650 --> 00:19:16.110
I'll also mention that there are actually
different types of pyroelectric infrared sensors

00:19:16.110 --> 00:19:23.419
that can detect more different types of movements
like this here quad type pyroelectric sensor

00:19:23.419 --> 00:19:28.270
that has four of these elements and that can
detect people moving across the field of view

00:19:28.270 --> 00:19:34.140
both sideways as well as up to down so there
are certain applications where you'd want

00:19:34.140 --> 00:19:38.539
to use this for instance on a ceiling.

00:19:38.539 --> 00:19:44.790
And the whilst we're talking about infrared
thermal sensors I should mention that a pyroelectric

00:19:44.790 --> 00:19:52.130
sensor is not a thermopile sensor a thermopile
sensor is what you'd see in say one of those

00:19:52.130 --> 00:19:59.090
non-contact infrared sensors and to be honest
this is more comparable to something like

00:19:59.090 --> 00:20:04.820
a microbolometer that I mentioned before that
you'd see in a thermal camera however this

00:20:04.820 --> 00:20:10.630
is working on a completely different sort
of theory it's similar but I have one of them

00:20:10.630 --> 00:20:16.120
so let's take it apart and see what's inside
we can see here that the external construction

00:20:16.120 --> 00:20:22.659
is very similar we have a metal can with a
window on top presumably it's probably silicon

00:20:22.659 --> 00:20:27.860
the sort of wavelengths are similar but it
probably doesn't have the 5 µm filter maybe

00:20:27.860 --> 00:20:30.460
a different type of filter.

00:20:30.460 --> 00:20:34.790
So taking a big cut out of the side of this
thing we can see that there's actually a silicon

00:20:34.790 --> 00:20:39.620
die inside here not just a ceramic element
there's also another thing off to the right

00:20:39.620 --> 00:20:45.530
here and the datasheet says that this is a
thermistor which is used for getting the ambient

00:20:45.530 --> 00:20:51.130
temperature of the die and that's used for
compensating it which we'll get into later

00:20:51.130 --> 00:20:55.770
and here we have a picture of a second one
I cut open a bit more carefully and this picture

00:20:55.770 --> 00:21:00.740
is actually taken on the microscope lens this
is again another composite although this one

00:21:00.740 --> 00:21:04.840
I did quite quickly so it's a bit screwed
up in some places but we can clearly make

00:21:04.840 --> 00:21:10.169
out this black square here in the middle of
this silicon die as well as all of the connections

00:21:10.169 --> 00:21:15.419
that go to it although only two of them are
actually connected to the outside world so

00:21:15.419 --> 00:21:21.220
let's take a closer look at that so here's
another composite microscope picture of that

00:21:21.220 --> 00:21:27.049
this time with the entire can cut off and
we can make out much more clearly that black

00:21:27.049 --> 00:21:32.310
area with all of the connections although
here I've actually damaged it a bit and that

00:21:32.310 --> 00:21:37.220
was just by blowing on it with some air to
clean it off and that's because if we look

00:21:37.220 --> 00:21:42.549
closely we can see that this black thing is
actually suspended on all of these connections

00:21:42.549 --> 00:21:47.630
this thing is actually floating in the middle
of it which as you can guess means the thermal

00:21:47.630 --> 00:21:53.950
mass of it is much much lower than the previous
piece of ceramic that was 0.1 mm thick I don't

00:21:53.950 --> 00:21:58.960
know how thin this is but I imagine it's a
lot thinner and taking an even closer look

00:21:58.960 --> 00:22:04.360
now we can see all these individual wires
going from the piece of silicon to this sensor

00:22:04.360 --> 00:22:10.110
region they're all individually separate you
might think that these are just for mechanical

00:22:10.110 --> 00:22:15.020
support to suspend it in the middle however
they're actually essential to how this thing

00:22:15.020 --> 00:22:18.470
works as a thermopile.

00:22:18.470 --> 00:22:23.970
To explain how this works I have this here
graphic from Shanghai Sunshine Technologies

00:22:23.970 --> 00:22:28.700
and we can see that this thing is just made
up of a whole bunch of thermoelectric elements

00:22:28.700 --> 00:22:35.350
just like a K-type thermocouple that has two
dissimilar metals this uses two dissimilar

00:22:35.350 --> 00:22:40.470
metal connections between the floating black
area in the middle that's heated up by the

00:22:40.470 --> 00:22:46.320
thermal radiation from whatever you're pointing
it at and the actual silicon surround which

00:22:46.320 --> 00:22:52.929
has a known temperature for cold junction
compensation which is derived from that thermistor

00:22:52.929 --> 00:22:57.789
on the metal body that we saw before and of
course the reason why there's more than just

00:22:57.789 --> 00:23:04.000
one junction is to get better sensitivity
and of course this black area here although

00:23:04.000 --> 00:23:08.760
it just looks black it's presumably some sort
of very specialized material that's very absorbent

00:23:08.760 --> 00:23:15.429
to IR wavelengths as well and that would increase
the sensitivity as well it also looks as if

00:23:15.429 --> 00:23:20.980
it has a sort of wavy texture to it it's not
completely flat I would assume that this has

00:23:20.980 --> 00:23:26.160
something to do with reducing reflections
inside the device as that may influence the

00:23:26.160 --> 00:23:32.690
readings if certain angles were to cause reflections
and increase the reading we can also see these

00:23:32.690 --> 00:23:39.750
etching windows in the diagram these are a
undesirable thing these are more just to facilitate

00:23:39.750 --> 00:23:47.220
the manufacture of the device these actually
result in the absorber area not being able

00:23:47.220 --> 00:23:52.270
to conduct heat through the substrate to the
hot junctions of the thermocouples so the

00:23:52.270 --> 00:23:57.470
shape and design of these is quite important
for sensitivity as well also just because

00:23:57.470 --> 00:24:01.990
we can't see them in the other one doesn't
mean they're not there the black absorber

00:24:01.990 --> 00:24:08.080
material on top of the substrate presumably
is filling those gaps in however in order

00:24:08.080 --> 00:24:12.580
to see that I'd probably need some sort of
like scanning electron microscope and yeah

00:24:12.580 --> 00:24:17.539
I don't think I'm going to buy one of them
anytime soon now just to show you how thin

00:24:17.539 --> 00:24:24.620
this thing is here I have a pair of tweezers
and just gently pushing on it we can see that

00:24:24.620 --> 00:24:30.880
it just completely disintegrates this thing
is extremely thin just blowing on it too hard

00:24:30.880 --> 00:24:35.630
will cause damage to it and taking the piece
of silicon out now and putting it on the end

00:24:35.630 --> 00:24:40.130
of some tweezers we can really see just how
small this thing is and turning it around

00:24:40.130 --> 00:24:45.549
we can actually see the back side of it has
been machined out or I guess etched I'm not

00:24:45.549 --> 00:24:51.080
really familiar with the exact principles
of how the sort of mems device would be manufactured

00:24:51.080 --> 00:24:55.880
but I can certainly appreciate just how cool
it is that we can manufacture stuff like this

00:24:55.880 --> 00:25:01.600
not only once off but on such a scale but
the average consumer can buy one of these

00:25:01.600 --> 00:25:06.750
for very little money especially when a couple
decades ago this sort of technology would

00:25:06.750 --> 00:25:13.840
be considered black magic as something like
this would be obscenely expensive to manufacture.

00:25:13.840 --> 00:25:17.880
And that's it for this video thank you for
watching hope you enjoyed it and I'll just

00:25:17.880 --> 00:25:20.160
leave you now with some pictures I took throughout
this project.     :)