en-AU.vtt

WEBVTT
Kind: captions
Language: en

00:00:00.000 --> 00:00:01.000
English (AU) (Spoken) [Manually Transcribed Captions]
github.com/WizardTim/WizardTim-captions

00:00:01.000 --> 00:00:04.800
Today we're going to be attempting a 
repair on this here washing machine.

00:00:05.440 --> 00:00:13.280
This is a Simpson brand washing machine from 
about 2010, this is part of their easy set range  

00:00:13.840 --> 00:00:21.760
and this particular unit has a bit of a problem. 
The fault condition this thing goes into is when  

00:00:21.760 --> 00:00:29.120
you turn it on it appears to work perfectly fine 
but after a few minutes it will just turn off  

00:00:29.120 --> 00:00:34.880
but when you turn it off at the power outlet 
after a couple seconds it will turn back on,  

00:00:35.440 --> 00:00:42.400
and then immediately back off, so I'm thinking it 
might have something to do with a power supply  

00:00:42.400 --> 00:00:50.000
problem so this might be salvageable. So getting 
into this there's just four screws at the back and  

00:00:50.000 --> 00:00:59.360
we can take off this top plastic cover and we can 
see here the line cable comes in it goes to a line  

00:00:59.360 --> 00:01:07.120
filter a power switch and then that goes to this 
here controller board which looks like it has the  

00:01:07.120 --> 00:01:14.400
power supply on it along with some other things 
possibly a VFD for the main motor there's two  

00:01:14.400 --> 00:01:20.640
solenoids here for the hot and cold water and 
then there's these two circuit boards here one for  

00:01:20.640 --> 00:01:26.640
all of the buttons LEDs and the seven segment 
display and we can see there's a buzzer on there  

00:01:26.640 --> 00:01:33.600
as well and then the other one is just a rotary 
encoder for the current mode so I'm suspecting  

00:01:33.600 --> 00:01:39.840
our problem is going to be on this power supply 
looking board so taking a closer look at that.  

00:01:40.960 --> 00:01:47.600
We can see here it's starting to look very 
power supply like, we have the mains input  

00:01:47.600 --> 00:01:52.560
here with all of the filtering a bridge 
rectifier a main bulk capacitor which is  

00:01:53.120 --> 00:01:59.840
quite large actually although this chip here quite 
large looks to be some sort of fully integrated  

00:01:59.840 --> 00:02:04.880
H-bridge or motor driver and that's probably 
what this connector here is going to that main  

00:02:05.600 --> 00:02:10.960
motor at the bottom of the washing machine. 
However we can see that there's a transformer  

00:02:10.960 --> 00:02:18.320
here, a lot of capacitors and some sort of probably 
switching transistor, so I suspect this here is our  

00:02:18.320 --> 00:02:26.800
power supply and we also can see that there's a 
controller here from International Rectifier  

00:02:27.360 --> 00:02:35.120
which is probably the main system controller for 
the entire washing machine or it might just be for  

00:02:35.120 --> 00:02:42.000
the motor because there is that other board but we 
might not even have to deal with any of that stuff. 

00:02:43.040 --> 00:02:48.480
Also of note is all of these opto-isolators 
and the fact that this has a transformer  

00:02:49.040 --> 00:02:54.320
so we can be pretty safe to assume that 
this is an isolated power supply and I  

00:02:54.320 --> 00:02:59.760
can just probe around in this low voltage 
stuff perfectly fine... or is it actually?  

00:03:00.560 --> 00:03:06.080
So maybe, maybe I should take a closer look at 
this before I start shoving my finger on that.  

00:03:06.080 --> 00:03:14.800
So with the board out I'm really not seeing 
where that isolation gap is I mean there's sort  

00:03:14.800 --> 00:03:23.520
of something here but what's all these things 
here going to the probably motor controller...  

00:03:25.280 --> 00:03:36.720
yeah... and the other side is a complete mess like 
where, where is this isolation? For reference here's  

00:03:36.720 --> 00:03:43.680
an ATX computer power supply and you can see 
that isolation gap is really obvious. So I  

00:03:43.680 --> 00:03:53.120
think maybe testing continuity and testing this 
bridge rectifier, yep the input is connected to  

00:03:53.120 --> 00:03:59.840
this bridge rectifier directly and how about 
the ground of this here crystal oscillator?

00:04:02.800 --> 00:04:09.600
Yep it's connected directly to neutral or at least 
one diode drop through the bridge rectifier.  

00:04:10.240 --> 00:04:17.360
So um yeah that, that could have been a bit of a 
tingle I mean it is neutral but there's plenty of  

00:04:17.360 --> 00:04:22.000
countries that have non-polarized plugs 
and there's plenty of houses that have  

00:04:22.000 --> 00:04:27.520
live and neutral reversed and I'm also not 
quite sure what other high voltage stuff is  

00:04:27.520 --> 00:04:31.120
on the supposedly "secondary" 
side of this power supply,  

00:04:31.760 --> 00:04:38.080
so yeah that's something to watch out for 
in these sorts of consumer appliances where  

00:04:38.080 --> 00:04:43.680
it's unreasonable to expect the end consumer is 
going to touch any of the electronic circuits.  

00:04:45.280 --> 00:04:51.040
And when you're working on stuff like this don't 
forget to discharge the main bulk capacitor on it, 

00:04:51.040 --> 00:04:58.720
this one's quite large i think it probably 
stores around 30 to 40 joules of energy so that could,  

00:04:58.720 --> 00:05:06.880
that could hurt quite a bit so now we can 
safely take a look at all of these soldered joints...  

00:05:07.760 --> 00:05:13.680
I'm not seeing anything here it looks all 
good we can of course see that this board is  

00:05:13.680 --> 00:05:20.080
nicely conformally coated so we can rule out any 
sort of contaminants affecting the performance  

00:05:20.800 --> 00:05:28.560
but the soldering quality on this looks quite 
good and I'm not seeing any broken solder joints  

00:05:29.280 --> 00:05:34.480
all of these surface mount devices also look 
quite good none of them are tombstone none of  

00:05:34.480 --> 00:05:41.600
them have cold joints. And on the other 
side taking a look at these capacitors  

00:05:42.560 --> 00:05:48.160
none of them are bulging, none of them 
are leaking, they all look perfectly fine  

00:05:50.000 --> 00:05:55.760
normally it's the capacitors because they are 
electrolytics, even this one here right next to  

00:05:55.760 --> 00:06:04.080
this transistor heatsink is perfectly fine so yeah 
it might not be an easy fix like a capacitor so...  

00:06:04.640 --> 00:06:11.840
So I guess it's time to whip out my secret 
weapon, my thermal camera and looking at the PCB 

00:06:11.840 --> 00:06:18.240
now installed into the device after about 
five minutes it decides to turn itself off  

00:06:19.040 --> 00:06:26.400
and going to the thermal view we can see that that 
transistor is getting quite hot by the looks of it  

00:06:27.520 --> 00:06:32.480
and coming in closer now I can see it's 
not actually that transistor but this here  

00:06:32.480 --> 00:06:37.040
DIP package device this thing 
here seems to be the problem  

00:06:37.920 --> 00:06:43.360
and getting the macro lens on it now we can 
see this is definitely the problem it's got a  

00:06:43.360 --> 00:06:51.200
massive hotspot on it and it is 128 degrees,
yeah this thing is not having a good day  

00:06:52.320 --> 00:06:57.760
this is definitely not meant to operate this hot 
because i'd expect at this temperature this FR-4 

00:06:57.760 --> 00:07:04.080
PCB would be heavily discolored but, this, this 
problem's only been happening for about a week  

00:07:04.080 --> 00:07:12.080
so it's been caught early on so it's not had time 
to damage the PCB. So yeah I'm suspecting this chip.  

00:07:12.640 --> 00:07:19.520
Although it could be something like a fault on 
the output is bringing this chip's output into  

00:07:19.520 --> 00:07:25.520
some sort of short circuit protection but I'd 
imagine there should be other hot spots and  

00:07:25.520 --> 00:07:31.600
a very hot PCB trace so I'm suspecting there's 
something maybe internally wrong with this chip  

00:07:32.640 --> 00:07:39.360
taking a closer look at that chip it is this 
one here and we can see the soldering on the  

00:07:39.360 --> 00:07:44.720
opposite side is perfectly acceptable so 
it's not that it's got a loose connection  

00:07:45.520 --> 00:07:51.600
so taking a closer look at the chip itself 
we can see it's one of these eight pin high-voltage  

00:07:51.600 --> 00:07:57.840
DIP packages because it has this missing 
pin here to get higher high voltage separation  

00:07:58.560 --> 00:08:06.240
so I'm suspecting this might be some sort of 
fully integrated off-line voltage switcher and 

00:08:06.240 --> 00:08:17.120
looking at the marking I'm not familiar with 
this logo by hand however the marking is "TNY278PN"

00:08:17.120 --> 00:08:23.360
and it appears to have been manufactured the 
32nd week 2010 which makes sense for this device  

00:08:24.080 --> 00:08:27.680
so I had a bit of a search round 
and I found this here datasheet  

00:08:28.640 --> 00:08:34.480
as it turns out this chip is from Power 
Integrations this is part of the TinySwitch-III  

00:08:34.480 --> 00:08:42.480
family which says it's an "energy efficient offline 
switcher with enhanced flexibility and extended  

00:08:42.480 --> 00:08:52.320
power range" and we can see here we have the "TNY278P"
which is either 16 or 28 watts depending on  

00:08:52.320 --> 00:08:58.400
presumably the cooling capabilities of the 
enclosure and taking a closer look into  

00:08:58.400 --> 00:09:06.320
this datasheet we can see a recommended circuit 
board layout which also includes a thermal check  

00:09:06.320 --> 00:09:12.720
note which says that the maximum operating 
temperature should not exceed 110 and further  

00:09:12.720 --> 00:09:19.200
in the data sheet we can see there is a thermal 
shutdown temperature of a 142 degrees which  

00:09:19.200 --> 00:09:25.360
I imagine in that thermal picture of it being 128 
degrees on the outside the die temperature was  

00:09:25.360 --> 00:09:32.400
probably certainly over that and that would also 
explain why this thing randomly turns on after  

00:09:32.400 --> 00:09:39.120
you unplug it from the wall probably some sort 
of die cooldown thing re-enabling the chip for  

00:09:39.120 --> 00:09:47.680
a second so I'm starting to get rather suspicious 
of this chip although I've just remembered I had a  

00:09:47.680 --> 00:09:55.520
ATX computer power supply from my old computer 
from 2011 that had a strange problem with what  

00:09:55.520 --> 00:10:02.640
I suspect was the standby power supply deciding 
not to work one day and if we take it apart we  

00:10:02.640 --> 00:10:13.680
can see it has exactly the same chip it is a "TNY278PN"
however this one's from the 18th week 2011.  

00:10:14.240 --> 00:10:19.120
now I never actually fixed this power supply 
because I just bought a new one however at the  

00:10:19.120 --> 00:10:25.600
time I was a little suspicious that this chip 
might be a problem and with this new washing  

00:10:25.600 --> 00:10:32.720
machine PCB I'm now extremely suspicious of this 
chip so I think the next best step is to just  

00:10:32.720 --> 00:10:38.640
buy a new chip and whack it in that's probably the 
easiest thing to do to check if that's the problem  

00:10:39.600 --> 00:10:48.400
and so I want this chip next day so taking a look 
on element14 we can see the "TNY278PN" is available  

00:10:49.120 --> 00:10:57.840
and it's only AU$2.64 so I shall buy some of them and 
we'll see if we can repair it tomorrow although it  

00:10:57.840 --> 00:11:05.280
will be the next second for you. So here's that new 
chip it's the same package although the engraving  

00:11:05.280 --> 00:11:11.920
is different we can see Power Integrations has 
changed their logo since this one is manufactured  

00:11:11.920 --> 00:11:20.640
the 42nd week 2021 so this should be quite simple 
to just get the desoldering gun out... uh oh okay...

00:11:24.800 --> 00:11:32.080
So I guess the the desoldering gun has decided 
to have a shorted heating element today  

00:11:32.640 --> 00:11:38.320
of all days I want to fix this today though 
so I guess we're going to be doing it manually  

00:11:38.960 --> 00:11:45.200
and I can't find my solder sucker either so I 
guess we're going to be demoing how to remove  

00:11:45.200 --> 00:11:51.760
a DIP package and put a new one in with a 
soldering iron and basic hand tools... oh goodie...  [sounds of despair]

00:11:53.200 --> 00:12:00.240
Okay so firstly we're going to be scraping 
off the conformal coating and adding  

00:12:00.240 --> 00:12:07.280
some leaded solder to make my life easier, also 
going to be very generous with the flux pen  

00:12:08.240 --> 00:12:14.720
and I might as well give this solder braid a 
go although I've never have any success with it  

00:12:14.720 --> 00:12:21.600
and uh yeah no success with that so I guess it's 
just going to be cutting the chip out just with  

00:12:21.600 --> 00:12:29.280
a set of side cutters pulling it out and then 
we can clean up the rest of the conformal coating  

00:12:29.280 --> 00:12:38.960
and then it's just as easy as heating up the pad 
on the other side and pulling out the legs.

00:12:38.960 --> 00:12:46.080
And now with that chip out we have to make sure that 
we can put the new ones in through the holes but  

00:12:46.080 --> 00:12:51.840
there currently aren't holes so this is a trick 
for you if you're in a very desperate situation  

00:12:52.400 --> 00:13:00.560
you can get the leg of a resistor and coat it 
in a generous amount of flux as well as the PCB

00:13:03.120 --> 00:13:08.480
you can then heat up the pad you want to make 
it sure it's very hot all the way through the  

00:13:08.480 --> 00:13:14.720
view and to the other side and then while 
still heating it put the resistor leg  

00:13:14.720 --> 00:13:22.640
in, you want to make sure to get it in all the way 
and then once you've got it in make sure to keep  

00:13:22.640 --> 00:13:29.280
moving it and keep adding heat moving it up 
and down until you get the entire leg coated in  

00:13:29.280 --> 00:13:36.000
the solder to pick up as much solder as possible 
and then inevitably when it cools down just keep  

00:13:36.000 --> 00:13:42.800
moving it especially in a circular motion and 
that will make sure that the solder does not  

00:13:43.360 --> 00:13:50.160
bond with the resistor leg so now 
you have a resistor stuck in the PCB 

00:13:50.160 --> 00:13:57.040
but now the PCB has a hole in it so just move it 
up and down like a file and eventually it will  

00:13:57.040 --> 00:14:05.120
come loose although you do have to be very careful 
not to damage the inner coating of the via that  

00:14:05.120 --> 00:14:10.560
can be quite fragile and if you damage it then 
that will be a very difficult repair however if  

00:14:10.560 --> 00:14:17.600
you're in a bind that can be quite a good way to 
get a through hole hole back to uh being able to  

00:14:17.600 --> 00:14:24.320
put a DIP package in and it looks like this 
has turned out reasonably well certainly not my  

00:14:24.320 --> 00:14:29.840
finest work but for today in a washing machine 
that's already 10 years old it should be fine  

00:14:30.720 --> 00:14:37.200
so putting that package in now and 
just soldering it on the other side  

00:14:38.880 --> 00:14:48.080
uh yeah turned out well enough. And so I remember 
that I've repaired this in the future I'm going to  

00:14:48.080 --> 00:14:54.960
also put a nice purple dot on it and then I think 
we can plug it in and test whether or not it works  

00:14:54.960 --> 00:15:04.000
and so if it plugged back in it's now consuming 
significantly less power down from 18.8 to 14.2  

00:15:04.960 --> 00:15:13.600
and that hot spot is now completely gone and 
we can see that that chip is no longer at  

00:15:13.600 --> 00:15:22.160
the scorching 128 degrees it's now a much more 
tame at 51.8 and this is after 20 minutes normally  

00:15:22.160 --> 00:15:28.560
it would turn itself off due to overheating after 
about five minutes and putting it on for a cycle  

00:15:29.280 --> 00:15:40.080
we can see it works perfectly fine now so I think 
we fixed it however you should know by now that  

00:15:40.080 --> 00:15:46.560
me on this channel is not quite happy with just 
fixing something i want to know exactly why it has  

00:15:46.560 --> 00:15:51.360
failed so we're going to be doing some destructive 
analysis on this chip to see if we can figure out  

00:15:51.360 --> 00:15:57.600
why it's stopped working so with these sorts 
of plastic dip packages you can heat them up  

00:15:58.160 --> 00:16:05.680
in this case to 400 degrees Celsius and chip 
away the plastic and expose the metal lead frame  

00:16:05.680 --> 00:16:12.880
which you can peel back and you can then 
remove the silicon die however in this case  

00:16:12.880 --> 00:16:18.720
400 degrees was not hot enough or I didn't let 
it dwell for long enough so I managed to crack the  

00:16:18.720 --> 00:16:25.040
die so we're also going to be doing it to one of 
the newer chips from 2021 and I'm going to bump  

00:16:25.040 --> 00:16:32.800
it up to 450 degrees and uh make sure it's nice 
and hot and uh that one yeah I managed to extract  

00:16:32.800 --> 00:16:40.320
the die intact for that one so here we have the 
two dies I extracted from the two separate chips  

00:16:40.320 --> 00:16:46.720
I was expecting there to actually be two separate 
dies in this chip for high voltage separation, 

00:16:46.720 --> 00:16:53.280
one for the control and one for the actual MOSFET but 
as it turns out it's just one big chip presumably  

00:16:53.280 --> 00:17:00.240
that helps keep the cots down quite a lot 
unfortunately the chip that failed I didn't heat  

00:17:00.240 --> 00:17:05.600
up enough and it has exploded everywhere although 
I managed to pick up some of the pieces I was  

00:17:05.600 --> 00:17:10.880
hoping though that I would be able to get it out 
in one piece and we could have a nice smoking gun  

00:17:10.880 --> 00:17:16.400
skid mark on the chip where the mosfet exploded 
or something really cool but unfortunately that's  

00:17:16.400 --> 00:17:23.360
not going to happen today and the other chip I 
managed to extract intact however I don't know  

00:17:23.360 --> 00:17:28.320
what happened I didn't get it hot enough but some 
of the plastic stuck to it so I have to chip that  

00:17:28.320 --> 00:17:36.800
off and then sand it down although unfortunately 
I wasn't able to consistently sand it very well so  

00:17:36.800 --> 00:17:44.400
part of the side has come off. Also here's the 
chip and the die appears to have been placed in  

00:17:44.400 --> 00:17:53.440
this area here and in this orientation and that 
correlates to the solder marks on the lead  

00:17:53.440 --> 00:18:00.160
frame as well as the thermal image and the wire 
bonding placement also makes sense for the pins  

00:18:02.000 --> 00:18:08.080
so here are the two dies or at least what's left 
of one of them and the other one and we can see  

00:18:08.080 --> 00:18:14.720
that there's two very obvious regions on the die 
one at the bottom for the quite large transistor  

00:18:15.440 --> 00:18:21.680
and the area on the top presumably has all of the 
control electronics for that, there's quite a lot  

00:18:21.680 --> 00:18:27.520
of it as well so let's have a closer look at these 
chips and see if we can find anything that might  

00:18:27.520 --> 00:18:34.400
be wrong with them or sort of uh speculate as 
to what the failure mode could have been seeing  

00:18:34.400 --> 00:18:43.360
as I lost half the other one so taking a look at 
the new chip we can see uh has a die marking here  

00:18:44.000 --> 00:18:51.040
it appears to have the old Power Integrations 
logo as well as the copyright mask copyright  

00:18:51.040 --> 00:18:58.640
of 2007 so this design definitely 
dates back to 2007 even in this newer chip  

00:18:59.440 --> 00:19:04.880
and it appears to have been marking a "DS58F4"

00:19:05.440 --> 00:19:12.080
and that matches with the older chip as well that 
has failed so presumably the die has not changed  

00:19:12.080 --> 00:19:21.120
between revisions there was also this other number 
on the top right of the new chip this is "DS58F"  

00:19:21.120 --> 00:19:28.160
although the "4" is missing although I might 
have sanded that off but I don't think so  

00:19:29.520 --> 00:19:34.720
but unfortunately on the chip that 
failed a nice little crack developed  

00:19:34.720 --> 00:19:38.560
exactly where that number was so I guess 
we'll never know i couldn't find that  

00:19:38.560 --> 00:19:43.760
shard in the pile of shards left over 
but I suspect it's probably the same

00:19:46.080 --> 00:19:53.440
and this here is one of the wire bonds for the 
transistor this is either the source or the drain  

00:19:53.440 --> 00:20:00.720
however you'll note that the large transistor is 
actually separated into three separate transistors  

00:20:01.760 --> 00:20:08.560
now this is important because I suspect this 
may actually be a failure mode that could happen  

00:20:09.440 --> 00:20:16.480
looking at the wire bonds we can see that they're 
actually different between the two devices on the  

00:20:16.480 --> 00:20:23.680
older device this is what the wire bonds look 
like there are directly bonded onto the main  

00:20:23.680 --> 00:20:32.320
area of the lead frame whereas the newer device 
has a different style of wire bonding which is  

00:20:32.320 --> 00:20:38.800
on one side of the lead frame so I'm 
wondering if maybe they had an issue with the  

00:20:39.440 --> 00:20:46.400
wire bonds failing and maybe uh one of 
these transistor areas became unavailable  

00:20:46.400 --> 00:20:53.120
which resulted in a high RDS(on) which caused the 
device to overheat and that's why they changed it  

00:20:53.840 --> 00:21:00.400
maybe? But that really is just baseless 
speculation it could be as simple as  

00:21:00.400 --> 00:21:06.080
they started manufacturing their chip at 
a different semiconductor fab house and  

00:21:06.080 --> 00:21:11.520
they just had a different wirebond machine the 
lead frame is completely different unfortunately  

00:21:11.520 --> 00:21:16.880
I've destroyed the lead frame in doing this and 
didn't take a picture of it but unfortunately  

00:21:16.880 --> 00:21:23.600
seeing as I managed to destroy the die of 
the faulty device I guess we'll never know  

00:21:24.400 --> 00:21:31.360
but I think that's a reasonable explanation 
so uh I guess thanks for watching and I'll  

00:21:31.360 --> 00:21:43.840
leave you with some video of the die of the 
broken one it's pretty cool features in here.