.Spoiler - it wasn't a capacitor.
If attempting a repair like this then take extra care to ensure the power is off to the PCB. It has a lot of exposed live tracks on it.
Do you remember when thermostats were cheap, simple and lasted for over 30 years? Welcome to the future where they are expensive, crammed with electronics and may not last a tenth of that.
I think this one managed to scrape by its one year warranty period before ceasing to sense.
I would categorise my test-bodge as a temporary measure to identify the fault. It would be better to get the proper component, although it's not an easy one to desolder. It may require more aggressive heat pen use and probably some careful masking with Kapton tape to protect adjacent components.
In use the unit seems to run pretty cool. The two hottest components visible to the thermal camera were the PSU diode and the PSU primary side MOSFET, and neither of them was particularly hot.
If you enjoy these videos you can help support the channel with a dollar for coffee, cookies and random gadgets for disassembly at:-
http://www.bigclive.com/coffee.htm
This also keeps the channel independent of YouTube's advertising algorithms allowing it to be a bit more dangerous and naughty.
#ElectronicsCreators

It's a internet of things, nest thermostat and it's one of our favorite types of internet of things. It's dead internet of things. This is apparently completely dead. I'd like to thank steve of oil of man for giving me this.

He said a couple of these fail and uh was interested to see if i could find out what's going wrong with them. So let's start by hooking the power up to this and giving a go see if it goes bang or does something weird. So that's the power at the left-hand side. Um right, tell you: what is this screwdriver going to fit in? I may have to use a smaller screwdriver yeah, i'm going to have to use a smaller screwdriver.

Where is my smaller screwdriver? No, that's not it! This one is promising a slight avalanche of screwdrivers, so let's see how easy it is to connect into this. In the first place. Where is a suitable cable? I had a cable hold up here. It is, although there will be effectively a ground somewhere uh.

I guess you're supposed to connect the ground externally. It does actually show an earth connection here. Oh no! That's a ground circuit internal circuitry, ground. Okay, just ignore what i said i'm talking.

I sometimes do that this just casts doubt already, when you're shoving wires into this little through these little holes, that it just strikes that you could have splayed wires touching each other inside. Maybe i am that would be good plug it in this internet of sparks. Let's get a guide to leds if any leds light press the button to activate manual mode okay right. So, let's plug this in, i shall unplug what's plugged in already azure plug this in and we'll see if any leds light nothing is it going to take a while to boot up push a button, nothing, it's completely dead.

So that's good! That's a good start because uh that's what you call a decisive fault. What's the bet, it's the power supply? Oh, is the power supply right to you, what let's open it? So i shall pop these wires out. I have unplugged it and we shall open it. What screws do we have in here, oh they're, very, very tiny screws? Is this even going to fit? This might not fit oh hold on hold on.

I think i've managed to force it into the screw head enough to actually get a screw. Oh they're, tiny screws. I have to say i've always had my doubts about internet of things stuff, it just seems less reliable a while back, i featured a thermostat, it was a honeywell thermostat and it was ancient. It used little sort of thermally, operated, bellows and the thing had been in use for decades.

This stuff won't last that long. There are so many weak points. There's so many things to feel. Let's go, i can just feel this wobbling.

Is it going to come out? Yes, it is okay, there's some more screws. Let's take all the screws out. I shall keep those ones separate. They look different to others, maybe i'm just being a cynical person thinking.

You know that the entire thing's also to me it's not a terribly secure thing. It's a hacker's playground and no matter how well they work in the software you're always going to get weaknesses. So it's going to come out just get a charge, capacitor i'll find out. When i touch it.
Oh there's a circuit board with a ribbon cable onto it. Is that for leds it is the leds right. Tell you what i'm going to have to remove this little ribbon cable. Is this one that slides out the way or back the way, or does it lift up it lifts up i've broken so many of those in the past.

In the early days before i got wise, what do we have? Oh that's, chunky, uh. This looks like the spicy section i may short those capacitors out just in case i will short those capacitors out just in case nothing finger test. Okay is the fuse intact? Well now, let's get the metering, let's get the metering put it to continuity. The meter is always on 20 volts dc or continuity.

It just never seems to be anything else, so there is a little fuse. The little fuse is intact and it's going up to a tiny, little class wag paster. Then a little thing is that an inductor, not sure that is that's going to the bridge right far and then to those capacitors right. So i'm guessing that this is going to be a little switching chip and that is going to be power supply and that these capacitors are immediate suspects for potentially not playing ball right.

Tell you what let's check that see if we get any sensible voltages, i'm just going to pause when i connect this up to the power again, keep in mind, it will all be live at spicy vultures. Now do so one moment please, the leads are connected, but i have noticed something that's a bit suspicious. I did some preliminary probing and the output capacitors. This is not powered up at the moment.

Uh the output capacitors are showing almost dead short circuit and you'd expect. Not a complete short circuit and the likewise the diode here is showing continuity. I wonder if the diodes failed or there's something shorter than the secondary side, so my inclination at the moment is to actually uh power, this up with a low voltage power supply current limited and actually see what sort of current it draws. It's probably going to draw a lot to current, but also see what heats up one moment.

Please, i'm just going to set that up right now, progress so far, it's a complete dead short. I didn't run it beyond 200 milliamps because i'm very suspicious that this diode may have failed, and if that diode is short-circuited, it means that all the current is going to be basically shunted through the winding of the transformer. And i don't want to damage the sector of the transformer, so what i'm going to do is i'm going to remove that diode and see if the fault departs best way to do this uh, i could try using the soldier, iron and just gently lifting one end Up i'll try that with tweezers, but after that, it's probably a case of using the hot air gun to do that, although there's other stuff never seen to it's just making a bit more cost that. So i shall try putting a little bit soda in this first and lifting that diode get some lead-based solder in this just because it makes it melt easier.
Maybe i'll just add some flux as well. That's always a good thing right, uh where's, some phlox uh. This will do. I shall just squirt some flux all over it.

The flux helps make the uh heat on and sort of permeate the soda. So to speak, that's not really good terminology. Is it? Oh? So i'm going to try and grip this i'm going to try and lift it. That's not working so far.

It's not not working! I may have to use the hot air gun to actually melt everything at once. Yeah. I think we'll have to do that right. Do you know what i'm gon na pause, because that could be quite time consuming because i'm going to have to heat it up.

One moment please: well, that's progress. The diode is off. It came off in two pieces. Unfortunately, it said i'll just zoom down this.

It's the type of diode that has the really large heat sink pad at one end, and that was making it very difficult to get off there. I perhaps should have just persisted a bit more with the hot air, but i didn't want to displace components in the vicinity, but it's off and the short circuit has disappeared. This is a good thing. Now the diode was an s3 100 70 volt, three amp.

The closest i've got to that is a uf5401, which is a a 3 amp dial, but look at the size of it compared to what was there already? It's really huge, but i also have um. I also have shortcuts, but only rated 20 volt. I think that's too low for this, because they're probably a reason they chose a hard voltage on, but i do have uf 4003s uh. The uf4003 is the high-speed version of the 1n403.

The standards of 1n range 1n. 401 to 1n407 have this high speed version designed for switch mode power supplies. So i'm going to stick one of these in and i'm going to see what temperature gets up to because, to be honest, i reckon that's a 12 volt bus and it's very unlikely to be used anywhere near one amp, but i could be wrong. I worked out that the relays are 12 volt relays and they're pretty much being powered directly from that, so i'm guessing that it starts off with 12 volts and then for the other modules.

It just steps it down locally with that little regulators. So now i have to work out how to get this diode into that cramped space. I think i may just sit it across the circuit board. What is a good convenient way to do this i'll get rid of these connections? For a start, let's get my little hot jump wires off interesting things about this.

I think that diode was deliberately overrated to produce a reliable product. Unfortunately, it's backfired there, that's just that particular diode was not reliable. It's also notable that these two capacitors are different styles and different brands, but the same value and they're in parallel, possibly as built-in redundancy in case. One turns out to be a duffer and it can then the other brand one uh kind of covers for it.
So to speak, which is quite a nice idea actually right. I want to keep this away from the rf sections. That is an rf section over there, so i don't want to disrupt that area too much. I could solder one end onto this big pad and the other one directly onto the pad of the transformer.

I think that's actually not a bad option here. That is not a bad option. Right tell you what i shall start by floating this down and to the side, so i can get it onto that big pad. I'm wondering what space i've got in here as well.

I think i've got decent clearance, but you just never know they tend to make these things so small. This would be sitting in like that. So this once this is sitting in there's not a huge amount of space in the back, it is going to have to be quite low profile, but this component sticks up a bit, so it should be okay. It depends how these ribs interfere with that in the back, when i put it back on, because they're going to be definitely making a bit bigger that back out again now right soda this, i wonder if this is a common fault in these, it would be interesting To know that, because uh, it could certainly save a lot of hassle if you could just basically swap that diode and get your unit back up and running.

Every product like this seems to have a weakness in some way, not a weakness designed by the manufacturer. In this instance uh, this is something that will have taken them by surprise. They will be annoyed if this has been happening a lot so now uh that's going to sit quite low. That's quite good! Now i'm going to fold this lead down and onto that pad and then solder on.

So this is going on to the switch mode power supply transformer. I'm going to have to be careful not to short onto an adjacent ground plane with it, try and stay in shot for this flow. Some solder on then i'll double check. I've got the diode around the right way.

I mean seriously. I should have been thinking about that in the beginning, but not to worry making a mess of this. I normally do it's small, it's fumbly, it's not it's, not an easy fix. The failure of the schottky diodes isn't uncommon, they're, the second biggest failure of power supplies.

After the capacitors, the usual puffy capacitors right that is on, and i think once it's cooled down a bit, i may just bend over a little bit. Well, that's not going to bend over a little bit. Let's try not to destroy it in the process, but i think that's low enough. Is it going to bend down a bit? I think that's more or less it right.

Well, i could power this up, but there are no indicator leds in it, but tell you what? Let's power up anyway and see what happens and i'll get the thermal imaging camera. Look at that diode and i shall try then, to actually activate both those relays. Uh. What do i need? I need mains, i need a mains cable, you think of these things and all the circuitry on them all the processors and all this wi-fi modules and stuff like that, and they can just be taken down completely by one tiny little component, usually in the power supply.
That's quite that's quite annoying, but it's very common. It keeps the electronic service guys in a job, keeps the plumbers and heating engineers in a job just swapping these modules completely. I wonder how much setting up is required for these probably quite a lot. I've never really looked into them too much.

It's not that you can just swap it in the wall and it instantly works right. What's gon na happen, there are no leds in this that i know of. I could connect that the indicator panel on it. Let's see what happens i plug this in uh, we'll see if we get a voltage there, i heard relays click if we get 12 volts uh.

Let's set this to 20 volts dc and gingerly probe on to untangle the leads first gingerly probe, i'm keeping my fingers. Well away from all the live bits gingerly probe onto there and there we have 11.6 volts, which is ample right. This is good right. Tell you what give me a second, i'm going to put this back in its case with this indicator lights and we'll see if we can get uh do a test in this and i'll get a temperature off this diode.

One moment please: okay, it's back in the leds are connected. I plug it in, and things activate. Now, let's put these relays into the both in the on position. I reckon that's going to be one of the highest power consumptions.

Although having said that, rated communication would probably also be quite a high power consumption, but let's stick this to continuity and check. There's two relays this one and this one. So if we can change them over to the other position, that would be both the coils energized. So, let's power this up that has energized both those coils right.

So now i'm going to get the thermal imaging camera and i'm going to check out what temperature that diode gets up to. It's bitten up, it'll, take a while to boot up as you'll pause while it boots up. The diode is looking very comfortable at about about 17 degrees celsius relative to ambient. It's not terribly hot in here it's well, it's about 90 degrees celsius in here uh.

So relative to ambient it's uh about 10 degrees above ambient that diode, i wonder if the circuit will actually have peaks of current consumption. There is a a device over here regulator. Perhaps it's pretty toastable, it's not torches 21.5 degrees celsius. Is there anything super hot in here? No there's, not everything is running pretty cool in this unit, so that looks like it uh.

I would guess that. Well, there's only one way to find out. I should shall leave that diode in use um. I should tame this down a bit taming down and focus down onto that.
So is the back going to fit on. That is a very fumbly connector for big fingers is the back going to fit on. I think it is, i think the back is going to fit on over that dive without any problems. Oh, yes, that's perfect, so that is it that's the fix, if you have one of these check that goes dead completely dead check across that diode for a dead short circuit, and if you find there is a dead short circuit, remove it and you may have to Get destructive you may have to break off because uh they're quite hard to do soda and replacing it with a well, preferably the same diode if you can get a hold of it.

The s3 100, which is 70 volt, 3 amp or well. In this case, i've put in uh what was it uf 4003, which was just what i had, which is actually only rated one amp, but is rated about 200 volts and it seems to be running cool. I'm not sure. If there's, i'm not sure if this device really increases in power consumption, though, when it's trying to wi-fi communicate, i'm not really sure that might happen, it's hard to say ideally the correct diode, for it will be the one they originally designed it for, hopefully, a more Reliable one though, but that is it, fixing the nest.

Oh look. How cleared out that is because it's white fixing the nest, uh thermostat, just a simple thing: it wasn't capacitors this time it was a diode.

12 thoughts on “Fixing a completely dead nest thermostat.”
  1. Avataaar/Circle Created with python_avatars Gregory Thomas says:

    I will never have any internet-connected IoT devices…though I am investigating running my own cloud server and using micro computer based devices instead.
    I already run my personal server on a local-only intranet which is totally independent from the internet connections so adding a cloud server would not be very difficult.
    The local intranet is hardwire based…no wifi…so any future connections will have to be with ethernet cables as well…most Arduino's already have a connector anyway.

  2. Avataaar/Circle Created with python_avatars Josh Bacon says:

    There’s planned obsolescence for you – where manufacturers deliberately shorten the lifespan of their products and/or make them hard to repair on purpose, meaning consumers become repeat customers earlier than they should, resulting in more profit for the company at the consumer’s expense…

  3. Avataaar/Circle Created with python_avatars Dick Cheesehead says:

    Thanks for the content and your reassuring calming personality. I always have been intrigued in the magic of micro-electronics. I never got into it and still I don't understand half of the things that are on a PCB. It's very fun watching you (read: oh holy expert) do the screwing around.

    Cheers, mate!

  4. Avataaar/Circle Created with python_avatars Stu C. says:

    Point of order Mr Clive, that's the control centre, not the thermostat.

    I've had two fail within the warranty period and Nest replaced them without any argument. The later ones seem to be much more reliable.

  5. Avataaar/Circle Created with python_avatars Paul Mear says:

    Hi Clive, I had the same problem with my nest. It is a design or manufacturing fault. Nest replaced mine out of warranty for free, sent me a replacement by 24hr courier once I got through their customer service triage. Happy days. P.s it still worked when powered via the micro usb socket.

  6. Avataaar/Circle Created with python_avatars 6581punk says:

    Well, not so much the thermostat but the switch box they include in the UK bundle. I had one die on me too. But I moved to Tado now as I use a Google free smartphone and the Nest app wouldn't sign in. I think the whole design of the Nest is rubbish, should be able to turn the thermostat down to an OFF position, but if you want to switch off the heating you have to faff about with clicks and turns. Also why have a rotary dial on the app/website? no point having skeuomorphism and making it harder to control. Tado has a slider like any sensible UX designer would use.

  7. Avataaar/Circle Created with python_avatars Peter Stimpel says:

    Oh, Glasgow did let you go, finally. Great! And hey, watching you fixing this microscopic SMD stuff using a proper THT thingy .. the industry hates this trick

  8. Avataaar/Circle Created with python_avatars rasmis says:

    In a perfect world electronic waste would be available for cheap for hobbyists. Unfortunately it's more profitable to ship it to poorer countries, shred it and extract the precious metals.

  9. Avataaar/Circle Created with python_avatars Dave Fredrock says:

    Open the pod bay doors Hal.

    The newer Nest thermostats are in excess of $200. I have a programable Honeywell one I got at a resale place for $3 that's been working well for a decade.

    Oh no, I can't use an app to see the temperature setting while I'm miles away. HA!

  10. Avataaar/Circle Created with python_avatars IncertusetNescio says:

    IoT devices are kept WELL away from my home network precisely because they are extremely vulnerable and almost never patched. Too many Bleeping Computer articles mention vulnerabilities in them.

  11. Avataaar/Circle Created with python_avatars Stephen Eyles says:

    Great to see you're back home Clive, and nice to see that scorched desk again!
    Nice fix on this one; gives us courage to investigate rather than chuck stuff out and buy a new one! :-))

  12. Avataaar/Circle Created with python_avatars Jeff Snyder says:

    9º? You keep your work desk at 9º (48ºF)?!? No wonder you're always in heavy flannels!

    Great fix!

    I love the Nest hardware; shame they were bought by Google. By example, Google just announced the discontinuation of the OnHub series of routers which require the Google Home app to manage and operate. Perfectly good devices will simply die when Google switches the ability to mange them off.

    The Nest home thermostats and controls will hopefully have a longer lifecycle, but perhaps the bean counters will spot this opportunity to prematurely deprecate perfectly good gear too.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.