en-AU.vtt

WEBVTT
Kind: captions
Language: en

00:00:00.001 --> 00:00:01.040
English (AU) (Spoken) [Manually Transcribed Captions]
github.com/WizardTim/WizardTim-captions

00:00:01.040 --> 00:00:06.799
This is a PMS7002
laser dust sensor from PLANTOWER

00:00:06.799 --> 00:00:09.440
I've had it for about
two years now and it's been running

00:00:09.440 --> 00:00:13.519
continuously and I've collected a fair
bit of data as you can see here

00:00:13.519 --> 00:00:17.039
in particular you can really see the
fires we had recently

00:00:17.039 --> 00:00:21.199
at the start of 2020 although on average
over the past two years

00:00:21.199 --> 00:00:26.320
it's been exposed to an average of 6.4
micrograms per cubic meter of dust at

00:00:26.320 --> 00:00:29.679
PM2.5
size so hasn't been in the most

00:00:29.679 --> 00:00:33.680
demanding scenario
although just looking at it it's pretty

00:00:33.680 --> 00:00:38.399
dirty so i thought
we'd take a look inside clean it

00:00:38.399 --> 00:00:41.440
and show you how it works firstly i'm
just gonna

00:00:41.440 --> 00:00:45.600
peel off this plastic protection that's
on the metalwork

00:00:45.600 --> 00:00:49.440
the metal cover on both sides are held
down

00:00:49.440 --> 00:00:54.559
by these little clips at either edge and
you can just get a spudger and insert

00:00:54.559 --> 00:00:58.800
them between the two
and pry up and bend the metal work and

00:00:58.800 --> 00:01:02.960
it comes apart
pretty easy however I do find that this

00:01:02.960 --> 00:01:07.040
style of spudger
that's thinner at the end is much easier

00:01:07.040 --> 00:01:10.240
to get between the two pieces and pry
them apart

00:01:10.240 --> 00:01:15.119
and once you've got the metal over all
of those clips it's just a case of

00:01:15.119 --> 00:01:18.960
pulling on it
until it opens up although you got to be

00:01:18.960 --> 00:01:22.720
careful and make sure you don't pull
too hard otherwise you'll damage the

00:01:22.720 --> 00:01:27.600
plastic or bend the metal too much
okay already this is pretty interesting

00:01:27.600 --> 00:01:31.439
just with one side off although
whilst we're at it we'll take the second

00:01:31.439 --> 00:01:35.040
side off which is now much easier now
that you don't have to get the spudger

00:01:35.040 --> 00:01:38.640
between the two pieces just coming from
the side like this

00:01:38.640 --> 00:01:42.079
now we have both sides of we can take a
look inside

00:01:42.079 --> 00:01:45.600
and the first thing we notice is this
metal work

00:01:45.600 --> 00:01:49.360
the case on the back of it is this black
sticker

00:01:49.360 --> 00:01:53.600
and that would be to prevent light
reflections either from the laser

00:01:53.600 --> 00:01:58.399
or from outside light getting onto the
sensor and affecting the readings

00:01:58.399 --> 00:02:03.119
and we can see inside this thing there's
a nice little centrifugal fan

00:02:03.119 --> 00:02:08.000
a laser and this little optical cavity
and we can also see that there's some

00:02:08.000 --> 00:02:12.319
springs clearly to
keep the chassis of the device grounded

00:02:12.319 --> 00:02:16.400
and i'm going to remove them before I
lose them

00:02:16.400 --> 00:02:21.840
the other side of the device has the PCB
and this is actually different to what I

00:02:21.840 --> 00:02:26.640
was expecting
I took one of these apart last year

00:02:26.640 --> 00:02:30.319
and it was a completely different design
here's a picture

00:02:30.319 --> 00:02:36.560
and as you can see this one today
it has a custom PLANTOWER chip

00:02:36.560 --> 00:02:42.879
whereas the other one has a
ST chip and some other eight pin

00:02:42.879 --> 00:02:46.160
SSOP package so I'm not quite sure
what's going on

00:02:46.160 --> 00:02:50.160
here is this is this other one of
counterfeit or something

00:02:50.160 --> 00:02:53.440
now the reason why I took this other one
apart

00:02:53.440 --> 00:02:57.760
last year was because I was a little bit
dumb and managed to plug it in the wrong

00:02:57.760 --> 00:03:01.599
way around
and yeah it doesn't like five volts with

00:03:01.599 --> 00:03:05.280
the wrong polarity
turns out I blew something up however

00:03:05.280 --> 00:03:09.040
looking at the thermal image of it it
was pretty easy to figure out what chip

00:03:09.040 --> 00:03:12.640
was getting
way too hot and it ended up just being

00:03:12.640 --> 00:03:16.640
this
little five pin SOT23 device here which

00:03:16.640 --> 00:03:22.500
i assumed was probably a regulator so I
replaced it with one of these

00:03:22.500 --> 00:03:24.640
Micrel 5504 regulators i had laying
around

00:03:24.640 --> 00:03:29.599
and it's been working ever since now if
you're like me you're probably wondering

00:03:29.599 --> 00:03:34.239
how on earth this thing actually works
and how it derives a value of micrograms

00:03:34.239 --> 00:03:38.879
per cubic meter well it's actually
very similar to how a smoke detector

00:03:38.879 --> 00:03:42.000
works
however unlike a smoke detector or

00:03:42.000 --> 00:03:46.720
cheaper dust sensor
this thing actually has to distinguish

00:03:46.720 --> 00:03:51.200
the difference between a small dust
particle and a large dust particle

00:03:51.200 --> 00:03:54.799
and to do that we have this whole
package here

00:03:54.799 --> 00:03:58.879
this is the block diagram that the
manufacturer gives us

00:03:58.879 --> 00:04:02.560
so we're going to look at how this light
scattering actually works

00:04:02.560 --> 00:04:07.040
unlike a smoke detector that usually
uses an infrared led

00:04:07.040 --> 00:04:12.640
this device has a surprisingly red laser
and it shines through here shines

00:04:12.640 --> 00:04:17.199
through these two light baffles
and then into this light trap at the end

00:04:17.199 --> 00:04:21.440
and that's just to stop
any stray reflections from hitting the

00:04:21.440 --> 00:04:25.520
photo diode at the bottom here
we also know from the data sheet that

00:04:25.520 --> 00:04:28.000
this device has a controlled flow rate
of

00:04:28.000 --> 00:04:34.320
0.1 liters per minute and that comes
out here across the sensing region

00:04:34.320 --> 00:04:37.919
and then out through here through the
centrifugal fan

00:04:37.919 --> 00:04:41.040
and that means we have a steady stream
of air

00:04:41.040 --> 00:04:44.320
with suspended dust particles coming
through here and

00:04:44.320 --> 00:04:47.360
as a dust particle crosses this laser
beam

00:04:47.360 --> 00:04:51.360
it will scatter the light onto this
photodiode and you can then pick that up

00:04:51.360 --> 00:04:54.479
and
count exactly how many dust particles

00:04:54.479 --> 00:04:57.520
have crossed that
laser beam and then you can infer

00:04:57.520 --> 00:05:02.639
through the 0.1 liters per minute
exactly how many dust particles are in

00:05:02.639 --> 00:05:08.479
one cubic meter of air and of course the
amount of light scattered by the dust

00:05:08.479 --> 00:05:11.360
particle
is proportional to the size of the dust

00:05:11.360 --> 00:05:15.840
particle as well as its
reflectance i suspect this is the method

00:05:15.840 --> 00:05:18.720
the
this unit uses in order to determine the

00:05:18.720 --> 00:05:21.360
particle sizes that we see in the data
output

00:05:21.360 --> 00:05:25.360
however some other more expensive
scientific

00:05:25.360 --> 00:05:29.919
dust measuring devices can also measure
how long the dust particle

00:05:29.919 --> 00:05:34.160
spends in the laser beam and at a
constant flow rate you can then infer

00:05:34.160 --> 00:05:38.080
the size of the dust particle
however that's a lot more difficult to

00:05:38.080 --> 00:05:41.120
measure so i doubt this unit's doing
that

00:05:41.120 --> 00:05:44.880
so taking a look at the pcb now and we
can see

00:05:44.880 --> 00:05:48.639
after the connector there's a surprising
amount of passives actually

00:05:48.639 --> 00:05:54.960
but there's two SOT23 5 pin packages
we know one of them is going to be a 3.3

00:05:54.960 --> 00:05:59.120
volt regulator for the
integrated circuits and powering the

00:05:59.120 --> 00:06:02.400
laser and everything
the other one i'm not quite sure what it

00:06:02.400 --> 00:06:05.680
does it's probably got something to do
with that fan

00:06:05.680 --> 00:06:09.039
might be some sort of PWM controller
because the

00:06:09.039 --> 00:06:13.360
microcontroller is able to regulate the
fan speed in order to get that

00:06:13.360 --> 00:06:20.080
0.1 liters per minute next we also have
here a surprising amount of passives i'm

00:06:20.080 --> 00:06:23.440
really not quite sure what they're doing
full of these passives and honestly i

00:06:23.440 --> 00:06:25.919
couldn't be bothered to reverse engineer
this thing

00:06:25.919 --> 00:06:30.400
especially with the black pcb although
that is kind of required to minimize

00:06:30.400 --> 00:06:33.919
light reflections
considering it's under the photodiode

00:06:33.919 --> 00:06:39.440
it's got something to do with that
next we have this rather large [LQFP]

00:06:39.440 --> 00:06:42.560
package that's been custom labeled PLANTOWER

00:06:42.560 --> 00:06:46.080
i doubt this is actual custom silicon so
that would be

00:06:46.080 --> 00:06:50.080
pretty expensive for something like this
that requires a

00:06:50.080 --> 00:06:53.599
reasonable amount of processing power
however

00:06:53.599 --> 00:06:57.520
my wizard senses are tingling and
telling me

00:06:57.520 --> 00:07:01.840
by the way the package is shaped with
that location dimple and the way the

00:07:01.840 --> 00:07:06.400
leads are formed
this is probably an STM32F part i would

00:07:06.400 --> 00:07:09.039
bet
but it could be anything it could be

00:07:09.039 --> 00:07:12.800
some chinese mcu manufacturer i've never
heard of

00:07:12.800 --> 00:07:16.639
over here we have some more passives as
well as

00:07:16.639 --> 00:07:20.160
what looks like a three pin SOT23
transistor

00:07:20.160 --> 00:07:23.919
this is clearly to drive the laser the
actual

00:07:23.919 --> 00:07:27.120
constant current part of the laser
driving though is

00:07:27.120 --> 00:07:33.199
on the back of the PCB next to the [TO-18]
laser and up here interestingly enough

00:07:33.199 --> 00:07:36.960
we actually have
a unpopulated device it looks like the

00:07:36.960 --> 00:07:41.520
footprint for an eight pin
dfn with an exposed pad my best guess

00:07:41.520 --> 00:07:45.039
what this would be
is probably some sort of memory option

00:07:45.039 --> 00:07:48.560
maybe it's some sort of EEPROM
that allows it to

00:07:48.560 --> 00:07:54.720
record locally some historic data
also here we have four holes through the

00:07:54.720 --> 00:07:57.360
pcb so that air can flow to the other
side

00:07:57.360 --> 00:08:00.560
i'm not quite sure why they've gone with
four holes and not a slot

00:08:00.560 --> 00:08:04.879
i suspect it's probably got something to
do for either a limitation with their

00:08:04.879 --> 00:08:08.400
pcb design package or maybe they're
trying to

00:08:08.400 --> 00:08:11.599
reduce the weight on the drill bit
drilling a slot

00:08:11.599 --> 00:08:14.879
now the pcb is held down by these screws
here

00:08:14.879 --> 00:08:18.319
but the other side is nothing it's just
traces

00:08:18.319 --> 00:08:22.639
now i open this device up to primarily
clean it so

00:08:22.639 --> 00:08:27.280
having a look at it now it looks like
there's a fair amount of dust there's a

00:08:27.280 --> 00:08:30.720
quite a fair bit on the blades of this
fan

00:08:30.720 --> 00:08:35.760
and there's a little bit round here
and it looks like there might actually

00:08:35.760 --> 00:08:40.640
be a little bit on the actual
photodiode now this is why they're

00:08:40.640 --> 00:08:44.720
so particular about how you orient this
device

00:08:44.720 --> 00:08:47.839
when you deploy it because if you deploy
it the wrong way around

00:08:47.839 --> 00:08:50.959
the dust can actually settle on the
photodiode so this

00:08:50.959 --> 00:08:54.000
has been in the position in which the
photodiode

00:08:54.000 --> 00:08:58.320
has been facing downwards so that no
dust can collect on it hence why it's so

00:08:58.320 --> 00:09:02.160
clean
on the other side on the pcb we can see

00:09:02.160 --> 00:09:06.640
that it's collected
a fair amount of dust as well i'm not

00:09:06.640 --> 00:09:10.320
quite sure why they actually have air
flowing over the pcb

00:09:10.320 --> 00:09:14.000
to me it seems like a big risk for
contamination

00:09:14.000 --> 00:09:18.399
although i suspect it allows them to
manufacture it a lot cheaper than

00:09:18.399 --> 00:09:21.680
having to put the pcb in a separate
compartment

00:09:21.680 --> 00:09:25.120
now i'm just cleaning this with a normal
art brush

00:09:25.120 --> 00:09:30.080
because i don't actually want to get my
nice esd brushes all dirty with dust

00:09:30.080 --> 00:09:35.440
but i'm not too worried about this this
fan here isn't actually held in with any

00:09:35.440 --> 00:09:38.959
screws it's just to press
fit so you can just push it out and

00:09:38.959 --> 00:09:42.640
clean it separately
and there's actually quite a lot of dust

00:09:42.640 --> 00:09:46.080
in this fan
i think there's probably more dust in

00:09:46.080 --> 00:09:51.279
the fan than anything else which
i suppose is kind of expected and

00:09:51.279 --> 00:09:55.120
i wonder if this has actually got
anything to do with the design of the

00:09:55.120 --> 00:09:59.360
PMSA003 which actually has the fan

00:09:59.360 --> 00:10:03.279
on the outside of the device and it's
holding with two screws and some

00:10:03.279 --> 00:10:09.040
clips so i wonder if that has anything
to do with making it easier to clean

00:10:09.040 --> 00:10:12.560
although one thing to note is the
plastic on this fan

00:10:12.560 --> 00:10:16.800
isn't as strong as the metal and i've
actually managed to damage one in the

00:10:16.800 --> 00:10:19.760
past so
that's one thing to be careful of and

00:10:19.760 --> 00:10:22.560
i'm just going to clean the rest of the
device here with

00:10:22.560 --> 00:10:29.440
that same art brush although in order to
clean that photo diode i'm using

00:10:29.440 --> 00:10:36.000
a q-tip soaked in 100% isopropyl alcohol
and with that done we can have a look

00:10:36.000 --> 00:10:39.200
around and everything looks much cleaner
now

00:10:39.200 --> 00:10:42.240
i don't know if this will have any
effect on the performance

00:10:42.240 --> 00:10:46.720
but it's good to just know it's not
horrifically dirty inside now

00:10:46.720 --> 00:10:51.760
and reassembly is easy enough it's just
a matter of popping the fan back in

00:10:51.760 --> 00:10:57.519
putting the connectors back on and
in order to put these metal shields back

00:10:57.519 --> 00:11:01.600
on
you can just bend the sides back in

00:11:01.600 --> 00:11:04.640
although making sure not to bend them
too far

00:11:04.640 --> 00:11:08.160
and then just pushing them back on they
clip into place

00:11:08.160 --> 00:11:11.200
i'm not sure how many times you can do
this before

00:11:11.200 --> 00:11:16.000
the metal starts to fatigue but i
imagine if you're doing this every like

00:11:16.000 --> 00:11:19.519
two years it's
not going to be a problem and so that's

00:11:19.519 --> 00:11:29.839
what's inside one of these and how to
clean it so thanks for watching

00:11:29.839 --> 00:11:31.920
you